# -*- coding: utf-8 -*-
# written by Ralf Biehl at the Forschungszentrum Jülich ,
# Jülich Center for Neutron Science 1 and Institute of Complex Systems 1
# Jscatter is a program to read, analyse and plot data
# Copyright (C) 2015-2024 Ralf Biehl
#
# This program is free software: you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation, either version 3 of the License, or
# (at your option) any later version.
#
# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
# GNU General Public License for more details.
#
# You should have received a copy of the GNU General Public License
# along with this program. If not, see <http://www.gnu.org/licenses/>.
#
"""
**dataArray**
dataArray contain a single matrix like dataset with attributes.
- matrix like data with columns and rows
- attributes are linked to the data e.g. from a measurement/simulation/fit parameters.
- all numpy array functionality preserved as e.g. slicing, index tricks.
- fit routine from scipy.optimize (least_square, differential-evolution, minimize, ...)
- read/write in human readable ASCII text including attributes (optional zipped).
- For programmers: a ndarray subclass
dataArray creation can be from read ASCII files or ndarrays as data=js.dA('filename.dat').
See :py:class:`~.dataArray` for details.
Hint for Beginners:
- The dataArray methods should not be used directly from this module.
- Instead create a dataArray and use the methods from this object as in the examples.
**Example Read data and plot** (see :ref:`Reading ASCII files` for more about reading data) ::
import jscatter as js
i5=js.dA(js.examples.datapath+'/iqt_1hho.dat',index=5) # read 5th data from file with multiple data
i5=js.dL(js.examples.datapath+'/iqt_1hho.dat')[5] # same as above (but read as dataList, then select 5th)
p=js.grace()
p.plot(i5)
**Example create/change/...** ::
import jscatter as js
import numpy as np
x=np.r_[0:10:0.5] # a list of values
D,A,q=0.45,0.99,1.2
# create dataArray from numpy array
data0 =js.dA(np.c_[x, x**2, A*np.sin(x)].T)
data=js.dA(np.vstack([x,np.exp(-q**2*D*x),np.random.rand(len(x))*0.05]))
data.D=D;data.A=A;data.q=q
data.Y=data.Y*data.A # change Y values
data[2]*=2 # change 3rd column
data.reason='just as a test' # add comment
data.Temperature=273.15+20 # add attribut
data.savetxt('justasexample.dat') # save data
data2=js.dA('justasexample.dat') # read data into dataArray
data2.Y=data2.Y/data2.A
# use a method (from fitting or housekeeping)
data2.interp(np.r_[1:2:0.01]) # for interpolation
**Example fit** ::
import jscatter as js
import numpy as np
data=js.dA(js.examples.datapath+'/exampledata0.dat') # load data into a dataArray
def parabola(q,a,b,c):
y = (q-a)**2+b*q+c
return y
data.fit( model=parabola ,freepar={'a':2,'b':4}, fixpar={'c':-20}, mapNames={'q':'X'})
data.showlastErrPlot()
The dataarray module can be run standalone in a new project.
**dataList**
dataList contain a list of dataArray.
- List of dataArrays allowing variable sizes and attributes.
- Basic list routines as read/save, appending, selection, filter, sort, prune, interpolate, spline...
- Multidimensional least square fit that uses the attributes of the dataArray elements.
- Higher dimesions (>1) are in the attributes.
- Read/Write in human readable ASCII text of multiple files in one run (gzip possible) or pickle.
- A file may contain several datasets and several files can be read.
- For programmers: Subclass of list
For Beginners:
- Create a dataList and use the methods from this object in *point notations*.::
data=js.dL('filename.dat').
data.prune(number=100)
data.attr
data.save('newfilename.dat')
- The dataList methods should not be used directly from this module.
See :py:class:`~.dataList` for details.
**Example**::
p=js.grace()
dlist2=js.dL()
x=np.r_[0:10:0.5]
D,A,q=0.45,0.99,1.2
for q in np.r_[0.1:2:0.2]:
dlist2.append(js.dA(np.vstack([x,np.exp(-q**2*D*x),np.random.rand(len(x))*0.05])) )
dlist2[-1].q=q
p.clear()
p.plot(dlist2,legend='Q=$q')
p.legend()
dlist2.save('test.dat.gz')
The dataarray module can be run standalone in a new project.
_end_
"""
import time
import sys
import os
import copy
import collections
import io
import gzip
import glob
import warnings
from functools import reduce
import types
import inspect
import pickle
import numbers
import multiprocessing as mp
import numpy as np
import scipy.optimize
import scipy.interpolate
import emcee
from . import parallel
class notSuccesfullFitException(Exception):
def __init__(self, value):
self.parameter = value
def __str__(self):
return repr(self.parameter)
# Control Sequence Introducer = "\x1B[" for print coloured text
# 30–37 Set text color 30 + x = Black Red Green Yellow[11] Blue Magenta Cyan White
# 40–47 Set background color 40 + x,
CSIr = "\x1B[31m" # red
CSIrb = "\x1B[31;40m" # red black background
CSIbr = "\x1B[30;41m" # black red background
CSIyr = "\x1B[33;41m" # yellow red background
CSIy = "\x1B[33m" # yellow
CSIyb = "\x1B[33;40m" # yellow black background
CSIg = "\x1B[32m" # green
CSIgb = "\x1B[32;40m" # green black background
CSIm = "\x1B[35m" # magenta
CSImb = "\x1B[35;40m" # magenta black background
CSIe = "\x1B[0m" # sets to default
#: update interval for errPlots in sec
errplotupdateinterval = 2
#: returns a log like distribution between mini and maxi with number points
def loglist(mini=0, maxi=0, number=10):
ll = np.r_[np.log((mini if mini != 0. else 1e-6)):
np.log((maxi if maxi != 0 else 1.)):
(number if number != 0 else 10) * 1j]
return np.exp(ll)
def _w2f(word):
"""
Converts strings if possible to float.
"""
try:
return float(word)
except ValueError:
return word
def _w2i(word):
"""
Converts strings if possible to integer.
"""
try:
return int(word)
except (ValueError, TypeError):
return word
def is_float(s):
try:
float(s)
return True
except ValueError:
return False
def is_dataline(words):
"""
Test if line words starts with float.
wf : list of strings
"""
try:
return is_float(words[1]) and is_float(words[0])
except IndexError:
return False
def _deletechars(line, deletechar):
# first try utf8 method
return line.translate({ord(st): None for st in deletechar})
def _readfile(xfile, encoding=None):
"""
Reads from normal file, gzip file or stringIO or returns list
"""
if isinstance(xfile, list) and all([isinstance(zz, str) for zz in xfile]):
# a list of strings
return xfile
try:
# test if xfile is IOString which should contain list of strings
zeilen = xfile.getvalue()
zeilen = zeilen.splitlines(True)
except AttributeError:
if os.path.isfile(xfile):
if xfile.endswith('.gz'):
_open = gzip.open
_readmode = 'rt' # to prevent default (binary) and force text
else: # normal file
_open = io.open
_readmode = 'r' # default is text
with _open(xfile, _readmode, encoding=encoding) as f:
zeilen = f.readlines()
else:
raise Exception('Nothing found in :', xfile)
return zeilen
def _append_temp2raw(raw_data, temp, single_words, xfile, ende):
"""
Internal of _read
appends new dataset temp to raw_data list and sets temp to empty structure
including the single words and the original filename xfile
ende is indicator if temp was last set in a _read file
temp is dict {'com':[],'_original_comline':[],'para':{},'val':[]}
raw_data is list of temp
"""
# this function is only visited if lines change from nondata to data or from data to nondata
# so we have data and para_com_words or only para_com_words or only data
if len(temp['val']) == 0 and (len(temp['para']) > 0 or len(temp['com']) > 0 or len(single_words) > 0) and ende:
# parameters found after the data lines at the end of a file
# append it to last raw_data
if len(raw_data) == 0:
raise ValueError('There were no data read; it was all about parameters ')
else:
for key in temp['para']: # discriminate multiple para with same name
if key in raw_data[-1]['para']:
num = 1
while key + str(num) in raw_data[-1]['para']: num += 1
keynum = str(num)
else:
keynum = ''
raw_data[-1]['para'][key + keynum] = temp['para'][key]
if len(single_words) > 0:
temp['com'].append(' '.join(single_words.split()))
for line in temp['com']: # append comments not already in raw_data
if line not in raw_data[-1]['com']:
raw_data[-1]['com'].append(line)
elif len(temp['val']) > 0 and (len(temp['para']) > 0 or len(temp['com']) > 0 or len(single_words) > 0):
# append to raw_data if a parameter and a data section was found
# ende guarantee that at last line appending to raw_data is forced
if '@name' not in temp['para']: # if not other given the filename is the name
temp['para']['@name'] = xfile
if len(single_words) > 0: # add single word to comment
temp['com'].append(' '.join(single_words.split()))
# ==>>>>> here we add new data from temp to raw_data
raw_data.append(temp)
try: # add the values to raw_data
# raw_data[-1]['val']=np.squeeze(temp['val']) # remove dimensions with length 1
raw_data[-1]['val'] = np.atleast_2d(temp['val'])
except TypeError: # if type error try this; but it will never be called
raw_data[-1]['val'] = []
for col in temp['val']:
raw_data[-1]['val'].append(np.squeeze(col))
else:
# return unprocessed data but increase len of single_words to indicate visit here
# this happens if only data lines separated by empty lines are in the file
single_words += ' '
return temp, single_words
# pass empty temp
single_words = ''
temp = {'com': [], '_original_comline': [], 'para': {}, 'val': []}
return temp, single_words
# a global that contains dataList during fit with bayes and differential evolution
# this speeds up using a pool of workers. Should be safe as only one fit at a time is running.
bayes_dL = None
def pool_error(*args, **kwargs):
"""
Error function defined on module level can be pickled in fast simple way
Use if fitting with pool of workers
"""
global bayes_dL
parameters, *largs = args
nargs = (parameters,) + (0,)
return bayes_dL._errorfunction(*nargs, **kwargs)
# noinspection PyBroadException
def _parsefile(xfile, block=None,
usecols=None,
skiplines=None,
replace=None,
ignore='#',
takeline=None,
delimiter=None,
lines2parameter=None,
encoding=None):
r"""
**How files are interpreted** :
| Reads simple formats as tables with rows and columns like numpy.loadtxt.
| The difference is how to treat additional information like attributes or comments and non float data.
**Line format rules**:
A dataset consists of **comments**, **attributes** and **data** (and optional other datasets).
First two words in a line decide what it is:
- string + value -> **attribute** with attribute name and list of values
- string + string -> **comment** ignore or convert to attribute by getfromcomment
- value + value -> **data** line of an array; in sequence without break, input for the ndarray
- single words -> are appended to **comment**
- string+\@unique_name-> **link** to other dataArray with unique_name
Even complex ASCII file can be read with a few changes as options.
Datasets are given as blocks of attributes and data.
**A new dataArray is created if**:
- a data block with a parameter block (preceded or appended) is found.
- a keyword as first word in line is found:
- Keyword can be eg. the name of the first parameter.
- Blocks are separated as start or end of a number data block (like a matrix).
- It is checked if parameters are prepended or append to the datablock.
- If both is used, set block to the first keyword in first line of new block (name of the first parameter).
Example of an ASCII file with attributes temp, pressure, name::
this is just a comment or description of the data
temp 293
pressure 1013 14
name temp1bsa
XYeYeX 0 1 2
0.854979E-01 0.178301E+03 0.383044E+02
0.882382E-01 0.156139E+03 0.135279E+02
0.909785E-01 0.150313E+03 0.110681E+02
0.937188E-01 0.147430E+03 0.954762E+01
0.964591E-01 0.141615E+03 0.846613E+01
0.991995E-01 0.141024E+03 0.750891E+01
0.101940E+00 0.135792E+03 0.685011E+01
0.104680E+00 0.140996E+03 0.607993E+01
this is just a second comment
temp 393
pressure 1011 12
name temp2bsa
XYeYeX 0 1 2
0.236215E+00 0.107017E+03 0.741353E+00
0.238955E+00 0.104532E+03 0.749095E+00
0.241696E+00 0.104861E+03 0.730935E+00
0.244436E+00 0.104052E+03 0.725260E+00
0.247176E+00 0.103076E+03 0.728606E+00
0.249916E+00 0.101828E+03 0.694907E+00
0.252657E+00 0.102275E+03 0.712851E+00
0.255397E+00 0.102052E+03 0.702520E+00
0.258137E+00 0.100898E+03 0.690019E+00
optional:
- string + @name:
Link to other data in same file with name given as "name".
Content of @name is used as identifier. Think of an attribute with 2dim data.
- Attribute xyeyx defines column index for ['X', 'Y', 'eY', 'eX', 'Z', 'eZ', 'W', 'eW'].
Non integer evaluates to None. If not given default is '0 1 2'
Line looks like ::
XYeYeX 0 2 3 - 1 - - -
**Reading of complex files**
with filtering of specific information
To read something like a pdb structure file with lines like ::
...
ATOM 1 N LYS A 1 3.246 10.041 10.379 1.00 5.28 N
ATOM 2 CA LYS A 1 2.386 10.407 9.247 1.00 7.90 C
ATOM 3 C LYS A 1 2.462 11.927 9.098 1.00 7.93 C
ATOM 4 O LYS A 1 2.582 12.668 10.097 1.00 6.28 O
ATOM 5 CB LYS A 1 0.946 9.964 9.482 1.00 3.54 C
ATOM 6 CG LYS A 1 -0.045 10.455 8.444 1.00 3.75 C
ATOM 7 CD LYS A 1 -1.470 10.062 8.818 1.00 2.85 C
ATOM 8 CE LYS A 1 -2.354 9.922 7.589 1.00 3.83 C
ATOM 9 NZ LYS A 1 -3.681 9.377 7.952 1.00 1.78 N
...
combine takeline, replace and usecols.
usecols=[6,7,8] selects the columns as x,y,z positions ::
# select all atoms
xyz = js.dA('3rn3.pdb',takeline=lambda w:w[0]=='ATOM',replace={'ATOM':1},usecols=[6,7,8])
# select only CA atoms
xyz = js.dA('3rn3.pdb',takeline=lambda w:(w[0]=='ATOM') & (w[2]=='CA'),replace={'ATOM':1},usecols=[6,7,8])
# in PDB files different atomic structures are separate my "MODEL","ENDMODEL" lines.
# We might load all by using block
xyz = js.dA('3rn3.pdb',takeline=lambda w:(w[0]=='ATOM') & (w[2]=='CA'),
replace={'ATOM':1},usecols=[6,7,8],block='MODEL')
"""
# read and parse file
# we separate values, parameters and comments and return raw data
# Returns
# ------
# list of dictionaries that will be converted to a dataArray
# [{
# 'val' :data array,
# 'para' :{list_of_parameters {'name':value}},
# 'com':['xxx','ddddddd',.....],
# 'original_comline':['xxx','ddddddd',.....]
# }]
if delimiter == '':
delimiter = None
# read the lines
zeilen = _readfile(xfile, encoding)
# convenience for takeline
if isinstance(takeline, (list, tuple, set)):
# multiple words
takelines = lambda words: any(w in words for w in takeline)
elif isinstance(takeline, str):
# single word
takelines = lambda words: takeline in words
elif callable(takeline):
# a function, catch IndexError
# noinspection PyStatementEffect
def takelines(z):
try:
return takeline(z.split(delimiter))
except IndexError:
False
else:
# anything else e.g. None
takelines = takeline
# convenience for skipping lines
if isinstance(skiplines, (list, tuple, set)):
skip = lambda words: any(w in words for w in skiplines)
elif isinstance(skiplines, str):
skip = lambda words: skiplines in words
elif callable(skiplines):
skip = skiplines
else:
skip = skiplines
# prepare output data
raw_data = [] # original read data
temp = {'com': [], '_original_comline': [], 'para': {}, 'val': []} # temporary dataset
single_words = '' # collection single words
if lines2parameter is not None:
if isinstance(lines2parameter, numbers.Number): lines2parameter = [lines2parameter]
for iline in lines2parameter:
# prepend a line string in front
zeilen[iline] = 'line_%i ' % iline + zeilen[iline]
if block is not None:
if isinstance(block, (list, tuple)):
# block has indices used as slice so convert it to slice
block = np.r_[block, [None, None, None]][:3]
block = slice(*[int(b) if isinstance(b, numbers.Number) else None for b in block])
if isinstance(block, slice):
zeilen = zeilen[block]
block = None
# block seems to be string
# to force a new dataset at the end
zeilen.append('')
# now sort it
lastlinewasnumber = False # a "last line was what" indicator
isheader = False # header at begin indicator
i = 0 # line number in original file
iz = 0 # nonempty line number in original file
for zeile in zeilen: # zeilen is german for lines
i += 1
if zeile.strip(): iz += 1 # count nonempty line
is_end = (i == len(zeilen)) # file end
if iz == 1: # is first line in file, is it header?
firstwords = zeile.split()
if len(firstwords) > 1 and firstwords[0] == '@name' and firstwords[1] == 'header_of_common_parameters':
isheader = True
# line drop
if ignore != '' and zeile.startswith(ignore): # ignore this line
continue
# take only specific lines
if takelines is not None and not is_end:
if not takelines(zeile): continue
# do char replacements in line
if isinstance(replace, dict):
for key in replace:
if isinstance(key, str):
zeile = zeile.replace(key, str(replace[key]))
else:
# key is a regular expression pattern (from re.compile)
try:
zeile = key.sub(str(replace[key]), zeile)
except AttributeError:
raise AttributeError('key in replace is not string or regular expression.')
worte = zeile.split(delimiter)
if skip is not None and skip(worte):
continue
isdataline = is_dataline(worte)
# block assignment
# test if block marker or change between data line and nondata line
# lastlinewasnumber shows status of previous line to detect change
if block is None:
if isheader and not zeile.strip():
# if isheader we append it as first
# later, the first is used as common data in dataList identified by @name content
if not temp['val']:
# with empty 'val' it would be rejected as first dataArray
temp['val'].append([])
temp, single_words = _append_temp2raw(raw_data, temp, single_words, xfile, is_end)
isheader = False
if ((isdataline and not lastlinewasnumber) or # change from nondata to data
(not isdataline and lastlinewasnumber)): # change from data to nondata
temp, single_words = _append_temp2raw(raw_data, temp, single_words, xfile, is_end)
lastlinewasnumber = True
elif zeile.startswith(block):
# a block marker is found
temp, single_words = _append_temp2raw(raw_data, temp, single_words, xfile, is_end)
lastlinewasnumber = False
# line assignment
if isdataline:
lastlinewasnumber = True
if isinstance(usecols, list):
try:
worte = [worte[ii] for ii in usecols]
except IndexError:
raise IndexError('list index out of range. There seems to be a line with missing entries! '
'usecols to large/wrong indices, try option skiplines.')
while len(worte) > len(temp['val']): # new columns needed
temp['val'].append([]) # create new column
try: # fill to last line
for row in np.arange(len(temp['val'][-2])):
temp['val'][-1].append(None)
except IndexError: # for first column no predecessor
pass
for col in range(len(worte)): # assign new data
try:
temp['val'][col].append(float(worte[col]))
except:
temp['val'][col].append(worte[col]) # replace on error (non float)
# do not change this!! sometimes data are something like u for up and d for down
continue
else:
# not a data line
lastlinewasnumber = False
if len(worte) == 0: # empty lines
continue
if len(worte) == 1: # single name
single_words += worte[0] + ' '
continue
if is_float(worte[1]) or worte[1][0] == '@' or worte[0] == '@name':
# is parameter (name number) or starts with '@'
if worte[0] in temp['para']:
num = 1
while worte[0] + str(num) in temp['para']: num += 1
keynum = str(num)
else:
keynum = ''
if worte[1][0] == '@' or worte[0] == '@name': # is link to something or is name of a link
temp['para'][worte[0] + keynum] = ' '.join(worte[1:])
elif len(worte[1:]) > 1:
temp['para'][worte[0] + keynum] = [_w2f(wort) for wort in worte[1:]]
else:
temp['para'][worte[0] + keynum] = _w2f(worte[1])
continue
else: # comment 1.+2. word not number
line = ' '.join(worte)
if line not in temp['com']: temp['com'].append(line)
if zeile != line: # store original zeile if different from line
if line not in temp['_original_comline']: temp['_original_comline'].append(zeile)
continue
print('A line didnt match a rule ' + str(i) + ' Zeile:\n' + zeile)
# append last set if not empty
_ = _append_temp2raw(raw_data, temp, single_words, xfile, True)
del zeilen
return raw_data
def _parse(filename,
index=slice(None),
usecols=None,
skiplines=None,
replace=None,
ignore='#',
XYeYeX=None,
delimiter=None,
takeline=None,
lines2parameter=None,
encoding=None,
block=None):
# This function is to be called in parallel read of multiple files during dataList creation
# in each read process the file is read, parsed, searched for internal links and dataArrays created.
# For single dataArrays use _parsefile
# read and parse file
data = _parsefile(filename,
block=block,
usecols=usecols,
skiplines=skiplines,
replace=replace,
ignore=ignore,
delimiter=delimiter,
takeline=takeline,
lines2parameter=lines2parameter,
encoding=encoding)
if len(data) == 0:
# an empty file was read
return []
# search for internal links of more complex parameters stored in same file
data = _searchForLinks(data)
# create according to index
if isinstance(data, str):
print(data)
return []
else:
# select according to index number
if isinstance(index, numbers.Integral):
data = [data[index]]
elif isinstance(index, slice):
# index is slice
data = data[index]
elif all([isinstance(a, numbers.Integral) for a in index]):
# index is a list of integer
data = [data[i] for i in index]
else:
raise TypeError('use a proper index or slice notation')
# separate common parameters if present
if data[0]['para']['@name'] == 'header_of_common_parameters':
commonParameters = data[0]['para']
data = data[1:]
else:
commonParameters = None
# return as list
return [dataArray(dat, XYeYeX=XYeYeX) for dat in data], commonParameters
def _searchForLinks(data):
"""
internal function
check for links inside data and returns a list without internal links
"""
i = 0
while i < len(data):
for parameter in data[i]['para']:
if isinstance(data[i]['para'][parameter], str) and data[i]['para'][parameter][0] == '@':
parname = data[i]['para'][parameter][1:]
for tolink in range(i + 1, len(data)):
if data[tolink]['para']['@name'] == parname:
data[i]['para'][parameter] = dataArray(data.pop(tolink))
break
i += 1
return data
def _maketxt(dataa, name=None, fmt='%.5e', exclude=[]):
"""
Converts dataArray to ASCII text
only ndarray content is stored; not dictionaries in parameters
format rules:
datasets are separated by a keyword line
given in blockempty; "empty lines" is the default
A dataset consists of comments, parameter and data (and optional to another dataset)
first two words decide for a line
string + value -> parameter[also simple list of parameter]
string + string -> comment
value + value -> data (line of an array; in sequence without break)
single words -> are appended to comments
optional:
1string+@1string -> as parameter but links to other dataArray
(content of parameter with name 1string) stored in the same
file after this dataset identified by parameter @name=1string
internal parameters starting with underscore ('_') are ignored for writing e.g._X, Y, ix
some others for internal usage too
content of @name is used as identifier or filename
passed to savetext with example for ndarray part:
fmt : str or sequence of str
A single format specifier as '%10.5e' with number of digits and precision.
If dictionaries are used add the key to name_key and store content as parameter.
"""
tail = []
partxt = []
comment = [dataa.comment] if isinstance(dataa.comment, str) else dataa.comment
comtxt = [com + '\n' for com in comment if com.strip()]
if name is not None:
setattr(dataa, '@name', str(name))
# add parameter lines
for parameter in dataa.attr:
if parameter in ['comment', 'raw_data', 'internlink', 'lastfit'] + protectedNames + exclude:
continue
if parameter[0] == '_': # exclude internals
continue
dataapar = getattr(dataa, parameter)
if isinstance(dataapar, dict):
# these are not saved
print(parameter, ' not saved; is a dictionary')
continue
if isinstance(dataapar, dataArray):
partxt += [parameter + ' @' + parameter + '\n']
tail += _maketxt(dataapar, parameter, fmt, exclude)
continue
if isinstance(dataapar, str):
partxt += [parameter + ' ' + dataapar + '\n']
continue
# noinspection PyBroadException
try:
ndataapar = np.array(dataapar).squeeze()
except:
print(parameter, ' not saved; is not a matrix format (np.array() returns error)')
continue
if isinstance(ndataapar, np.ndarray):
if ndataapar.ndim == 0:
# float as float and others as string
# TODO maybe complex numbers?
try:
partxt += [parameter + ' ' + fmt % ndataapar + '\n']
except (TypeError, ValueError):
partxt += [parameter + ' ' + '%s' % ndataapar + '\n']
elif ndataapar.ndim == 1:
try:
nfmt = (' ' + fmt) * ndataapar.shape[0]
partxt += [parameter + ' ' + nfmt % tuple(ndataapar) + '\n']
except (TypeError, ValueError):
nfmt = (' %s') * ndataapar.shape[0]
partxt += [parameter + ' ' + nfmt % tuple(ndataapar) + '\n']
elif ndataapar.ndim == 2:
partxt += [parameter + ' @' + parameter + '\n']
tail += _maketxt(dataArray(ndataapar), parameter, fmt, exclude)
else:
raise IOError('to many dimensions; only ndim<3 supported ')
# add existing columnIndices to recover them on reading
if 'XYeYeX' not in exclude:
partxt += 'XYeYeX ' + ''.join(['{} '.format(getattr(dataa, ina, '-')) for ina in protectedIndicesNames]) + '\n'
# prepare for final output
output = io.BytesIO()
# write the array as ndarray
try:
np.savetxt(output, dataa.array.T, fmt)
except TypeError:
np.savetxt(output, dataa.array.T, '%s')
datatxt = output.getvalue() # this contains '\n' at the end of each line within this single line
output.close()
# return list of byte ascii data by using encode to write later only ascii data
return [c.encode() for c in comtxt] + [p.encode() for p in partxt] + [datatxt] + tail
def shortprint(values, threshold=6, edgeitems=2):
"""
Creates a short handy representation string for array values.
Parameters
----------
values
threshold: int default 6
number of elements to switch to reduced form
edgeitems : int default 2
number of elements shown in reduced form
"""
opt = np.get_printoptions()
np.set_printoptions(threshold=threshold, edgeitems=edgeitems)
if isinstance(values, np.ndarray):
valuestr = np.array_str(values)
else:
# assume list of arrays
if len(values)<6:
valuestr = '[' + '\n'.join([np.array_str(val) for val in values]) + ']'
else:
valuestr = '[' + '\n'.join([np.array_str(val) for val in values[:edgeitems]])
valuestr += '...\n'
valuestr += '\n'.join([np.array_str(val) for val in values[:edgeitems]]) + ']'
np.set_printoptions(**opt)
return valuestr
def inheritDocstringFrom(cls):
"""
Copy docstring from parent.
"""
def docstringInheritDecorator(fn):
if isinstance(fn.__doc__, str):
prepend = fn.__doc__ + '\noriginal doc from ' + cls.__name__ + '\n'
else:
prepend = ''
if fn.__name__ in cls.__dict__:
fn.__doc__ = prepend + getattr(cls, fn.__name__).__doc__
return fn
return docstringInheritDecorator
#: Defined protected names which are not allowed as attribute names.
protectedNames = ['X', 'Y', 'eY', 'eX', 'Z', 'eZ', 'W', 'eW']
#: Indices attributes of protected names
protectedIndicesNames = ['_i' + pN.lower() for pN in protectedNames]
class attributelist(list):
"""
A list of attributes extracted from dataList elements with additional methods for easier attribute list handling.
Mainly to handle arrays with some basic properties respecting that missing values are allowed.
"""
_isatlist = True
def __init__(self, objekt):
list.__init__(self, objekt)
return
@inheritDocstringFrom(list)
def __getitem__(self, index):
if isinstance(index, numbers.Integral):
# return list item
return list.__getitem__(self, index)
elif isinstance(index, list):
# return attributelist
return attributelist([self[i] for i in index])
elif isinstance(index, np.ndarray):
# array indexing like numpy arrays
if index.dtype == np.dtype('bool'):
# this converts bool in integer indices where elements are True
index = np.r_[:len(index)][index]
return attributelist([self[i] for i in index])
elif isinstance(index, tuple):
# this includes the slicing of the underlying dataArrays whatever is in index1
index0, index1 = index[0], index[1:]
if isinstance(index0, numbers.Integral):
return self[index0][index1]
else:
return attributelist([element[index1] for element in self[index0]])
# a default if nothing of above is used
return list.__getitem__(self, index)
@property
def array(self):
"""returns ndarray if possible or list of arrays"""
try:
return np.asarray(self)
except:
return [da for da in self]
@property
def unique(self):
"""returns ndarray if possible or list of arrays"""
return np.unique(self.flatten)
@property
def flatten(self):
"""returns flattened ndarray"""
return np.hstack(self)
@property
def mean(self):
"""returns mean"""
return np.mean(self.flatten)
@property
def std(self):
"""returns standard deviation from mean"""
return np.std(self.flatten)
@property
def sum(self):
"""returns sum"""
return self.flatten.sum()
@property
def min(self):
"""minimum value"""
return np.min(self.flatten)
@property
def max(self):
"""maximum value"""
return np.max(self.flatten)
@property
def hasNone(self):
"""
This can be used to test if some dataArray elements do not have the attribute
"""
return np.any([ele is None for ele in self])
@property
def shape(self):
""" Length of attributelist and shape of elements """
return len(self), tuple([a.shape for a in self])
# This is the base dataArray class without plotting (only dummies)
# noinspection PyIncorrectDocstring,PyDefaultArgument,PyArgumentList
class dataListBase(list):
"""See init """
def __init__(self, objekt=None,
block=None,
usecols=None,
delimiter=None,
takeline=None,
index=slice(None),
replace=None,
skiplines=None,
ignore='#',
XYeYeX=None,
lines2parameter=None,
encoding=None):
r"""
A list of dataArrays with attributes for analysis, fitting and plotting.
- Allows reading, appending, selection, filter, sort, prune, least square fitting, ....
- Saves to human readable ASCII text format (optional gziped). For file format see dataArray.
- The dataList allows simultaneous fit of all dataArrays dependent on attributes.
- and with different parameters for the dataArrays (see fit).
- dataList creation parameters (below) mainly determine how a file is read from file.
- .Y are used as function values at coordinates [.X,.Z,.W] in fitting.
Parameters
----------
objekt : strings, list of array or dataArray
Objects or filename(s) to read.
- Filenames with extension '.gz' are decompressed (gzip).
- Filenames with asterisk like exda=dataList(objekt='aa12*') as input for multiple files.
- An in-memory stream for text I/O (Python3 -> io.StringIO).
lines2parameter : list of integer
List of line numbers to use as attribute with attribute name 'line_i'.
Used to mark lines with parameters without name (only numbers in a line as in .pdh files in the header).
E.g. to skip the first lines.
takeline : string,list of string, function
Filter lines to be included according to keywords or filter function.
If the first 2 words contain non-float it should be combined with: replace
(e.g. replace starting word by number {'ATOM':1} to be detected as data)
and usecols to select the needed columns.
Examples (function gets words in line):
- lambda words: any(w in words for w in ['ATOM','CA']) # one of both words somewhere in line
- lambda w: (w[0]=='ATOM') & (w[2]=='CA') # starts with 'ATOM' and third is 'CA'
For word or list of words first example is generated automatically.
skiplines : boolean function, list of string or single string
Skip if line meets condition. Function gets the list of words in a data line.
Examples:
- lambda words: any(w in words for w in ['',' ','NAN',''*****]) # remove missing data, with exact match
- lambda words: any(float(w)>3.1411 for w in words)
- lambda words: len(words)==1 # e.g. missing data with incomplete number of values
If a list is given, the lambda function is generated automatically as in above example.
If single string is given, it is tested if string is a substring of a word ( 'abc' in '123abc456')
replace : dictionary of [string,regular expression object]:string
String replacement in read lines as {'old':'new',...} (after takeline).
String pairs in this dictionary are replaced in each line.
This is done prior to determining line type and can be used to convert strings to number or ',':'.'.
If dict key is a regular expression object (e.g. rH=re.compile('H\d+') ),it is replaced by string.
See python module re for syntax.
ignore : string, default '#'
Ignore lines starting with string e.g. '#'.
delimiter : string, default any whitespace
Separator between words (data fields) in a line.
E.g. '\t' tabulator
usecols : list of integer
Use only given columns and ignore others (evaluated after skiplines).
block : string, slice (or slice indices), default None
Indicates separation of dataArray in file if multiple blocks of data are present.
- None : Auto detection of blocks according to change between datalines and non-datalines.
A new dataArray is created if data and attributes are present.
- string : If block is found at beginning of line a new dataArray is created and appended.
block can be something like "next" or the first parameter name of a new block as block='Temp'
- slice or slice indices :
block=slice(2,100,3) slices the file lines in file as lines[i:j:k] .
If only indices are given these are converted to slice.
index : integer, slice list of integer, default is a slice for all.
Selects which dataArray to use from read file if multiple are found.
Can be integer , list of integer or slice notation.
XYeYeX : list integers, default=[0,1,2,None,None,None]
Sets column indices for X, Y, eY, eX, Z, eZ, W, eW.
Change later by: data.setColumnIndex .
encoding : None, 'utf-8', 'cp1252', 'ascii',...
The encoding of the files read. By default, the system default encoding is used.
Others: python2.7 'ascii', python3 'utf-8'
For files written on Microsoft Windows use 'cp1252' (US),'cp1251' (with German öäüß)
'latin-1' codes also the first 256 ascii characters correctly.
Returns
-------
dataList : list of dataArray
Notes
-----
**Attribute access as attributelist**
Attributes of the dataArray elements can be accessed like in dataArrays by .name notation.
The difference is that a dataList returns *attributelist* -a subclass of *list*- with some additional methods
as the list of attributes in the dataList elements.
This is necessary as it is allowed that dataList elements miss an attribute (indicated as None) or
have different type. A numpy ndarray can be retrieved by the array property (as .name.array).
**Global attributes**
We have to discriminate attributes stored individual in each dataArray and in the dataList
as a kind of global attribute. dataArray attributes belong to a dataArray and are saved
with the dataArray, while global dataList attributes are only saved with
the whole dataList at the beginning of a file. If dataArrays are saved as single files global attributes
are lost.
Examples
--------
For more about usage see :ref:`Beginners Guide / Help`.
::
import jscatter as js
ex=js.dL('aa12*') # read aa files
ex.extend('bb12*') # extend with other bb files
ex.sort(...) # sort by attribute e.g. "q"
ex.prune(number=100) # reduce number of points; default is to calc the mean in an interval
ex.filter(lambda a:a.Temperature>273) # to filter for an attribute "Temperature" or .X.mean() value
# do a linear fit
ex.fit(model=lambda a,b,t:a*t+b,freepar={'a':1,'b':0},mapNames={'t':'X'})
# fit using parameters in example the Temperature stored as parameter.
ex.fit(model=lambda Temperature,b,x:Temperature*x+b,freepar={'b':0},mapNames={'x':'X'})
::
import jscatter as js
import numpy as np
t=np.r_[1:100:5];D=0.05;amp=1
# using list comprehension creating a numpy array
i5=js.dL([np.c_[t,amp*np.exp(-q*q*D*t),np.ones_like(t)*0.05].T for q in np.r_[0.2:2:0.4]])
# calling a function returning dataArrays
i5=js.dL([js.dynamic.simpleDiffusion(q,t,amp,D) for q in np.r_[0.2:2:0.4]])
# define a function and add dataArrays to dataList
ff=lambda q,D,t,amp:np.c_[t,amp*np.exp(-q*q*D*t),np.ones_like(t)*0.05].T
i5=js.dL() # empty list
for q in np.r_[0.2:2:0.4]:
i5.append(ff(q,D,t,amp))
Get elements of dataList with specific attribute values.
::
i5=js.dL([js.dynamic.simpleDiffusion(q,t,amp,D) for q in np.r_[0.2:2:0.4]])
# get q=0.6
i5[i5.q.array==0.6]
# get q > 0.5
i5[i5.q.array > 0.5]
**Rules for reading of ASCII files**
"""
self._block = block
if objekt is None:
# return empty dataList
list.__init__(self, [])
else:
# read objekt
temp = self._read_objekt(objekt,
index,
usecols=usecols,
replace=replace,
skiplines=skiplines,
ignore=ignore,
XYeYeX=XYeYeX,
delimiter=delimiter,
takeline=takeline,
lines2parameter=lines2parameter,
encoding=encoding)
if len(temp) > 0:
list.__init__(self, temp)
else:
# return empty list not to break the flow
list.__init__(self, [])
self._limits = {}
self._isdataList = True
self._constrains = []
self._errplot = None
self._bayes_sampler = None
self._freepar = None
self._mapNames = None
self._fixpar = None
self._lasterrortime = 0
self._lasterrortimecommandline = 0
self.model = None
self.numberOfModelEvaluations = 0
self._fitmethod = None
self._code = None
self._output = None
self._xslice = None
self._nozeroerror = None
self._chi2trace = []
self._ln_prior = None
# add docstring from _read
__init__.__doc__ += _parsefile.__doc__
# add docstring from __new__ to class docstring to show this in help
__doc__ = __init__.__doc__
def __getattribute__(self, attr):
"""--
"""
# priority to access attributes
# protectedNames > special list attributes > element attributes > list attributes
if attr in protectedNames + ['name']:
return attributelist([getattr(element, attr, None) for element in self])
elif attr in ['lastfit']:
return super().__getattribute__(attr)
elif attr[0] == 'º' and attr[1:] in self.dlattr():
# this will not be documented for now as it is a dirty thing
return super().__getattribute__(attr[1:])
elif np.any([attr in element.attr for element in self]):
return attributelist([getattr(element, attr, None) for element in self])
else:
return super().__getattribute__(attr)
def __getdatalistattr__(self, attr):
"""
get attributes from dataList
"""
return super().__getattribute__(attr)
def __setattr__(self, attr, val):
"""
set attribute in datList
"""
if attr not in protectedNames + ['lastfit']:
self.__setlistattr__(attr, val)
else:
raise NameError('{0} is reserved keyword '.format(attr))
def __setlistattr__(self, attr, val):
"""internal usage
this separate method to bypass __setattr__ is used
to set dataList attributes directly
"""
list.__setattr__(self, attr, val)
# noinspection PyBroadException
def __delattr__(self, attr):
"""del attribute in elements or in dataList"""
try:
for ele in self:
ele.__delattr__(attr)
except:
list.__delattr__(self, attr)
def _read_objekt(self, objekt=None,
index=slice(None),
usecols=None,
skiplines=None,
replace=None,
ignore='#',
XYeYeX=None,
delimiter=None,
takeline=None,
lines2parameter=None,
encoding=None):
"""
internal function to read data
reads data from ASCII files or already read stuff in output format of "_parsefile"
and returns simple dataArray list
see _parsefile for details of parameters
"""
# check input objekt what it is and return list of dataArray
if isinstance(objekt, dataList):
return objekt[index]
elif isinstance(objekt, dataArray):
return [objekt]
elif isinstance(objekt, np.ndarray):
return [dataArray(objekt, XYeYeX=XYeYeX)]
elif isinstance(objekt, dict):
# single element from _parsefile
if 'val' in objekt:
return [dataArray(objekt, XYeYeX=XYeYeX)]
else:
warnings.warn('Nothing useful found in "' + str(objekt) + '"')
elif isinstance(objekt, (list, tuple)):
# call recursively
datalist = []
for obj in objekt:
datalist.extend(self._read_objekt(obj, index=index, usecols=usecols, replace=replace,
skiplines=skiplines, ignore=ignore, XYeYeX=XYeYeX,
delimiter=delimiter, takeline=takeline,
lines2parameter=lines2parameter, encoding=encoding))
return datalist
else:
# get filenames and parse them all
filelist = glob.glob(objekt)
if len(filelist) == 0:
raise ValueError('No file with ', objekt, ' pattern found. ')
# parse all returning a list of lists and sort into a single list
datalist = []
# the following parallel read is 9times slower for <100 files and for 10000 files equal (at least on a SSD)
# try:
# # assert that we are not in a subprocess
# assert multiprocessing.current_process().name == 'MainProcess'
# parsed = parallel.doForList(_parse, filelist, loopover='filename', index=index, usecols=usecols,
# skiplines=skiplines, replace=replace, ignore=ignore,
# XYeYeX=XYeYeX, delimiter=delimiter, takeline=takeline,
# lines2parameter=lines2parameter, encoding=encoding, block=self._block,
# output=False, ncpu=0)
# except (AssertionError, pickle.PicklingError) as e:
# # To catch if we are in a subprocess
parsed = parallel.doForList(_parse, filelist, loopover='filename', index=index, usecols=usecols,
skiplines=skiplines, replace=replace, ignore=ignore,
XYeYeX=XYeYeX, delimiter=delimiter, takeline=takeline,
lines2parameter=lines2parameter, encoding=encoding, block=self._block,
output=False, ncpu=1) # ncpu=1 avoids multiprocessing
for data, common in parsed:
if common:
# if common parameters are found add them to self (will overwrite)
_ = common.pop('@name', None)
for k, v in common.items():
setattr(self, k, v)
datalist.extend(data)
if len(datalist) == 0:
warnings.warn('Nothing useful found in file(s) with input "' + str(objekt) + '"')
return datalist
@inheritDocstringFrom(list)
def __setitem__(self, index, objekt, i=0, usecols=None):
"""puts the objekt into self
needs to be a dataArray object
"""
if isinstance(objekt, dataArray):
list.__setitem__(self, index, objekt)
else:
raise TypeError('not a dataArray object')
@inheritDocstringFrom(list)
def __getitem__(self, index):
if isinstance(index, numbers.Integral):
return list.__getitem__(self, index)
elif isinstance(index, list):
out = dataList([self[i] for i in index])
return out
elif isinstance(index, np.ndarray):
if index.dtype == np.dtype('bool'):
# this converts bool in integer indices where elements are True
index = np.r_[:len(index)][index]
out = dataList([self[i] for i in index])
return out
elif isinstance(index, tuple):
# this includes the slicing of the underlying dataArrays whatever is in index1
index0, index1 = index[0], index[1:]
if isinstance(index0, numbers.Integral):
# return the single dataArray but sliced by index1
out = self[index0][index1]
else:
out = [element[index1] for element in self[index0]]
if np.all([hasattr(element, '_isdataArray') for element in out]):
out = dataList(out)
return out
out = dataList(list.__getitem__(self, index))
return out
@inheritDocstringFrom(list)
def __delitem__(self, index):
list.__delitem__(self, index)
@inheritDocstringFrom(list)
def __setslice__(self, i, j, objekt):
self[max(0, i):max(0, j):] = objekt
@inheritDocstringFrom(list)
def __delslice__(self, i, j):
del self[max(0, i):max(0, j):]
# TODO remove getslice
@inheritDocstringFrom(list)
def __getslice__(self, i, j):
return self[max(0, i):max(0, j):]
@inheritDocstringFrom(list)
def __add__(self, other):
if hasattr(other, '_isdataList'):
out = dataList(list.__add__(self, other))
elif hasattr(other, '_isdataArray'):
out = dataList(list.__add__(self, [other]))
else:
out = dataList(list.__add__(self, [dataArray(other)]))
return out
def __getstate__(self):
"""
Needed to remove model and _code from dict for serialization (pickle)
if these cannot be serialized because model was defined as lambda or not defined at top level of module.
"""
state = self.__dict__.copy()
if 'model' in state:
try:
state['model'] = pickle.dumps(state['model'])
except (pickle.PicklingError, AttributeError):
state['model'] = 'removed during serialization, not pickabel function'
if '_code' in state:
del state['_code']
return state
def __deepcopy__(self, memo):
cls = self.__class__
result = cls([copy.deepcopy(da, memo) for da in self])
memo[id(self)] = result
for k, v in self.__dict__.items():
# copy only attributes in dataList
if k[0] != '_': # but nothing private
# bypass setattr to include lastfit
result.__setlistattr__(k, copy.deepcopy(v, memo))
return result
def copy(self):
"""
Deepcopy of dataList
To make a normal shallow copy use copy.copy
"""
return copy.deepcopy(self)
def nakedCopy(self):
"""
Returns copy without attributes, thus only the data.
"""
cls = self.__class__
return cls([ele.nakedCopy() for ele in self])
@property
def whoHasAttributes(self):
"""
Lists which attribute is found in which element.
Returns
-------
dictionary of attributes names: list of indices
keys are the attribute names
values are indices of dataList where attr is existent
"""
attrInElements = set()
for ele in self: attrInElements.update(ele.attr)
whohasAttribute = {}
for attr in attrInElements:
whohasAttribute[attr] = [i for i, j in enumerate(getattr(self, attr)) if j is not None]
return whohasAttribute
@property
def shape(self):
"""
Tuple with shapes of dataList elements.
"""
return tuple([a.shape for a in self])
def dlattr(self, attr=None):
"""
Get attribute or list of existing attribute names excluding common attributes from dataArrays.
Parameters
----------
attr : string
Name of dataList attribute to return.
If None a list of all attribute names is returned
"""
if attr is None:
attrlist = [k for k in list(super().__getattribute__('__dict__')) if (k[0] != '_')
and (k not in ('@name', 'raw_data'))]
return sorted(attrlist)
else:
return super().__getattribute__(attr)
@property
def attr(self):
"""
Returns all attribute names (including commonAttr of elements) of the dataList.
"""
attr = [k for k in list(super().__getattribute__('__dict__')) + self.commonAttr
if (k[0] != '_') and (k not in ('@name', 'raw_data'))]
return sorted(attr)
# noinspection PyBroadException
def showattr(self, maxlength=75, exclude=None):
"""
Show data specific attributes for all elements.
Parameters
----------
maxlength : integer
Truncate string representation
exclude : list of str
List of attribute names to exclude from show
"""
if exclude is None:
exclude = ['comment', 'lastfit']
for element in self:
print('------------------------------------------------')
element.showattr(maxlength=maxlength, exclude=exclude)
print('==================================================')
for attr in self.dlattr():
if attr not in exclude:
values = super().__getattribute__(attr)
try:
valstr = shortprint(values).split('\n')
print('{:>24} = {:}'.format(attr, valstr[0]))
for vstr in valstr[1:]:
print('{:>25} {:}'.format('', vstr))
except:
print('%24s = %s' % (attr, str(values)[:maxlength]))
print('------------------------------------------------')
def copyattr2elements(self, maxndim=1, exclude=['comment']):
"""
Copy dataList specific attributes to all elements.
Parameters
----------
exclude : list of str
List of attr names to exclude from show
maxndim : int, default 2
Maximum dimension e.g. to prevent copy of 2d arrays like covariance matrix
Notes
-----
Main use is for copying fit parameters.
"""
commonAttr = self.commonAttr
for attr in self.dlattr():
if attr not in exclude + protectedNames + ['lastfit', 'raw_data']:
val = super().__getattribute__(attr)
if (hasattr(val, '__iter__') and len(val) == len(self)) and attr[0] != '_':
for ele, va in zip(self, val):
if np.array(va).ndim <= maxndim:
setattr(ele, attr, va)
else:
for ele in self:
if np.array(val).ndim <= maxndim:
setattr(ele, attr, val)
def getfromcomment(self, attrname, convert=None, ignorecase=False):
"""
Extract a non number parameter from comment with attrname in front
If multiple names start with parname first one is used.
Used comment line is deleted from comments
Parameters
----------
attrname : string without spaces
Name of the parameter in first place.
convert : function
Function to convert the remainder of the line to the desired attribut value. E.g. ::
# line "Frequency - 3.141 MHz "
.getfromcomment('Frequency',convert=lambda a: float(a.split()[1]))
ignorecase : bool
Ignore case of attrname.
Notes
-----
A more complex example with unit conversion ::
f={'GHz':1e9,'MHz':1e6,'KHz':1e3,'Hz':1}
# line "Frequency - 3.141 MHz "
.getfromcomment('Frequency',convert=lambda a: float(a.split()[1]) * f.get(a.split()[2],1))
"""
for element in self:
element.getfromcomment(attrname=attrname, convert=convert, ignorecase=ignorecase)
# noinspection PyBroadException
@property
def commonAttr(self):
"""
Returns list of attribute names existing in elements.
"""
common = []
try:
for attr in self[0].attr:
if np.all([attr in element.attr for element in self]):
common.append(attr)
except:
return []
return common
@property
def names(self):
"""
List of element names.
"""
return [element.name for element in self]
@inheritDocstringFrom(list)
def append(self, objekt=None,
index=slice(None),
usecols=None,
skiplines=None,
replace=None,
ignore='#',
XYeYeX=None,
delimiter=None,
takeline=None,
lines2parameter=None,
encoding=None):
r"""
Reads/creates new dataArrays and appends to dataList.
See dataList for description of all keywords.
If objekt is dataArray or dataList all options except XYeYeX,index are ignored.
Parameters
----------
objekt,index,usecols,skiplines,replace, ignore,delimiter,takeline,lines2parameter : options
See dataArray or dataList
"""
obj = self._read_objekt(objekt, index=index, usecols=usecols, skiplines=skiplines,
replace=replace, ignore=ignore, XYeYeX=XYeYeX, delimiter=delimiter,
takeline=takeline, lines2parameter=lines2parameter, encoding=encoding)
list.extend(self, obj)
# extend is same as append
extend = append
@inheritDocstringFrom(list)
def insert(self, i, objekt=None,
index=0,
usecols=None,
skiplines=None,
replace=None,
ignore='#',
XYeYeX=None,
delimiter=None,
takeline=None,
lines2parameter=None,
encoding=None):
"""
Reads/creates new dataArrays and inserts in dataList.
If objekt is dataArray or dataList all options except XYeYeX,index are ignored.
Parameters
----------
i : int, default 0
Position where to insert.
objekt,index,usecols,skiplines,replace,ignore,delimiter,takeline,lines2parameter : options
See dataArray or dataList
"""
obj = self._read_objekt(objekt, usecols=usecols, skiplines=skiplines,
replace=replace, ignore=ignore, XYeYeX=XYeYeX, delimiter=delimiter,
takeline=takeline, lines2parameter=lines2parameter, encoding=encoding)
list.insert(self, i, obj[index])
@inheritDocstringFrom(list)
def pop(self, i=-1):
""" """
out = list.pop(self, i)
return out
@inheritDocstringFrom(list)
def delete(self, index):
"""
Delete element at index
"""
self.__delitem__(self, index)
@inheritDocstringFrom(list)
def index(self, value, start=0, stop=-1):
""" """
for i in range(len(self[start:stop])):
if self[i] is value:
return i
raise ValueError('not in list')
@property
def aslist(self):
"""
Return as simple list.
"""
return [ele for ele in self]
@inheritDocstringFrom(list)
def reverse(self):
"""Reverse dataList -> INPLACE!!!"""
list.reverse(self)
# noinspection PyMethodOverriding
def sort(self, key=None, reverse=False):
"""
Sort dataList -> INPLACE!!!
Parameters
----------
key : function
A function that is applied to all elements and the output is used for sorting.
e.g. 'Temp' or lambda a:a.Temp
convenience: If key is attributename this attribute is used
reverse : True, False
Normal or reverse order.
Examples
--------
::
dlist.sort('q',True)
dlist.sort(key=lambda ee:ee.X.mean() )
dlist.sort(key=lambda ee:ee.temperatur )
dlist.sort(key=lambda ee:ee.Y.mean() )
dlist.sort(key=lambda ee:ee[:,0].sum() )
dlist.sort(key=lambda ee:getattr(ee,parname))
dlist.sort(key='parname')
"""
if isinstance(key, str):
self.sort(key=lambda ee: getattr(ee, key), reverse=reverse)
return
try:
list.sort(self, key=key, reverse=reverse)
except ValueError:
print('You have to define how to compare dataList elements for sorting; see help\n')
@inheritDocstringFrom(list)
def __repr__(self):
if len(self) > 0:
attr = self.commonAttr[:7]
shape = np.shape(self)
if all([sh == shape[0] for sh in shape[1:]]):
shape = 'all ==> ' + str(shape[0])
elif len(shape) > 12:
shape = shape[:5] + ('...', '...') + shape[-5:]
desc = """dataList->\nX = {0}\n,Y = {1},\nfirst attributes={2}...,\nshape=[{3}] {4} """
return desc.format(shortprint(self.X.array), shortprint(self.Y.array), attr, len(self), shape)
else:
return """dataList-> empty"""
@property
def dtype(self):
"""return dtype of elements"""
return [element.dtype for element in self]
def filter(self, filterfunction=None, **kwargs):
"""
Filter elements according to filterfunction and kwargs.
Parameters
----------
filterfunction : function or lambda function returning boolean
Return those items of sequence for which function(item) is true.
kwargs :
Any given keyword with value is combined with filterfunction (logical AND).
Examples
--------
::
i5=js.dL('exampleData/iqt_1hho.dat')
i1=i5.filter(lambda a:a.q>0.1)
i1=i5.filter(lambda a:(a.q>0.1) )
i5.filter(lambda a:(a.q>0.1) & (a.average[0]>1)).average
i5.filter(lambda a:(max(a.q*a.X)>0.1) & (a.average[0]>1))
# with kwargs
i5.filter(q=0.5,conc=1)
"""
if not filterfunction:
filterkwargs = lambda a: np.all([getattr(a, k) == v for k, v in kwargs.items()])
return dataList([item for item in self if filterkwargs(item)])
elif not kwargs:
return dataList([item for item in self if filterfunction(item)])
else:
# with kwargs generate from keywords
filterkwargs = lambda a: np.all([getattr(a, k) == v for k, v in kwargs.items()])
return dataList([item for item in self if filterfunction(item) & filterkwargs(item)])
def setColumnIndex(self, *arg, **kwargs):
"""
Set the columnIndex where to find X,Y,Z,W, eY, eX, eZ
Default is ix=0,iy=1,iey=2,iz=None,iex=None,iez=None,iw=None,iew=None as it is the most used.
There is no limitation and each dataArray can have different ones.
Parameters
----------
ix,iy,iey,iex,iz,iez,iw,iew : integer, None, default= [0,1,2,None,None,None,None,None]
Set column index, where to find X, Y, eY.
- Default from initialisation is ix,iy,iey,iex,iz,iez,iw,iew=0,1,2,None,None,None,None,None.
(Usability wins iey=2!!)
- If first ix is dataArray the ColumnIndex is copied, others are ignored.
- If first ix is list [0,1,3] these are used as [ix,iy,iey,iex,iz,iez,iw,iew].
Remember that negative indices always are counted from back,
which changes the column when adding a new column.
Notes
-----
- integer column index as 0,1,2,-1 , should be in range
- None as not used e. g. iex=None -> no errors for x
- anything else does not change
Shortcut sCI
Examples
--------
::
data.setColumnIndex(ix=2,iy=3,iey=0,iex=None)
# remove y error in (only needed if 3rd column present)
data.setColumnIndex(iey=None)
# add Z, W column for 3D data
data.setColumnIndex(ix=0, iz=1, iw=2, iy=3)
"""
for element in self:
element.setColumnIndex(*arg, **kwargs)
sCI = setColumnIndex
def savetxt(self, name=None, exclude=[], fmt='%.5e'):
"""
Saves dataList to ASCII text file, optional compressed (gzip).
- Saves dataList with attributes to one file that can be reread retrieving data and attributes.
- Dynamic created attributes as e.g. X, Y, eY, are not saved.
- If name extension is '.gz' the file is compressed (gzip).
Parameters
----------
name : string
Filename
exclude : list of str, default []
List of attribute names to exclude from being saved.
- To exclude dataList attributes (in beginning of dataList files) `exclude = ['XYeYeX']` which also excludes
the specific lines to recover columnIndex ("XYeYeX 1 2 3 - - - - ").
- To exclude all attributes `exclude = data.attr`
fmt : string, default '%.5e'
Format specifier for writing float as e.g. '%.5e' is exponential with 5 digits precision.
Notes
-----
**Format rules**:
dataList/dataArray in ASCII text files consists of tabulated data with attributes and comments.
Multiple dataArrays are separated by empty lines,
attributes and comments come before corresponding data.
First two strings decide for a line if reread:
- string + value -> attribute as attribute name + list of values
- string + string -> comment line
- value + value -> data (line of an array; in sequence without break)
- single words -> are comment line
optional:
- Attributes containing a dataArray or a 2D ndarray are linked by `value = @attrname`
pointing to a later dataArray with `.name = attrname`
- internal parameters starting with underscore ('_') are ignored for writing, also X,Y,Z,eX,eY,eZ and lastfit
- only ndarray content is stored; no dictionaries in parameters.
- @name is used as identifier or filename, it can be accessed as `.name`.
- attributes of dataList are saved as common attributes marked with a line "@name header_of_common_parameters"
Saving only the data without attributes use numpy.savetxt.
Remember that all attribute information is lost.
::
np.savetxt('filename.dat',i5[0].array.T, fmt='%.5e')
"""
if isinstance(exclude, str):
exclude = [exclude]
if name is None:
raise IOError('filename for dataset missing! first original name in list is ', getattr(self[0], '@name'))
if os.path.splitext(name)[-1] == '.gz':
_open = gzip.open
else: # normal file
_open = open
with _open(name, 'wb') as f:
# prepend dataList attr if present
if self.dlattr() and 'XYeYeX' not in exclude:
temp = dataArray(np.array([[0, 0, 0]]).T)
for attr in self.dlattr():
if attr not in exclude:
setattr(temp, attr, super().__getattribute__(attr))
f.writelines(_maketxt(temp, name='header_of_common_parameters', fmt=fmt, exclude=exclude))
f.writelines(['\n'.encode()]) # .encode converts to byte
for element in self:
f.writelines(_maketxt(element, name=name, fmt=fmt, exclude=exclude))
f.writelines(['\n'.encode()])
return
savetext = savetxt
save = savetxt
# noinspection PyNoneFunctionAssignment
def merge(self, indices, isort=None, missing=None):
"""
Merges elements of dataList.
The merged dataArray is stored in the lowest indices. Others are removed.
Parameters
----------
indices : integer,'all'
List of indices to merge
'all' merges all elements into one.
isort : integer, default None
Sort after merge along specified column e.g.isort='X', 'Y', or 0,1,2
missing : None, 'error', 'drop', 'skip', 'first' default=None
Determines how to deal with missing attributes.
- Insert None
- Raise AttributeError
- 'drop' attribute value for missing
- 'skip' attribute for all
- Use 'first' value
Notes
-----
Attributes are copied as lists in the merged dataArray.
"""
if indices == 'all':
indices = range(len(self))
index = list(indices)
index.sort(reverse=True)
first = index.pop()
# keep this and don't change because of inspection
self[first] = self[first].merge([self[i] for i in index], isort=isort, missing=missing)
for this in index:
self.__delitem__(this)
def mergeAttribut(self, parName, limit=None, isort=None, func=np.mean):
"""
Merges elements of dataList if attribute values are closer than limit (in place).
If attribute is list the average is taken for comparison.
For special needs create new parameter and merge along this.
Parameters
----------
parName : string
name of a parameter
limit : float
The relative limit value.
If limit is None limit is determined as standard deviation of sorted differences
as limit=np.std(np.array(data.q[:-1])-np.array(data.q[1:]))/np.mean(np.array(self.q)
isort : 'X', 'Y' or 0,1,2..., default None
Column for sort.
func : function or lambda, default np.mean
A function to create a new value for parameter.
see extractAttribut
stored as .parName+str(func.func_name)
Examples
--------
::
i5=js.dL('exampleData/iqt_1hho.dat')
i5.mergeAttribut('q',0.1)
# use qmean instead of q or calc the new value
print(i5.qmean)
"""
self.sort(key=parName)
if limit is None:
try:
# relative standard deviation of the parameter differences as limit
parval = getattr(self, parName)
limit = np.std(np.diff(parval)) / parval.mean
except:
raise TypeError('cannot determine limit; please specify')
# define a criterion for merging dataset
def allwithinlimit(ml, limit):
return abs(np.std(ml)) < limit * np.mean(ml)
mergelist = [0] # a first value to start
while mergelist[-1] < (len(self) - 1):
# append if still within limits
if allwithinlimit([getattr(self[ml], parName) for ml in mergelist + [mergelist[-1] + 1]], limit):
mergelist += [mergelist[-1] + 1]
elif len(mergelist) == 1:
# only one element; no merge but parname should be a list as the others
setattr(self[mergelist[-1]], parName, [getattr(self[mergelist[-1]], parName)])
# next element for test in list
mergelist = [mergelist[0] + 1]
else:
# mergelist >1 so merge and start next element
self.merge(mergelist, isort=isort)
mergelist = [mergelist[0] + 1]
# care about last element if it was a single one
if len(mergelist) > 1:
self.merge(mergelist, isort=isort)
else:
setattr(self[mergelist[-1]], parName, [getattr(self[mergelist[-1]], parName)])
# extract with func from the merged
if func is not None:
self.extractAttribut(parName, func=func, newParName=parName + str(func.__name__))
def extractAttribut(self, parName, func=None, newParName=None):
"""
Extract a simpler attribute from a complex attribute in each element of dataList.
e.g. extract the mean value from a list in an attribute
Parameters
----------
parName : string
Name of the parameter to process
func : function or lambda
A function (e.g. lambda ) that creates a new content for the
parameter from the original content
e.g. lambda a:np.mean(a)*5.123
The function gets the content of parameter whatever it is
newParName :string
New parname, if None old parameter is overwritten.
"""
if newParName is None:
for element in self:
setattr(element, parName, func(getattr(element, parName)))
else:
for element in self:
setattr(element, newParName, func(getattr(element, parName)))
# noinspection PyUnboundLocalVariable
def bispline(self, func=None, invfunc=None, tx=None, ta=None, deg=[3, 3], eps=None, addErr=False, **kwargs):
"""
Weighted least-squares bivariate spline approximation for interpolation of Y
at given attribute values for X values.
Uses scipy.interpolate.LSQBivariateSpline .
Weights are (1/eY**2) if .eY is present.
Parameters
----------
kwargs :
Keyword arguments
The first keyword argument found as attribute is used for interpolation.
E.g. conc=0.12 defines the attribute 'conc' to be interpolated to 0.12
Special kwargs see below.
X : array
List of X values were to evaluate.
If X not given the .X of first element are used as default.
func : numpy ufunction or lambda
Simple function to be used on Y values before interpolation.
see dataArray.polyfit
invfunc : numpy ufunction or lambda
To invert func after extrapolation again.
tx,ta : array like, None, int
Strictly ordered 1-D sequences of knots coordinates for X and attribute.
If None the X or attribute values are used.
If integer<len(X or attribute) the respective number of equidistant points
in the interval between min and max are used.
deg : [int,int], optional
Degrees of the bivariate spline for X and attribute. Default is 3.
If single integer given this is used for both.
eps : float, optional
A threshold for determining the effective rank of an over-determined
linear system of equations. `eps` should have a value between 0 and 1,
the default is 1e-16.
addErr : bool
If errors are present spline the error column and add it to the result.
Returns
-------
dataArray
Notes
-----
- The spline interpolation results in a good approximation if the data are narrow.
Around peaks values are underestimated if the data are not dense enough as the
flank values are included in the spline between the maxima. See Examples.
- Without peaks there should be no artifacts.
- To estimate new errors for the spline data use .setColumnIndex(iy=ii,iey=None) with ii as index of errors.
Then spline the errors and add these as new column.
- Interpolation can not be as good as fitting with a prior known
model and use this for extrapolating.
Examples
--------
::
import jscatter as js
import numpy as np
import matplotlib.pyplot as plt
from mpl_toolkits.mplot3d import Axes3D
fig = plt.figure()
ax1 = fig.add_subplot(211, projection='3d')
ax2 = fig.add_subplot(212, projection='3d')
i5=js.dL([js.formel.gauss(np.r_[-50:50:5],mean,10) for mean in np.r_[-15:15.1:3]])
i5b=i5.bispline(mean=np.r_[-15:15:1],X=np.r_[-25:25:1],tx=10,ta=5)
fig.suptitle('Spline comparison with different spacing of data')
ax1.set_title("Narrow spacing result in good interpolation")
ax1.scatter3D(i5.X.flatten, np.repeat(i5.mean,[x.shape[0] for x in i5.X]), i5.Y.flatten,s=20,c='red')
ax1.scatter3D(i5b.X.flatten,np.repeat(i5b.mean,[x.shape[0] for x in i5b.X]), i5b.Y.flatten,s=2)
ax1.tricontour(i5b.X.flatten,np.repeat(i5b.mean,[x.shape[0] for x in i5b.X]), i5b.Y.flatten)
i5=js.dL([js.formel.gauss(np.r_[-50:50:5],mean,10) for mean in np.r_[-15:15.1:15]])
i5b=i5.bispline(mean=np.r_[-15:15:1],X=np.r_[-25:25:1])
ax2.set_title("Wide spacing result in artifacts between peaks")
ax2.scatter3D(i5.X.flatten, np.repeat(i5.mean,[x.shape[0] for x in i5.X]), i5.Y.flatten,s=20, c='red')
ax2.scatter3D(i5b.X.flatten,np.repeat(i5b.mean,[x.shape[0] for x in i5b.X]), i5b.Y.flatten,s=2)
ax2.tricontour(i5b.X.flatten,np.repeat(i5b.mean,[x.shape[0] for x in i5b.X]), i5b.Y.flatten)
plt.show(block=False)
# fig.savefig(js.examples.imagepath+'/bispline.jpg')
.. image:: ../../examples/images/bispline.jpg
:align: center
:width: 50 %
:alt: bispline
"""
if 'X' in kwargs:
X = np.atleast_1d(kwargs['X'])
else:
X = self[0].X
if isinstance(deg, numbers.Integral):
deg = [deg, deg]
par = None
for kw, val in kwargs.items():
if kw == 'X':
continue
if kw in self.attr:
par = kw
newparval = np.atleast_1d(val)
newparval.sort()
break
uniqueX = self.X.unique
if isinstance(tx, numbers.Integral) and tx < uniqueX.shape[0]:
tx = np.r_[uniqueX.min():uniqueX.max():tx * 1j]
if tx is None:
tx = uniqueX
uniquepar = getattr(self, par).unique
if isinstance(ta, numbers.Integral) and ta < uniquepar.shape[0]:
ta = np.r_[uniquepar.min():uniquepar.max():ta * 1j]
if ta is None:
ta = uniquepar
# create par coordinate P with shape of .X
P = np.repeat(getattr(self, par), [x.shape[0] for x in self.X])
if np.all(self.eY):
w = 1 / self.eY.flatten ** 2 # error weight
else:
w = None
Y = self.Y.flatten
if func is not None:
Y = func(Y)
f = scipy.interpolate.LSQBivariateSpline(x=self.X.flatten, y=P, z=Y, tx=tx, ty=ta,
w=w, kx=deg[0], ky=deg[1], eps=eps)
# get new values
fY = f(X, newparval)
if invfunc is not None:
fY = invfunc(fY)
if addErr and w is not None:
ferr = scipy.interpolate.LSQBivariateSpline(x=self.X.flatten, y=P, z=self.eY.flatten, tx=tx, ty=ta,
kx=deg[0], ky=deg[1], eps=eps)
eY = ferr(X, newparval)
else:
eY = np.zeros_like(fY)
# prepare output dataList
result = dataList()
for p, fy, e in zip(newparval, fY.T, eY.T):
if addErr and w is not None:
result.append(np.c_[X, fy, e].T)
else:
result.append(np.c_[X, fy].T)
setattr(result[-1], par, p)
return result
# noinspection PyUnboundLocalVariable
def interpolate(self, func=None, invfunc=None, deg=1, col='Y', **kwargs):
"""
2D interpolation of .Y values at new attribute and .X values using piecewise spline interpolation.
Uses twice an interpolation (first along .X then along attribute).
Common and equal attributes are copied automatically to the interpolated dataList.
Parameters
----------
**kwargs : keyword arguments with float or array-like values.
The first keyword argument found as attribute is used for interpolation.
E.g. conc=0.12 defines the attribute 'conc' to be interpolated to 0.12
X : array
List of new X values were to evaluate (linear interpolation for X).
If X < or > self.X the corresponding min/max border is used.
If X not given the .X of first dataList element are used as default.
col : index or char for column, default 'Y'
Which column to interpolate. Can be column index
func : function or lambda
Function to be used on Y values before interpolation.
See dataArray.polyfit.
invfunc : function or lambda
To invert func after extrapolation again.
deg : [int,int], optional
Degrees of the spline for X and attribute. Default is 1 for linear.
If single integer given this is used for both.
Outliers result in Nan. See scipy.interpolate.interp1d for more options.
Returns
-------
dataArray
Notes
-----
- Values outside the range of .X, .Y are extrapolated.
Values outside the range of the interpolated attribute are also extrapolated.
Both migth produce strange result if to far away.
- This interpolation results in a good approximation if the data are narrow.
Around peaks values are underestimated if the data are not dense enough. See Examples.
- To estimate new errors for the spline data use .setColumnIndex(iy=ii,iey=None) with ii as index of errors.
Then spline the errors and add these as new column.
- Interpolation can not be as good as fitting with a prior known
model and use this for extrapolating.
Examples
--------
::
import jscatter as js
import numpy as np
import matplotlib.pyplot as plt
from mpl_toolkits.mplot3d import Axes3D
fig = plt.figure()
ax1 = fig.add_subplot(211, projection='3d')
ax2 = fig.add_subplot(212, projection='3d')
# try different kinds of polynominal degree
deg=2
# generate some data (gaussian with mean)
i5=js.dL([js.formel.gauss(np.r_[-50:50:5],mean,10) for mean in np.r_[-15:15.1:3]])
# interpolate for several new mean values and new X values
i5b=i5.interpolate(mean=np.r_[-15:20:1],X=np.r_[-25:25:1],deg=deg)
#
fig.suptitle('Interpolation comparison with different spacing of data')
ax1.set_title("Narrow spacing result in good interpolation")
ax1.scatter3D(i5.X.flatten, np.repeat(i5.mean,[x.shape[0] for x in i5.X]), i5.Y.flatten,s=20,c='red')
ax1.scatter3D(i5b.X.flatten,np.repeat(i5b.mean,[x.shape[0] for x in i5b.X]), i5b.Y.flatten,s=2)
ax1.tricontour(i5b.X.flatten,np.repeat(i5b.mean,[x.shape[0] for x in i5b.X]), i5b.Y.flatten)
#
i5=js.dL([js.formel.gauss(np.r_[-50:50:5],mean,10) for mean in np.r_[-15:15.1:15]])
i5b=i5.interpolate(mean=np.r_[-15:20:1],X=np.r_[-25:25:1],deg=deg)
#
ax2.set_title("Wide spacing result in artifacts between peaks")
ax2.scatter3D(i5.X.flatten, np.repeat(i5.mean,[x.shape[0] for x in i5.X]), i5.Y.flatten,s=20,c='red')
ax2.scatter3D(i5b.X.flatten,np.repeat(i5b.mean,[x.shape[0] for x in i5b.X]), i5b.Y.flatten,s=2)
ax2.tricontour(i5b.X.flatten,np.repeat(i5b.mean,[x.shape[0] for x in i5b.X]), i5b.Y.flatten)
plt.show(block=False)
"""
interp1d = scipy.interpolate.interp1d
if isinstance(deg, numbers.Integral):
deg = [deg, deg]
if 'X' in kwargs:
X = np.atleast_1d(kwargs['X'])
del kwargs['X']
else:
X = self[0].X
for kw, val in kwargs.items():
if kw in self.attr:
par = kw
newparval = np.atleast_1d(val)
break
raise ValueError('No parameter as given found in data. Check with .attr')
# first interpolate each to new X values
if func is not None:
YY = np.array([interp1d(ele.X, func(ele[col]), kind=deg[0], fill_value='extrapolate')(X) for ele in self])
else:
YY = np.array([interp1d(ele.X, ele[col], kind=deg[0], fill_value='extrapolate')(X) for ele in self])
# attribute array
parval = getattr(self, par).flatten
# calc the poly coefficients for all YY and call it with newparval
newY = interp1d(parval, YY.T, kind=deg[1], fill_value='extrapolate')(newparval)
if invfunc is not None:
newY = invfunc(newY)
result = dataList()
for p, fy in zip(newparval, newY.T):
result.append(np.c_[X, fy].T)
# set new attribute at interpolated value
setattr(result[-1], par, p)
# add attributes that are common to all dataList elements
for attr in self.commonAttr:
if attr == par:
continue
attrlist = getattr(self, attr)
# test if all common Attributes are equal by counting first
try:
# attributes which can simply be compared/counted
if attrlist.count(attrlist[0]) == len(attrlist):
for res in result:
setattr(res, attr, attrlist[0])
except ValueError:
# in case it was a numpy array
if all((np.array_equal(attrlist[0], a) for a in attrlist[1:])):
for res in result:
setattr(res, attr, attrlist[0])
return result
def polyfit(self, func=None, invfunc=None, xfunc=None, invxfunc=None, exfunc=None, **kwargs):
r"""
Inter/Extrapolated .Y values along attribute for all given X values using a polyfit.
To extrapolate along an attribute using twice a polyfit (first along X then along attribute).
E.g. from a concentration series to extrapolate to concentration zero.
Parameters
----------
**kwargs :
Keyword arguments
The first keyword argument found as attribute is used for extrapolation
e.g. q=0.01 attribute with values where to extrapolate to
Special kwargs see below.
X : arraylike
list of X values were to evaluate
func : function or lambda
Function to be used in Y values before extrapolating.
See Notes.
invfunc : function or lambda
To invert function after extrapolation again.
xfunc : function or lambda
Function to be used for X values before interpolating along X.
invxfunc : function or lambda
To invert xfunction again.
exfunc : function or lambda
Weight for extrapolating along X
degx,degy : integer default degx=0, degy=1
polynom degree for extrapolation in x,y
If degx=0 (default) no extrapolation for X is done and values are linear interpolated.
Returns
-------
dataArray
Notes
-----
funct/invfunc is used to transfer the data to a simpler smoother or polynominal form.
- Think about data describing diffusion like :math:`I=exp(-q^2Dt)` and we want to interpolate along attribute q.
If funct is np.log we interpolate on a simpler parabolic q**2 and linear in t.
- Same can be done with X axis e.g. for subdiffusion :math:`I=exp(-q^2Dt^a) \ with \ a < 1`.
Examples
--------
::
# Task: Extrapolate to zero q for 3 X values for an exp decaying function.
# Here first log(Y) is used (problem linearized), then linear extrapolate and exp function used for the result.
# This is like lin extrapolation of the exponent.
#
i5.polyfit(q=0,X=[0,1,11],func=lambda y:np.log(y),invfunc=lambda y:np.exp(y),deg=1)
#
# Concentration data with conc and extrapolate to conc=0.
data.polyfit(conc=0,X=data[0].X,deg=1)
#
# Interpolate for specified X and a list of attributes. ::
i5=js.dL(js.examples.datapath+'/iqt_1hho.dat')
i5.polyfit(X=np.r_[1:5.1],q=i5.q)
"""
if 'X' in kwargs:
X = np.atleast_1d(kwargs['X'])
del kwargs['X']
else:
X = self[0].X
for kw in kwargs:
if kw in self.attr:
par = kw
parval = np.atleast_1d(kwargs[kw])
break
else:
raise ValueError('No parameter found in data check with .attr')
degx = kwargs.pop('degx', 0)
degy = kwargs.pop('degy', 1)
if xfunc is None:
xfunc = lambda y: y
exfunc = None
if exfunc is None:
exfunc = lambda y: y
if invxfunc is None:
invxfunc = lambda y: y
if func is None:
func = lambda y: y
if invfunc is None:
invfunc = lambda y: y
if degx > 0:
# interpolate to needed X values
YY = np.array([ele.polyfit(X, deg=degx, function=xfunc, efunction=exfunc).Y for ele in self])
else:
YY = np.array([np.interp(X, ele.X, ele.Y) for ele in self])
# calc the poly coefficients for all YY
poly = np.polyfit(np.array(getattr(self, par)).flatten(), func(invxfunc(YY)), deg=degy)
# and calc the values at parval
pnn = np.array([np.poly1d(polyi)(parval) for polyi in poly.T]).T
result = dL()
for p, fy in zip(parval, pnn):
result.append(np.c_[X, invfunc(fy)].T)
setattr(result[-1], par, p)
return result
#: alternative name for polyfit
extrapolate = polyfit
def prune(self, *args, **kwargs):
"""
Reduce number of values between upper and lower limits.
Prune reduces a dataset to reduced number of data points in an interval
between lower and upper by selection or by averaging including errors.
Parameters
----------
*args,**kwargs :
arguments and keyword arguments see below
lower : float
Lower bound
upper : float
Upper bound
number : int
Number of points in [lower,upper] resulting in number intervals.
kind : 'log','lin', array, default 'lin'
Determines how new points were distributed.
- explicit list/array of new values as [1,2,3,4,5].
Interval borders were chosen in center between consecutive values.
Outside border values are symmetric to inside.
*number*, *upper*, *lower* are ignored.
The value in column specified by *col* is the average found in the interval.
The explicit values given in kind, can be set after using prune for the column given in col.
- 'log' closest values in log distribution with *number* points in [lower,upper]
- 'lin' closest values in lin distribution with *number* points in [lower,upper]
- If *number* is None all points between [lower,upper] are used.
type : {None,'mean','error','mean+error'} default 'mean'
How to determine the value for a point.
- None next original value closest to column col value.
- 'mean' mean values in interval between 2 points;
- 'mean+std' calcs mean and adds error columns as standard deviation in intervals (no weight).
Can be used if no errors are present to generate errors as std in intervals.
For single values the error is interpolated from neighbouring values.
! For less pruned data error may be bad defined if only a few points are averaged.
col : 'X','Y'....., or int, default 'X'
Column to prune along X,Y,Z or index of column.
weight : None, protectedNames as 'eY' or int
Column for weight as 1/err**2 in 'mean' calculation, weight column gets new error sqrt(1/sum_i(1/err_i**2))
- None is equal weight
- If weight not existing or contains zeros equal weights are used.
keep : list of int
List of indices to keep in any case e.g. keep=np.r_[0:10,90:101]
Returns
-------
dataArray with values pruned to number of values
Examples
--------
::
import jscatter as js
import numpy as np
x=np.r_[0:10:0.01]
data=js.dA(np.c_[x,np.sin(x)+0.2*np.random.randn(len(x)),x*0+0.2].T) # simulate data with error
p=js.grace()
p.plot(data)
p.plot(data.prune(lower=1,upper=9,number=100,type='mean+'))
Notes
-----
| Attention !!!!
| dependent on the distribution of original data a lower number of points can be the result
| e.g. think of noisy data between 4 and 5 and a lin distribution from 1 to 10 of 9 points
| as there are no data between 5 and 10 these will all result in 5 and be set to 5 to be unique
"""
out = dataList()
for element in self:
out.append(element.prune(*args, **kwargs))
return out
def transposeAttribute(self, attr):
"""
Use attribute as new X axis (like transpose .X and attribute).
It is necessary that all X have same values and length.
This can be achieved by polyfit, interpolate or prune to shape the dataList.
Parameters
----------
attr : str
Attribute to use
Returns
-------
dataList with attribute x as old X values
Examples
--------
::
i5=js.dL(js.examples.datapath+'/iqt_1hho.dat')
# polyfit and interpolate produce the same .X with control over used values
i6=i5.polyfit(X=np.r_[1:5.1],q=i5.q).transposeAttribute('q')
i7=i5.interpolate(X=i5[-1].X,q=i5.q).transposeAttribute('q')
# .prune allows to use X borders to cut X range
i5.prune(lower=1,upper=5).transposeAttribute('q')
"""
if attr not in self.attr:
raise ValueError('Attribute not found in data. Check with .attr')
result = dL()
for X, Y in zip(self.X.array.T, self.Y.array.T):
result.append(np.c_[getattr(self, attr), Y].T)
setattr(result[-1], 'x', X[0])
return result
def modelValues(self, **kwargs):
"""
Get modelValues allowing simulation with changed parameters.
Model parameters are used from dataArray attributes or last fit parameters after a fit.
Given arguments overwrite parameters and attributes to simulate modelValues
e.g. to extend X range.
Parameters
----------
**kwargs : parname=value
Overwrite parname with value in the dataList attributes or fit results
e.g. to extend the parameter range or simulate changed parameters.
debug : internal usage documented for completes
dictionary passed to model to allow calling model as model(**kwargs) for debugging
Returns
-------
dataList of modelValues with parameters as attributes.
Notes
-----
Example: extend time range ::
data=js.dL('iqt_1hho.dat')
diffusion=lambda A,D,t,wavevector: A*np.exp(-wavevector**2*D*t)
data.fit(diffusion,{'D':[2],'amplitude':[1]},{},{'t':'X'}) # do fit
# overwrite t to extend range
newmodelvalues=data.modelValues(t=numpy.r_[0:100]) #with more t
Example: 1-sigma interval for D ::
import jscatter as js
import numpy as np
data=js.dL(js.examples.datapath + '/iqt_1hho.dat')
diffusion=lambda A,D,t,q: A*np.exp(-q**2*D*t)
data.fit(diffusion,{'D':[0.1],'A':[1]},{},{'t':'X'}) # do fit
# add errors of D for confidence limits
upper=data.modelValues(D=data.D+data.D_err)
lower=data.modelValues(D=data.D-data.D_err)
data.showlastErrPlot()
data.errPlot(upper,sy=0,li=[2,1,1])
data.errPlot(lower,sy=0,li=[2,1,1])
"""
imap = {'X': 'ix', 'eX': 'iex', 'Z': 'iz', 'eZ': 'iez', 'W': 'iw', 'eW': 'iew'}
if not hasattr(self, 'model'):
raise ValueError('First define a model to calculate model values!!')
# undocumented default values, a dictionary with parnames and values
default = kwargs.pop('default', None)
debug = kwargs.pop('debug', False)
# get args and put into mappedArgs
mappedArgs = {} # all args to sent to model
mappedNames = self._mapNames # from calling fit
modelParameterNames = self._code.co_varnames[:self._code.co_argcount] # names from model definition
# map the data names to model names
for name in modelParameterNames:
# check mapped names
if name in mappedNames:
pname = mappedNames[name]
else:
pname = name
# and get the right attribute values
pval = getattr(self, pname, None)
if pval is not None:
mappedArgs[name] = pval
elif default is not None and name in default:
mappedArgs[name] = default[name]
# add the fixed parameters to the mappedArgs
mappedArgs.update(self._fixpar)
# the fit algorithm calls modelValues with kwargs through _errorfunction
# and kwargs are used for simulation
mappedArgs.update(kwargs)
# create full dataArrays in dataList as return values
values = dataList()
columnIndex = {'iy': -1, 'iey': None} # last column for Y and no errors in simulated data
# do calculation of model independently for self.X because of different length in X[i]
# singleArgs contains the kwarguments for model
singleArgs = {}
# loop over dataList elements to fill values
for i in range(len(self)):
# xxArgs will have X,Z,eX and eZ values and appended Y values
xxArgs = []
# shape the singleArgs and fill them with values
for key, item in mappedArgs.items():
if key in mappedNames and mappedNames[key] in ('X', 'Z', 'W', 'eX', 'eZ', 'eW'):
# noinspection PyBroadException
try:
singleArgs[key] = item[i][self._xslice[i]]
except:
# for new X from keywords equal for all sets
singleArgs[key] = item
xxArgs.append(singleArgs[key])
columnIndex[imap[mappedNames[key]]] = len(xxArgs) - 1
elif isinstance(item, numbers.Number): # single numbers
singleArgs[key] = item
elif len(item) == 1: # list with one element like independent parameter
singleArgs[key] = item[0] # more for convenience to avoid like [0]
elif isinstance(item, (list, np.ndarray, attributelist)): # lists
singleArgs[key] = item[i]
else:
print('strange parameter found : ', key, item, len(item), isinstance(item, list), type(item))
if isinstance(singleArgs[key], attributelist):
singleArgs[key] = singleArgs[key].array
for key in singleArgs:
# soft limits increase chi2 in _errorfunction
# here set hard limits to avoid breaking of limits
if key in self._limits:
# set minimum hard border
if self._limits[key][2] is not None and np.any(singleArgs[key] < self._limits[key][2]):
singleArgs[key] = self._limits[key][2]
# set maximum hard border
if self._limits[key][3] is not None and np.any(singleArgs[key] > self._limits[key][3]):
singleArgs[key] = self._limits[key][3]
if debug:
return singleArgs
# calc the model values
fX = self.model(**singleArgs)
if isinstance(fX, numbers.Integral) and fX < 0:
# error in model
return fX
elif hasattr(fX, '_isdataArray') and fX.ndim > 1:
values.append(fX)
else:
xxArgs.append(np.asarray(fX)) # fX should be array
values.append(np.vstack(xxArgs))
values[-1].setColumnIndex(**columnIndex)
values[-1].setattr(fX) # just in case there are attributes in return value fX
# add parameters + values to dataList (consider _mapNames)
for key, item in mappedArgs.items():
# add only the used parameters and ignore others
if key in self._fixpar.keys() or key in self._freepar.keys() or key in self._mapNames.keys():
if key in self._mapNames:
if self._mapNames[key] in ['X', 'Z', 'W', 'eX', 'eZ', 'eW']:
setattr(values, key, ['@->' + self._mapNames[key]] * len(values))
else:
setattr(values, key, item)
else:
setattr(values, key, item)
return values
def _getError(self, modelValues):
# calc error and put it together
# check if output was ok
if isinstance(modelValues, numbers.Integral) and modelValues < 0:
# there was an error in model but we return something not to break the fit
# also _lenerror is determined without model evaluation
error = np.hstack([y[xslice] * 1000 for y, xslice in zip(self.Y, self._xslice)])
evalOK = False
elif not np.all(np.isfinite(modelValues.Y.flatten)):
# we have nans or inf in the result
error = np.hstack([y[xslice] * 1000 for y, xslice in zip(self.Y, self._xslice)])
evalOK = False
elif self._nozeroerror and self._fitmethod not in ['bayes']:
err = [((val - y[xslice]) / ey[xslice]) for val, y, ey, xslice in
zip(modelValues.Y, self.Y, self.eY, self._xslice)]
error = np.hstack(err)
evalOK = True
else:
err = [(val - y[xslice]) for val, y, xslice in zip(modelValues.Y, self.Y, self._xslice)]
error = np.hstack(err)
evalOK = True
self._lenerror = len(error) # set number of data points
chi2 = np.sum(error ** 2)
return error, chi2, evalOK
def _errorfunction(self, *args, **kwargs):
"""
The fit scipy.optimize fit algorithm calls this function with the variable fit parameter in front of args
We split these and call our model and make some output....
Calculates the weighted error for least square fitting using model from fit
as (val-y)/ey if ey is given, otherwise unweighted with ey=1.
for Bayesian estimation we return log_likelyhood with a flat prior in limits
If makeErrPlot is used an intermediate stepwise output is created
as y value plot with residuals.
"""
self.numberOfModelEvaluations += 1
# _p contains the variable parameters from the fit algorithm
self._p, args = args # remaining args are not used (but it should not be empty -> (0,))
# distribute variable parameters to kwargs, check limits
i = 0
for name, par0 in self._freepar.items():
l = len(np.atleast_1d(par0))
kwargs[name] = self._p[i:i + l]
i += l
limitweight, limits, hardlimits, nconstrain = self._checklimits(self._p)
# calc modelValues, hardlimits are used there
# this might be not necessary if hardlimits are there
modelValues = self.modelValues(**kwargs)
# error determination including check of proper model evaluation
error, chi2, evalOK = self._getError(modelValues)
chi2 = chi2 / self._dof # reduced chi2
self._lastchi2 = chi2
# optional errPlot if calculation longer than errplotupdateinterval seconds ago
now = time.time()
if mp.current_process().name == 'MainProcess':
# output only if main process
if self._errplot and self._lasterrortime < now - errplotupdateinterval and evalOK:
# last calculation time
self._lasterrortime = now
self.showlastErrPlot(modelValues=modelValues, kwargs=kwargs)
# output to commandline all 0.2 s
if self._lasterrortimecommandline < now - 0.2:
self._lasterrortimecommandline = now
self._show_output(chi2, limitweight, limits, hardlimits, nconstrain, kwargs)
# save tor trace if main process
self._chi2trace.append(np.r_[self.numberOfModelEvaluations, chi2, self._p])
# now define the return dependent on the fit method
if self._fitmethod in ['lm', 'trf', 'dogbox']:
# this is for scipy.optimize.least_square, it wants residuals and calcs chi2 itself
return error * limitweight
elif self._fitmethod in ['bayes']:
# for bayesian estimation
# here error is the residual, but we return ln_likelihood + ln_prior
if self._ln_prior is None:
# default uninformative prior
# ln_prior = -inf outside limits, otherwise constant
ln_prior = -np.inf if limitweight > 1 else 0
else:
# ln_prior = -inf outside limits, otherwise given prior
ln_prior = -np.inf if limitweight > 1 else self._lnprior(**kwargs)
# sigma**2 ,
# we may later modify s2 to test error models as in emcee fit example
s2 = np.hstack([ey[xslice] for ey, xslice in zip(self.eY, self._xslice)])**2
# this is log of a Gaussian distribution, so the log of the probability
# the last log is from normalisation
ln_likelyhood = -0.5 * np.sum(error**2 / s2 + np.log(2*np.pi*s2))
# combine and return it
return (ln_prior + ln_likelyhood) if np.isfinite(ln_prior) else -np.inf
else:
# this returns chi2 for all algorithm in scipy.optimize.minimize and differential_evolution
return chi2 * limitweight
def _lnprior(self, **kwargs):
# only used in fit method 'bayes'
# update call arguments with fixpar
kwargs.update(self._fixpar)
# the prior described only prior knowledge about parameters,
# so we skip protectedNames as these describe data like X or Y
# and use all required parameters for ln_prior
codevars = self._ln_prior.__code__.co_varnames[:self._ln_prior.__code__.co_argcount]
kw = {}
for k in codevars:
if (k in kwargs) and (k not in protectedNames):
kw.update([(k, kwargs[k])])
# now try prior evaluation
lnprior = self._ln_prior(**kw)
return lnprior
def _checklimits(self, parameters):
"""
Checks the parameters if limits are reached and increases limitweight.
Returns
-------
limitweight,limits,hardlimits,nconstrain
"""
# add _p to corresponding kwargs[name] values to reproduce change in fit algorithm
i = 0
limitweight = 1 # factor for chi2
limits = [] # names of soft limited parameters
hardlimits = [] # names of hard limited parameters
nconstrain = 0 # names of constrain
kwargs = {}
for name, par0 in self._freepar.items():
l = len(np.atleast_1d(par0))
par = parameters[i:i + l]
kwargs[name] = par
# here determine upper and lower bound
if name in self._limits:
# soft limits just increase chi2 by a factor limitweight >1
if self._limits[name][0] is not None and np.any(par < self._limits[name][0]): # set minimum border
# increase with distance to border and number of parameters above border
wff = sum(abs(par - self._limits[name][0]) * (par < self._limits[name][0]))
limitweight += 1 + wff * 10
limits.append(name)
if self._limits[name][1] is not None and np.any(par > self._limits[name][1]): # set maximum border
wff = sum(abs(par - self._limits[name][1]) * (par > self._limits[name][1]))
limitweight += 1 + wff * 10
limits.append(name)
# hard limits are set in modelValues here only tracking for output and increase weight
if self._limits[name][2] is not None and np.any(par < self._limits[name][2]): # set minimum hard border
wff = sum(abs(par - self._limits[name][2]) * (par < self._limits[name][2]))
limitweight += 10 + wff * 10
hardlimits.append(name)
if self._limits[name][3] is not None and np.any(par > self._limits[name][3]): # set maximum hard border
wff = sum(abs(par - self._limits[name][3]) * (par > self._limits[name][3]))
limitweight += 10 + wff * 10
hardlimits.append(name)
i += l
if self.hasConstrain:
# combines actual fitpar and the fixpar
kwargs = dict(kwargs, **self._fixpar)
largs = {d: k for d, k in kwargs.items() if np.size(k) > 1} # list kwargs
fargs = {d: k for d, k in kwargs.items() if np.size(k) == 1} # float kwargs
constrain = []
for cfunc in self._constrains:
code = cfunc.__code__
cf_names = code.co_varnames[:code.co_argcount]
if largs:
for i in range(len(self)):
kargs = {name: largs[name][i] for name in cf_names if name in largs}
kargs = dict(kargs, **{name: fargs[name] for name in cf_names if name in fargs})
constrain.append(cfunc.__call__(**kargs))
else:
kargs = {name: fargs[name] for name in cf_names if name in fargs}
constrain.append(cfunc.__call__(**kargs))
nconstrain = sum(np.array(constrain) is False) # count evaluations which are False
limitweight += 10 * nconstrain
return limitweight, limits, hardlimits, nconstrain
def _show_output(self, chi2, limitweight=1, limits=[], hardlimits=[], nconstrain=0, kwargs={}):
if self._output is None or self._output is False:
# suppress output
return
print('chi^2 = %.5g * %.1g (limit weight) after %i evaluations'
% (chi2, limitweight, self.numberOfModelEvaluations))
outlist = ''.join(['%-8s= %s %s %s\n' % (
(item, '', value, '') if item not in limits + hardlimits else
((item, CSIr, value, ' !limited' + CSIe) if item not in hardlimits else
(item, CSIy, value, ' !hard limited' + CSIe)))
for item, value in sorted(kwargs.items())])
outlist += '-----fixed-----\n'
for name, values in sorted(self._fixpar.items()):
# noinspection PyBroadException
try:
outlist += '%-8s=[' % name + ''.join([' %.4G' % val for val in values]) + ']\n'
except:
outlist += '%-8s=[%.4G]\n' % (name, values)
if nconstrain > 0:
outlist += 'Constrains violated : %d \n' % nconstrain
print(outlist, )
return
def getChi2Trace(self):
"""
Get the trace of function evaluations after a fit.
Maybee useful in 'differential evolution' to get last ensembles or to see fit convergence.
Or to restart fits that break because of other conditions.
Works only if fit method does not use multiprocessing like e.g. 'bayes'.
Multiprocessing in the model is allowed.
Returns
-------
array with [index, chi2, par1, par2,...]
Examples
--------
::
import jscatter as js
import numpy as np
data=js.dA(js.examples.datapath+'/exampledata0.dat')
def parabola(q,a,b,c):
return (q-a)**2+b*q+c
par = {'a':2,'b':4}
data.fit( model=parabola ,freepar=par, fixpar={'c':-20}, mapNames={'q':'X'})
trace = data.getChi2Trace()
# to restart a fit or simulate the model values with the last parameters
lastpar = {k.strip():v for k,v in zip(trace.columnname.split(';')[2:], trace.array[2:,-1])}
data.fit( model=parabola ,freepar=lastpar, fixpar={'c':-20}, mapNames={'q':'X'})
"""
chi2trace = dA(np.array(self._chi2trace).T)
chi2trace.setColumnIndex(iey=None)
chi2trace.columnname = 'i; chi2;' + ''.join([f' {k};' for k in self._freepar.keys()])
lastfreepar = dA()
for k, v in zip(self._freepar.keys(), self._chi2trace[-1][2:]):
setattr(lastfreepar, k.strip(), v)
chi2trace.lastfreepar = lastfreepar
return chi2trace
def setConstrain(self, *args):
"""
Set inequality constrains for constrained minimization in fit.
Inequality constrains are accounted by an exterior penalty function increasing chi2.
Equality constrains should be incorporated in the model function
to reduce the number of parameters.
Parameters
----------
args : function or lambda function
Function that defines constrains by returning boolean with free and fixed parameters as input.
The constrain function should return True in the accepted region and return False otherwise.
Without function all constrains are removed.
Notes
-----
Warning:
The fit will find a best solution with violated constrains
if the constrains forbid to find a good solution.
A 3 component model with fractional contributions n1,n2,n3
Constrains are:
- n1+n2+n3=1
- 0=<ni<=1 for i=1, 2, 3
Use n3=1-n1-n2 to reduce number of parameters in model function.
Set constrain::
data.setconstrain(lambda n1,n2:(0<=n1<=1) & (0<=n2<=1) & (0<=1-n1-n2<=1))
"""
if not args:
self._constrains = []
else:
for func in args:
if isinstance(func, types.FunctionType):
self._constrains.append(func)
else:
print('This is not a function')
@property
def hasConstrain(self):
"""
Return list with defined constrained source code.
"""
if self._constrains:
return [inspect.getsource(fconst) for fconst in self._constrains]
else:
return None
def setLimit(self, **kwargs):
"""
Set upper and lower limits for parameters in least square fit.
Use as ``.setlimit(parname=(lowlimit, uplimit,lowhardlimit, uphardlimit))``
Parameters
----------
parname : [value x 4] , list of 4 x (float/None), default None
- lowlimit, uplimit : float, default None
soft limits: chi2 increased with distance from limit, non-float resets limit
- lowhardlimit, uphardlimit: hardlimit float, None
values are set to border , chi2 is increased strongly
Notes
-----
Penalty methods are a certain class of algorithms for solving constrained optimization problems.
Here the penalty function increases chi2 by a factor chi*f_constrain.
- no limit overrun : 1
- softlimits : + 1+abs(val-limit)*10 per limit
- hardlimits : +10+abs(val-limit)*10 per limit
Examples
--------
::
setlimit(D=(1,100),A=(0.2,0.8,0.0001)) # to set low=1 and up=100
# A with a hard limit to avoid zero
setlimit(D=(None,100)) # to reset lower and set upper=100
setlimit(D=(1,'thisisnotfloat','',)) # to set low=1 and reset up
"""
if 'reset' in kwargs or len(kwargs) == 0:
self._limits = {}
return
for key in kwargs:
limits = [None, None, None, None]
try:
limits[0] = float(kwargs[key][0])
except (IndexError, TypeError):
pass
try:
limits[1] = float(kwargs[key][1])
except (IndexError, TypeError):
pass
try:
limits[2] = float(kwargs[key][2])
except (IndexError, TypeError):
pass
try:
limits[3] = float(kwargs[key][3])
except (IndexError, TypeError):
pass
self._limits[key] = limits
setlimit = setLimit
@property
def hasLimit(self):
"""
Return existing limits
without limits returns None
"""
if isinstance(self._limits, dict) and self._limits != {}:
return self._limits
return None
has_limit = hasLimit
def fit(self, model, freepar={}, fixpar={}, mapNames={}, method='lm', xslice=slice(None), condition=None,
output=True, **kw):
r"""
Least square fit of model that minimizes :math:`\chi^2` (uses scipy.optimize methods) or Bayesian analysis.
- A fit of scalar .Y values dependent on coordinates (X,Z,W) and attributes (multidimensional fitting).
- Data attributes are used automatically in model if they have the same name as a parameter.
- Resulting errors are 1-sigma errors as estimated from the covariance matrix diagonal for
:math:`\chi^2` minimization or std deviation of the distribution for Bayesian inference.
Errors be accessed for fit parameter D as D_err (see **Fit result attributes** below).
- Results can be simulated with changed parameters in ``.modelValues``, ``.showlastErrPlot`` or ``.simulate``.
Parameters
----------
model : function or lambda
Model function, should accept arrays as input (use numpy ufunctions).
- Return value should be dataArray (.Y is used) or only Y values.
- Failed model evaluation should return single negative integer.
Example ( see :ref:`How to build simple models` ): ::
diffusion=lambda A,D,t,wavevector: A*np.exp(-wavevector**2*D*t)
freepar : dictionary
Fit parameter names with start values.
- ``{'D':2.56,..}`` float, one common value for all
- ``{'D':[1,2.3,4.5,...],..}`` list of float, individual parameters for independent fit.
- ``[..]`` is extended with missing values equal to last given value. [2,1] -> [2,1,1,1,1,1]
- It is sufficient to add [] around a float to switch between common value and independent fit values.
fixpar : dictionary
Fixed parameters (see freepar for syntax). Overwrites data attributes with same name.
mapNames : dictionary
Map parameter names from model to attribute names in data e.g. {'t':'X','wavevector':'q',}
method : 'lm', 'trf', 'dogbox', ‘Nelder-Mead’, 'bayes', 'differential_evolution', ‘BFGS’,
or scipy.optimize.minimize methods
Type of solver for minimization
See later **fit methods** for usage hints and a speed comparison. Fastest are 'lm', "trf", "dogbox".
- **'lm'** (default) and what you typically expect for fitting and should be first choice without bounds.
It is a wrapper around MINPACK’s lmdif and lmder algorithms which are a modification
of the **Levenberg-Marquardt algorithm**. *Returns errors*. With limits 'trf' is used.
- **'trf'** (default with limits) trust-region reflective algorithm, similar to 'lm' but with bounds.
*Returns errors*.
- **'dogbox'** a trust-region algorithm, but considers rectangular trust regions. *Returns errors*.
- **'Nelder-Mead'** allows optimization of integer variables. Additionally it is sometime more robust if
gradient methods (like above) break early or stick in a local minimum. NO errors.
- **'BFGS'** quasi-Newton method of Broyden, Fletcher, Goldfarb, and Shanno.
Uses the first derivatives only. *Returns errors*.
- **'bayes'** uses Bayesian inference for modeling and the MCMC algorithms for sampling
(we use `emcee <https://emcee.readthedocs.io>`_).
Requires a larger amount of function evaluations but *returns errors* from Bayesian statistical analysis.
Check additional description and parameters below.
Needs limits, which are set as for differential_evolution.
See :py:func:`getBayesSampler` with example.
The model should not use multiprocessing as 'bayes' uses this already.
The prior can be changed using the additional parameter ``ln_prior``.
- **'differential_evolution'** is a stochastic population based method that is useful for global
optimization problems. Needs bounds for all parameters as used from .setlimit.
- If no bounds are set bounds are generated automatic around start value x0 as [x0/10**0.5,x0*10**0.5].
- The optional *workers* argument allows parallel processing in a pool of workers.
``workers=-1`` uses all cpus, otherwise the number of cpus (default all).
NO errors.
- All methods use bounds set by *.setlimits* to allow bounds as described there.
For 'trf', 'dogbox' the bounds implemented in the method are used taken from setlimit soft bounds.
- For additonal options passed to scipy.optimize least_square ('lm', 'trf', 'dogbox') or minimize (others)
see `scipy.optimize <https://docs.scipy.org/doc/scipy/reference/optimize.html>`_.
For some methods the Jacobian is required.
xslice : slice object
Select datapoints to include by slicing ::
xslice=slice(2,-3,2) To skip first 2,last 3 and take each second.
condition : function or lambda
A function to determine which datapoints to include.
- The function should evaluate to boolean with dataArray as input
and combines with xslice used on full set (first xslice then the condition is used)
- local operation on numpy arrays as "&"(and), "|"(or), "^"(xor) ::
sel = lambda a:(a.X>1) & (a.Y<1)
sel = lambda a:(a.X>1) & (a.X<100)
sel = lambda a: a.X>a.q * a.X
Rg = 4 # nm
gunier = lambda a: a.X**2*Rg**2/3 < 1
Use as ``condition = sel`` or ``condition=lambda a:(a.X>1) & (a.Y<1)``
output : 'last','best', False, default True
By default write some text messages (fit progress).
- 'last' return lastfit and text messages
- 'best' return best (parameters,errors) and text messages
- False : No printed output.
debug : 1,2,3,int
Debug modus returns:
- 1 simulation mode: parameters sent to model, errPlot and modelValues without fitting.
- 2 Free and fixed parameters but not mappedNames.
- 3 Fitparameters in modelValues as dict to call model as model(**kwargs) with mappedNames.
- 4 Prints parameters sent to model and returns the output of model without fitting.
- >4 -> 1
kw : additional keyword arguments
Additional kw forwarded to fit method as given in
`scipy.optimize <https://docs.scipy.org/doc/scipy/reference/optimize.html#module-scipy.optimize>`_
for least_square or minimize or `emcee <https://emcee.readthedocs.io>`_.
See simple `additional kwargs`_
Returns
-------
By default no return value.
- Final results with errors are in ``.lastfit``
- Fitparameters are additional in dataList object as ``.parname`` and corresponding errors as ``.parname_err``.
- If the fit fails an exception is raised and last parameters are printed.
!!_These_are_NOT_a_valid_fit_result_!!.
Notes
-----
* For :math:`\mathbf{\chi^2 minimization}` the unbiased estimate
.. math:: \chi^2 = \frac{1}{n-p} \sum_i \frac{[X_i-f(X_i,a)]^2}{\sigma_i^2}
with number of datapoints :math:`n`, number of parameters :math:`p` and function :math:`f(X_i,a)`
dependent on parameters :math:`a_i` is minimized.
Differences are weighted by measurement errors :math:`\sigma_i^2`.
Methods from `scipy.optimize <https://docs.scipy.org/doc/>`_ are used.
For **Bayesian analysis ('bayes')** the log probability :math:`\ln\,p(y\,|\,x,\sigma,a) + log(p(a_i))`
is maximized with the log likelihood :math:`ln (p(y\,|\,x,\sigma,a))`
.. math:: \ln\,p(y\,|\,x,\sigma,a) = -\frac{1}{2} \sum_i \left[\frac{(X_i-f(X_i,a))^2}{\sigma_i^2}
+ \ln \left ( 2\pi\,\sigma_i^2 \right )\right]`
and the prior :math:`p(a_i)`. By default an uniform (so-called "uninformative") prior is used
.. math:: log(p(a_i)) = \left\{ \begin{array}{ll}
0 & \mbox{if $a_i$ in limits};\\
-inf & \mbox{otherwise}.\end{array} \right.
which can be changed using the parameter ``ln_prior`` to a more informative prior.
See `emcee <https://emcee.readthedocs.io>`_ for details how this works and further analysis
or options.
* For :math:`\mathbf{\chi^2}` **minimization** the resulting parameter errors are 1-sigma errors as
determined from the covariance matrix (see [1]_).
This holds under some (and some more) assumptions :
- We have a well-behaved likelihood function (same as in 'bayes') which is asymptotically Gaussian
near its maximum (equal to minimum in :math:`\chi^2`).
- The error estimate is reasonable (1-sigma measurement errors in data).
- The model is the correct model.
- The model parameters are linear close to the :math:`\chi^2` minimum.
These reasons might also limit the best :math:`\chi^2` values reached in a fit
(e.g. it reaches only 2 instead of 1) if a simplified model is used.
Practically, the resulting parameter error is (roughly) independent of the absolute value of errors
in the data as long as the relative contributions are well represented.
Thus scaling of the errors by a factor leads to same 1-sigma error,
respectively the 1-sigma errors are independent of the absolute error scale.
Please try this by modifying the examples given, it can be proved analytically.
* For **Bayesian estimations ('bayes')** errors are determined from the likelihood and represent the
standard deviation of the mean in the real likelyhood. (see [1]_ 3.4 ).
By default an uninfomative prior within set limits is used.
This can be changed passing a log_prior function to the parameter ``ln_prior`` which contains
any prior information about the parameters (NOT the data).
This might be e.g. that some parameters are itself distributed like a Gaussian around a mean or
their differences.
The log_prior function gets as parameters all ``freepar`` and ``fixpar`` arrays
(but not the .X values from data).
If likelyhood and prior are correctly weighted ( for each :math:`\mathbf{\chi^2}` is 1 ) both contribute equally
to the probability. An example with a prior is given in :ref:`Bayesian inference for fitting`.
A sufficient requirement for the here used methods from `emcee <https://emcee.readthedocs.io>`_ is that
the sample number :math:`N > tolerance * \tau_f` with the autocorrelation time :math:`\tau_f` and
the tolerance =50, which means having just enough samples to get reasonable statistics.
The fit method tries in steps of *bayesnsteps* to determine :math:`\tau_f` and
proceeds until the requirement is satisfied. *tolerance* is by default =50 but can be reduced for testing
before a production run. Also, *nwalkers* (default 2*number of parameters) can be changed.
:math:`2\tau_f` is discarded from the chain for analysis to remove bias.
See :py:func:`getBayesSampler` to get the emcee sampler and for example usage.
The **disadvantage of 'bayes'** is that simple problems need a quite large number of function evaluations.
E.g. the below comparison needs around 10000 evaluations for a single dataArray compard to
around 30 for method 'lm' for the full dataList.
More complex models that need longer than 1s to evaluate need hours to days to finish.
For single datasets with fast evaluated functions 'bayes' works reasonable well.
For these reasons 'bayes' works by default with a pool of workers (multiprocessing) which wins largely
on a multicore machine with more than 16 cpus and is not well for a notebook.
For good data with small errors 'bayes' should not be used as there is no advantage compared to e.g. 'lm'
('lm' results in good error estimate in this case).
If your data and model have trouble to find a :math:`\chi^2` minimum or find often local minima give 'bayes'
a try. After testing (small tolerance) the sampling might run for longer times if needed.
The samples can be accessed after the fitting to examine the results. See :py:func:`getBayesSampler`.
My personal impression is that if you have a chance to improve your measurement do this and use 'lm' or 'trp'
before spending the time on long 'bayes' estimation.
For unique data (your spectral analysis of Halley's comet) one can improve analysis using 'bayes'.
* If data errors exist (:math:`\sigma_i` = *.eY*) and are not zero, the error weighted :math:`\chi^2`
is minimized (or respective weighted likelihood is maximized).
Without error (or with single errors equal zero) unweighted values respective equal weights are used.
* Using unweighted error means basically *equal weight* as :math:`\sigma_i^2 =1` in :math:`\chi^2` above.
This might lead to biased results.
* If no errors are directly available it is useful (or a better error estimate than equal weights)
to introduce a weight that represents the statistical nature of the measurement
(at least the dominating term in error propagation).
* equal errors :math:`\sigma_i \propto const`
* equal relative error :math:`\sigma_i \propto .Y`
* statistical :math:`\sigma_i \propto .Y^{0.5}` e.g. Poisson statistics on neutron/Xray detector.
* with bgr :math:`\sigma_i \propto .Y+b`
* any other :math:`\sigma_i \propto f(.X,.Y,...)`
To use one or the other a column needs to be added with the respective values
and use .setColumnIndex(iey=...) to mark it as error column `.eY` .
Set values as e.g. ``data.eY=0.01*data.Y`` for equal relative errors.
* The concept of dataLists is to use data attributes as fixed parameters for the fit (multidimensional fit).
This is realized by using data attributes with same name as fixed parameters
if not given explicitly in freepar or fixpar.
* Fit parameters can be set equal for all elements ``'par':1`` or independent ``'par':[1]``
just by writing the start value as a single float or as a list of float.
The same for fixed parameters.
* Changing between free and fixed parameters is easily done by moving ``'par':[1]`` between freepar and fixpar.
* Limits for parameters can be set prior to the fit as ``.setlimit(D=[1,4,0,10])``.
The first two numbers (min,max) are softlimits (increase :math:`\chi^2`) and
second are hardlimits to avoid extreme values.
(hard set to these values if outside interval and increasing :math:`\chi^2`).
* The change of parameters can be simulated by ``.modelValues(D=3)``
which overrides attributes and fit parameters.
* ``.makeErrPlot()`` creates an errorplot with residuals prior to the fit for intermediate output.
* The last errPlot can be recreated after the fit with ``.showlastErrPlot()``.
* The simulated data can be shown in errPlot with ``.showlastErrPlot(D=3)``.
* Each dataArray in a dataList can be fit individually (same model function) like this ::
# see Examples for dataList creation
data[3].fit(model,freepar,fixpar,.....)
# or
for dat in data:
dat.fit(model,freepar,fixpar,.....)
Most important _`additional kwargs` for 'least_square' methods 'lm', 'trf', 'dogbox' ::
arguments passed to least_squares (see scipy.optimize.least_square)
ftol default 1.e-8 Relative error desired in the sum of squares (also tol accepted)
xtol default 1.e-8 Relative error desired in the approximate best parameters.
gtol default 1.e-8 Tolerance for termination by the norm of the gradient.
max_nfev default ~100*N(*N 'lm') Maximum model evaluations (also maxiter accepted)
diff_step default None, relative step size = x*diff_step for finite differences
approx. of the Jacobian.
Most important additional kwargs for 'minimize' methods 'Nelder-Mead', 'BFGS',... ::
arguments passed to minimize
tol tolerance for termination. Depends on algorithm. (also max_nfev excepted)
maxiter maximum model evaluations (also max_nfev accepted)
**Fit result attributes** ::
# exda are fitted example data
exda.D freepar 'D' ; same for fixpar but no error.
use exda.lastfit.attr to see all attributes of model
exda.D_err 1-sigma error of freepar 'D'
# full result in lastfit
exda.lastfit.X X values in fit model
exda.lastfit.Y Y values in fit model
exda.lastfit[i].D free parameter D result in best fit
exda.lastfit[i].D_err error of free parameter D as 1-sigma error from diagonal in covariance matrix.
exda.lastfit.chi2 chi2 = sum(((.Y-model(.X,best))/.eY)**2)/dof; should be around 1 with proper weight.
exda.lastfit.cov covariance matrix = hessian**-1 * chi2
exda.lastfit.dof degrees of freedom = len(y)-len(best)
exda.lastfit.func_name name of used model
exda.lastfit.func_code where to find code of used model
Examples
--------
**How to make a model**:
The model function gets ``.X`` (``.Z, .W, .eY, ....``) as ndarray and parameters
(from attributes and freepar and fixpar) as scalar input.
It should return a ndarray as output (only ``.Y`` values) or dataArray (``.Y`` is used automatically).
Therefore, it is advised to use numpy ufunctions in the model. Instead of ``math.sin`` use ``numpy.sin``,
which is achieved by ``import numpy as np`` and use ``np.sin``
see https://numpy.org/doc/stable/reference/ufuncs.html#available-ufuncs
See :ref:`How to build simple models` and :ref:`How to build a more complex model`
A bunch of examples can be found in *formel.py*, *formfactor.py*, *stucturefactor.py*.
**Basic examples with synthetic data.**
Usually data are loaded from a file.
For the following also see :ref:`1D fits with attributes` and :ref:`2D fitting` .
- An error plot with residuals can be created for intermediate output.
The model is here a lambda function ::
import jscatter as js
import numpy as np
data=js.dL(js.examples.datapath+'/iqt_1hho.dat')
diffusion=lambda t,wavevector,A,D,b:A*np.exp(-wavevector**2*D*t)+b
data.setlimit(D=(0,2)) # set a limit for diffusion values
data.makeErrPlot() # create errorplot which is updated
data.fit(model=diffusion ,
freepar={'D':0.1, # one value for all (as a first try)
'A':[1,2,3]}, # extended to [1,2,3,3,3,3,...3] independent parameters
fixpar={'b':0.} , # fixed parameters here, [1,2,3] possible
mapNames= {'t':'X', # maps time t of the model as .X column for the fit.
'wavevector':'q'}, # and map model parameter 'wavevector' to data attribute .q
condition=lambda a:(a.Y>0.1) ) # set a condition
- Fit sine to simulated data. The model is inline lambda function. ::
import jscatter as js
import numpy as np
x=np.r_[0:10:0.1]
data=js.dA(np.c_[x,np.sin(x)+0.2*np.random.randn(len(x)),x*0+0.2].T) # simulate data with error
data.fit(lambda x,A,a,B:A*np.sin(a*x)+B,{'A':1.2,'a':1.2,'B':0},{},{'x':'X'}) # fit data
data.showlastErrPlot() # show fit
print( data.A,data.A_err) # access A and error
- Fit sine to simulated data using an attribute in data with same name ::
x=np.r_[0:10:0.1]
data=js.dA(np.c_[x,1.234*np.sin(x)+0.1*np.random.randn(len(x)),x*0+0.1].T) # create data
data.A=1.234 # add attribute
data.makeErrPlot() # makes errorPlot prior to fit
data.fit(lambda x,A,a,B:A*np.sin(a*x)+B,{'a':1.2,'B':0},{},{'x':'X'}) # fit using .A
- Fit sine to simulated data using an attribute in data with different name and fixed B ::
x=np.r_[0:10:0.1]
data=js.dA(np.c_[x,1.234*np.sin(x)+0.1*np.random.randn(len(x)),x*0+0.1].T) # create data
data.dd=1.234 # add attribute
data.fit(lambda x,A,a,B:A*np.sin(a*x)+B,{'a':1.2,},{'B':0},{'x':'X','A':'dd'}) # fit data
data.showlastErrPlot() # show fit
- Fit sine to simulated dataList using an attribute in data with different name and fixed B from data.
first one common parameter then as parameter list in []. ::
import jscatter as js
import numpy as np
x=np.r_[0:10:0.1]
data=js.dL()
ef=0.1 # increase this to increase error bars of final result
for ff in [0.001,0.4,0.8,1.2,1.6]: # create data
data.append( js.dA(np.c_[x,(1.234+ff)*np.sin(x+ff)+ef*ff*np.random.randn(len(x)),x*0+ef*ff].T) )
data[-1].B=0.2*ff/2 # add attributes
#
# fit with a single parameter for all data, obviously wrong result
data.fit(lambda x,A,a,B,p:A*np.sin(a*x+p)+B,{'a':1.2,'p':0,'A':1.2},{},{'x':'X'})
data.showlastErrPlot() # show fit
#
# now allowing multiple p,A,B as indicated by the list starting value
data.fit(lambda x,A,a,B,p:A*np.sin(a*x+p)+B,{'a':1.2,'p':[0],'B':[0,0.1],'A':[1]},{},{'x':'X'})
#data.savelastErrPlot(js.examples.imagepath+'/4sinErrPlot.jpg')
# plot p against A , just as demonstration
p=js.grace()
p.plot(data.A,data.p,data.p_err)
.. image:: ../../examples/images/4sinErrPlot.jpg
:align: center
:width: 50 %
:alt: 4sinErrPlot
- **2D/3D/xD fit** for scalar Y
For 2D fit we calc Y values from X,Z coordinates, for 3D fits we use X,Z,W coordinates.
For 2D plotting of the result we need data in X,Z,Y column format.
This can be combined with attribute dependence to result in higher dimensional fits.
::
import jscatter as js
import numpy as np
#
# create 2D data with X,Z axes and Y values as Y=f(X,Z)
x,z=np.mgrid[-5:5:0.25,-5:5:0.25]
xyz=js.dA(np.c_[x.flatten(),
z.flatten(),
0.3*np.sin(x*z/np.pi).flatten()+0.01*np.random.randn(len(x.flatten())),
0.01*np.ones_like(x).flatten() ].T)
# set columns where to find X,Y,Z )
xyz.setColumnIndex(ix=0,iz=1,iy=2,iey=3)
#
def mymodel(x,z,a,b):
return a*np.sin(b*x*z)
xyz.fit(mymodel,{'a':1,'b':1/3.},{},{'x':'X','z':'Z'})
# inspect the result
fig = js.mpl.showlastErrPlot2D(xyz)
#fig.savefig(js.examples.imagepath+'/2dfit.jpg')
.. image:: ../../examples/images/2dfit.jpg
:align: center
:width: 70 %
:alt: 2dfit
- Fit for **vector valued results**
Vectors (multidimensional results or vector Y values) like a vector field are fitted by minimizing
a specific norm (L2 or other) of the difference between measured vectors and model vectors.
Doing this the fit is reduced to a scalar fit as above.
- **Comparison of fit methods** ::
import numpy as np
import jscatter as js
diffusion=lambda A,D,t,elastic,wavevector=0:A*np.exp(-wavevector**2*D*t)+elastic
i5=js.dL(js.examples.datapath+'/iqt_1hho.dat')
i5.makeErrPlot(title='diffusion model residual plot')
# default
i5.fit(model=diffusion,freepar={'D':0.2,'A':1}, fixpar={'elastic':0.0},
mapNames= {'t':'X','wavevector':'q'},
condition=lambda a:a.X>0.01, method='lm')
# 22 evaluations; error YES -> 'lm'
# with D=[0.2] => 130 evaluations and chi2 = 0.992
i5.fit(model=diffusion,freepar={'D':0.2,'A':1}, fixpar={'elastic':0.0},
mapNames= {'t':'X','wavevector':'q'},
condition=lambda a:a.X>0.01, method='trf')
# 22 evaluations; error YES
# with D=[0.2] => 145 evaluations and chi2 = 0.992
i5.fit(model=diffusion,freepar={'D':0.2,'A':1}, fixpar={'elastic':0.0},
mapNames= {'t':'X','wavevector':'q'},
condition=lambda a:a.X>0.01, method='dogbox')
# 22 evaluations; error YES
# with D=[0.2] => 145 evaluations and chi2 = 0.992
i5.fit(model=diffusion,freepar={'D':0.2,'A':1}, fixpar={'elastic':0.0},
mapNames= {'t':'X','wavevector':'q'},
condition=lambda a:a.X>0.01 ,method='Nelder-Mead' )
# 72 evaluations, error NO
i5.fit(model=diffusion,freepar={'D':0.2,'A':1}, fixpar={'elastic':0.0},
mapNames= {'t':'X','wavevector':'q'},
condition=lambda a:a.X>0.01, method='differential_evolution', workers=-1)
# >400 evaluations, error NO ; needs >20000 evaluations using D=[0.2]
# profits strongly from worker>1 (-1 = all) to use multiple processes (not on Windows)
# use only with low number of parameters and polish result with methods yielding errors.
i5.fit(model=diffusion,freepar={'D':0.2,'A':1}, fixpar={'elastic':0.0},
mapNames= {'t':'X','wavevector':'q'},
condition=lambda a:a.X>0.01, method='bayes', tolerance=50, bayesnsteps=1000)
# >10000 evaluations; error YES
# tolerance should be >= 50 not smaller
# The full dataset takes some time (for testing -> tolerance=20 or i6=i5[::3])
# use only with low number of parameters and polish result when you really need it
i5.fit(model=diffusion,freepar={'D':0.2,'A':1}, fixpar={'elastic':0.0},
mapNames= {'t':'X','wavevector':'q'},
condition=lambda a:a.X>0.01 ,method='Powell' )
# 121 evaluations; error NO
i5.fit(model=diffusion,freepar={'D':0.2,'A':1}, fixpar={'elastic':0.0},
mapNames= {'t':'X','wavevector':'q'},
condition=lambda a:a.X>0.01 ,method='SLSQP' )
# 37 evaluations, error NO
i5.fit(model=diffusion,freepar={'D':0.2,'A':1}, fixpar={'elastic':0.0},
mapNames= {'t':'X','wavevector':'q'},
condition=lambda a:a.X>0.01 ,method='BFGS' )
# 52 evaluations, error YES
# with D=[0.2] => 931 evaluations and chi2 = 0.992
i5.fit(model=diffusion,freepar={'D':0.2,'A':1}, fixpar={'elastic':0.0},
mapNames= {'t':'X','wavevector':'q'},
condition=lambda a:a.X>0.01 ,method='COBYLA' )
# 269 evaluations, error NO
References
----------
About error estimate from covariance Matrix M with the Fischer matrix :math:`F` (like gradien methods)
.. math:: cM_{i,j}=(\frac{\partial^2 log(F) }{ \partial x_i\partial x_j})^{-1} .
.. [1] https://arxiv.org/pdf/1009.2755.pdf
"""
debug = kw.pop('debug', False)
# store all we need for fit with attributes
self.model = model
self.numberOfModelEvaluations = 0
self._chi2trace = []
try:
# is wrapped in decorator as e.g.sas.smear but we want to look here at the wrapped code
self._code = self.model.__wrapped__.__code__
except AttributeError:
# seems to be normal function
self._code = self.model.__code__
# test for protected names
if len(set(protectedNames) & set(self._code.co_varnames[:self._code.co_argcount])) != 0:
raise NameError(' model should not have a parameter name of: ' + ' '.join(protectedNames))
# save parameters internally
self._freepar = collections.OrderedDict(sorted(freepar.items(), key=lambda t: t[0]))
self._mapNames = collections.OrderedDict(sorted(mapNames.items(), key=lambda t: t[0]))
self._fixpar = collections.OrderedDict(sorted(fixpar.items(), key=lambda t: t[0]))
self._lasterrortime = 0 # to limit frequency for optional output in _errorfunction,0 so first is plotted
self._lasterrortimecommandline = 0 # to limit frequency for output on commandline
self._output = output # disable output
self._ln_prior = kw.pop('ln_prior', None) # ln_prior function for bayes
# we need a list of slices to select values to be included for fit
if isinstance(xslice, slice):
xslice = [xslice]
xslice.extend([xslice[-1]] * (len(self) - len(xslice))) # extend to len(self) with last element
self._xslice = xslice
# overwrite _xslice if a condition was given
if condition is not None:
for i in range(len(self)):
# intersection of condition and full slice over data to use both
cond = condition(self[i])
if isinstance(cond, bool):
cond = np.full(len(self[i].X), cond, dtype=bool)
self._xslice[i] = np.intersect1d(np.where(cond)[0], np.arange(len(self[i].X))[self._xslice[i]])
# ensure length of parameter list =length self
for key in self._freepar:
if isinstance(self._freepar[key], list):
self._freepar[key].extend([self._freepar[key][-1]] * (len(self) - len(self._freepar[key])))
for key in self._fixpar:
if isinstance(self._fixpar[key], list):
self._fixpar[key].extend([self._fixpar[key][-1]] * (len(self) - len(self._fixpar[key])))
# only with nonzero errors we cal in _errorfunction weighted chi**2
if any([ey is None for ey in self.eY]):
self._nozeroerror = False
else:
# test if Zero (is False) in eY
self._nozeroerror = np.all([np.all(ey[xslice]) for ey, xslice in zip(self.eY, self._xslice)])
if not self._nozeroerror:
warnings.warn('Errors equal zero detected. Using non-weighted chi**2', UserWarning)
# list of free parameters for fit routine as 1d array as start parameters
freeParValues = np.r_[[sval for k, val in self._freepar.items() for sval in np.atleast_1d(val)]]
self._len_p = len(freeParValues) # set number of free parameters
# in _getError(-1) number of datapoints is set
_ = self._getError(-1)
self._dof = self._lenerror - self._len_p
if debug:
if debug == 2:
return dict(self._freepar, **self._fixpar)
elif debug == 3:
return self.modelValues(**dict(self._freepar, **self._fixpar))
elif debug == 4:
# show parameter sent to modeValues and returns output of modelValues
print('sent to model from fit in debug mode 3:')
outlist = ''.join(['%-8s= %s %s %s\n' % (item, '', value, '')
for item, value in sorted(dict(self._freepar, **self._fixpar).items())])
print(outlist)
return self.modelValues(**dict(self._freepar, **self._fixpar))
else:
print('sent to model in simulation mode:')
outlist = ''.join(['%-8s= %s %s %s\n' % (item, '', value, '')
for item, value in sorted(dict(self._freepar, **self._fixpar).items())])
print(outlist)
_mV = self.modelValues(**dict(self._freepar, **self._fixpar, **kw))
self.showlastErrPlot(modelValues=_mV)
return _mV
else:
# remove lastfit if existing
try:
# delete a previous fit result
del self.lastfit
except AttributeError:
pass
if self._dof < 1:
if self._output:
print('Degrees of freedom < 1 ; We need more data points or less parameters.')
return -1
# this is the fit
if self._output: print(f'^^^^^^^^^^^^^^ start {method} fit ^^^^^^^^^^^^^^')
startfittime = time.time()
global bayes_dL
if method in ['lm', 'trf', 'dogbox']:
if 'max_nfev' not in kw and 'maxiter' in kw:
kw['max_nfev'] = kw.pop('maxiter')
if 'ftol' not in kw and 'tol' in kw:
kw['ftol'] = kw.pop('tol')
if self.has_limit and method in ['lm']:
method = 'trf'
_ = kw.pop('bounds', None) # remove them from kwargs
if method in ['trf', 'dogbox']:
# use soft limit as bounds but stay at hard limits, 'lm' has not included bounds
lb = []
ub = []
for name, values in self._freepar.items():
lv = 1 if isinstance(values, numbers.Number) else len(values)
if name in self._limits:
lb.extend([self._limits[name][0] if self._limits[name][0] is not None else -np.inf] * lv)
ub.extend([self._limits[name][1] if self._limits[name][1] is not None else np.inf] * lv)
else:
lb.extend([-np.inf] * lv)
ub.extend([np.inf] * lv)
bounds = (lb, ub)
else:
# default -> no bounds
bounds = (- np.inf, np.inf)
self._fitmethod = method
res = scipy.optimize.least_squares(self._errorfunction, x0=freeParValues, args=(0,), bounds=bounds,
method=method, **kw)
ier = res.success
mesg = res.message
best = res.x
chi2 = sum(res.fun**2) / self._dof
# calc cov and errors from jacobian as approximation
hessian = np.dot(res.jac.T, res.jac)
try:
hess_inv = np.linalg.inv(hessian)
cov = hess_inv * chi2
best_err = np.sqrt(cov.diagonal())
except (AttributeError, np.linalg.LinAlgError):
try:
# try with pseudo inverse like scipy minpack curve_fit
hess_inv = np.linalg.pinv(hessian)
cov = hess_inv * chi2
# there can be small neg numbers in diagonal or dependent parameters
# if the resulting relative error is
covdiag = cov.diagonal() # write protected
_sel = np.abs(covdiag/best) < 1e-10
cov[_sel, _sel] = 0
best_err = np.sqrt(cov.diagonal())
except np.linalg.LinAlgError:
cov = None
best_err = None
elif method[:3] == 'dif':
self._fitmethod = 'differential_evolution'
if 'max_nfev' in kw and 'maxiter' not in kw:
kw['maxiter'] = kw.pop('max_nfev')
_ = kw.pop('bounds', None) # remove them
bounds = []
for name, values in self._freepar.items():
if name in self._limits:
# use the soft or hard limits with for len of values
bb = (self._limits[name][0] if self._limits[name][0] is not None else self._limits[name][2],
self._limits[name][1] if self._limits[name][1] is not None else self._limits[name][3])
# do for all values
for val in np.atleast_1d(values):
bounds.append(bb)
else:
for val in np.atleast_1d(values):
bounds.append((val / 10 ** 0.5, val * 10 ** 0.5))
# stuff to allow usage of a pool
bayes_dL = self
workers = kw.pop('workers', 1)
disp = kw.pop('disp', True)
polish = kw.pop('polish', False)
output = self._output
self._output =False
if workers == 1 or 'fork' not in mp.get_all_start_methods():
res = scipy.optimize.differential_evolution(func=pool_error,
bounds=bounds, workers=1, disp=disp, polish=polish,
args=(0,), **kw)
else:
# differential_evolution uses Pool with default context 'spawn' on macOS, also for python >3.14
# here I force 'fork'
ncpu = mp.cpu_count() if workers<1 else min(workers, mp.cpu_count())
with mp.get_context('fork').Pool(ncpu) as pool:
res = scipy.optimize.differential_evolution(func=pool_error,
bounds=bounds, workers=pool.map, disp=disp, polish=polish,
updating='deferred', args=(0,), **kw)
self._output = output
ier = res.success
mesg = res.message
best = res.x
chi2 = res.fun
cov = None
best_err = None
elif method in ['bayes']:
self._fitmethod = 'bayes'
output = self._output
self._output =False
_ = kw.pop('max_nfev', None)
_ = kw.pop('maxiter', None)
# setup limits if not given for log prior, keep hard limits
for name, values in self._freepar.items():
val = [values] if isinstance(values, numbers.Number) else values
if name in self._limits:
low = (min(val) / 10**0.5) if self._limits[name][0] is None else self._limits[name][0]
up = (max(val) * 10**0.5) if self._limits[name][1] is None else self._limits[name][1]
self.setlimit(**{name: [low, up, self._limits[name][2], self._limits[name][3]]})
else:
self.setlimit(**{name: [min(val) / 10**0.5, max(val) * 10**0.5]})
# setup sampler
ndim = self._len_p
nwalkers = kw.pop('nwalkers', 2 * ndim)
nsteps = kw.pop('bayesnsteps', 300)
tolerance = kw.pop('tolerance', 50)
# this is ugly but speeds up dramatically as it avoids pickling of the data for each call
bayes_dL = self
if 'fork' in mp.get_all_start_methods():
with mp.get_context('fork').Pool() as pool:
sampler = emcee.EnsembleSampler(nwalkers=nwalkers, ndim=ndim,
log_prob_fn=pool_error, pool=pool, args=(0,))
self._bayes_sampler = sampler
# init in small sphere around start parameters
init = freeParValues * (1 + 1e-4 * np.random.standard_normal(size=(nwalkers, len(freeParValues))))
# run mcmc in steps until trust tolerance is reached
# default of emcee is tolerance = 50 * tau
while True:
state = sampler.run_mcmc(initial_state=init, nsteps=nsteps, progress=True, **kw)
# next time run_mcmc will start from last end
init = None # this takes the last like init = sampler.get_last_sample()
tau = sampler.get_autocorr_time(tol=tolerance, quiet=True).max()
if sampler.get_chain().shape[0]/tau > tolerance:
break
else:
sampler = emcee.EnsembleSampler(nwalkers=nwalkers, ndim=ndim,
log_prob_fn=pool_error, pool=None, args=(0,))
self._bayes_sampler = sampler
# init in small sphere around start parameters
init = freeParValues * (1 + 1e-4 * np.random.standard_normal(size=(nwalkers, len(freeParValues))))
# run mcmc in steps until trust tolerance is reached
# default of emcee is tolerance = 50 * tau
while True:
state = sampler.run_mcmc(initial_state=init, nsteps=nsteps, progress=True, **kw)
# next time run_mcmc will start from last end
init = None # this takes the last like init = sampler.get_last_sample()
tau = sampler.get_autocorr_time(tol=tolerance, quiet=True).max()
if sampler.get_chain().shape[0] / tau > tolerance:
break
self._output = output
tau = sampler.get_autocorr_time(tol=tolerance)
flat_samples = sampler.get_chain(discard=int(max(tau) * 2), flat=True)
log_prob = sampler.get_log_prob(discard=int(max(tau) * 2), flat=True).mean()
ier = True
mesg = ''
best = np.mean(flat_samples, axis=0)
chi2 = (-2 * log_prob - np.sum(np.log(2 * np.pi * self.eY.flatten ** 2))) / self._dof
best_err = np.std(flat_samples, axis=0)
# covariance matrix from samples
cov = np.cov(flat_samples, rowvar=False, ddof=self._dof)
# generate labels
labels =[]
for k, v in self._freepar.items():
if isinstance(v, numbers.Number):
labels.append(k)
else:
labels.extend([k+f'{i:.0f}' for i in range(len(v))])
self._bayes_sampler.parlabels = labels
else:
self._fitmethod = 'minimize_' + method
if 'max_nfev' in kw and 'maxiter' not in kw:
kw['maxiter'] = kw.pop('max_nfev')
if 'tol' not in kw and 'ftol' in kw:
kw['tol'] = kw.pop('ftol')
# sort options and additional kwargs and remove some of them
options = kw.pop('options', {})
for k in list(kw.keys()):
# anything that is not minimize kwarg will go to options which depends on method
if k not in scipy.optimize.minimize.__code__.co_varnames:
options.update({k: kw.pop(k)})
res = scipy.optimize.minimize(self._errorfunction, x0=freeParValues, args=(0,),
method=method, options=options, **kw)
ier = res.success
mesg = res.message
best = res.x
chi2 = res.fun
try:
cov = res.hess_inv * chi2
best_err = np.sqrt(cov.diagonal())
except AttributeError:
cov = None
best_err = None
# remove from global
bayes_dL = None
if ier not in [True, 1, 2, 3, 4]:
# NOT successful fit
if self._output: print(CSIr + 'Error ' + str(mesg) + CSIe)
if self._output: print(CSIr + 'last parameters ' + CSIbr + '!! NOT a fit result !!: ' + CSIe)
i = 0
for name, value in self._freepar.items():
l = len(np.ravel(value))
if self._output: print(name, best[i:i + l])
i += l
if self._output: print(CSIbr + 'fit NOT successful!!' + CSIe)
raise notSuccesfullFitException(mesg)
# -------------------------------------------
# successful fit -->
# add fitted ParNames to self with correct name
i = 0
resultpar = {}
for name, value in self._freepar.items():
l = len(np.ravel(value))
resultpar[name] = best[i:i + l]
self.__setlistattr__(name, best[i:i + l])
if best_err is not None:
self.__setlistattr__(name + '_err', best_err[i:i + l])
i += l
# write lastfit into attribute directly where modelValues uses the parameters set with __setlistattr__
modelValues = self.modelValues(**resultpar)
# a negative integer indicates error was returned from model
if isinstance(modelValues, numbers.Integral):
if self._output:
print(CSIr + 'fit NOT successful!!' + CSIe)
raise notSuccesfullFitException('model returns single integer. Error occurred')
# add modelValues as lastfit attribute
self.__setlistattr__('lastfit', modelValues)
# add results of freepar to lastfit with errors
i = 0
for name, value in self._freepar.items():
l = len(np.ravel(value))
self.lastfit.__setlistattr__(name, best[i:i + l])
if best_err is not None:
self.lastfit.__setlistattr__(name + '_err', best_err[i:i + l])
i += l
# add fixpar to lastfit without error
for key, val in self._fixpar.items():
self.lastfit.__setlistattr__(key, np.atleast_1d(val))
# update the errorplot if existing
if self._errplot:
self.showlastErrPlot(modelValues=modelValues)
# put everything into lastfit
self.lastfit.__setlistattr__('chi2', chi2)
self.lastfit.__setlistattr__('dof', self._dof)
self.lastfit.__setlistattr__('func_code', str(self._code))
self.lastfit.__setlistattr__('func_name', str(self.model.__name__))
if self._output: print(
CSIg + 'fit finished after %.3g s --->> result --->>' % (time.time() - startfittime) + CSIe)
limitweight, limits, hardlimits, nconstrain = self._checklimits(best)
# noinspection PyArgumentEqualDefault
self._show_output(chi2, 1, limits, hardlimits, nconstrain, resultpar)
if self._output:
print('degrees of freedom = ', self._dof)
if cov is not None:
# this output only if cov and errors are defined
self.lastfit.__setlistattr__('cov', cov)
# correlation matrix R_ij = C_ij/C_ii^0.5/C_jj^0.5
# inverse sigma from diagonal
covdiag = cov.diagonal()
# if one covdiag is zero the correlation is undefined and results here as zero
invcii = np.divide(1, covdiag, where=(covdiag != 0)) # 1/cov.diagonal()**0.5
correlation = invcii * cov * invcii[:, None]
correlation = (correlation - np.diag(correlation.diagonal()))
dim = np.shape(correlation)[0]
imax = correlation.argmax()
correlationmax = correlation.max()
# freeparnames as in freeparvalues
freeParNames = reduce(list.__add__, [[k] * len(np.atleast_1d(v)) for k, v in self._freepar.items()])
message = 'nondiag correlation matrix maximum ' + '%.3g' % correlationmax + ' between ' + \
str(freeParNames[imax // dim]) + ' and ' + str(freeParNames[imax % dim]) + '\n'
if self._nozeroerror and self._output:
if abs(correlationmax) < 0.5:
print(CSIg + message + ' <0.5 seems to be OK' + CSIe)
elif 0.5 < abs(correlationmax) < .8:
print(CSIy + message + ' >0.5 seems to be too large' + CSIe)
elif 0.8 < abs(correlationmax):
print(CSIm + message + ' => strong dependent parameters ' + CSIe)
if np.any(covdiag==0):
print(CSIm + ' => correlated parameter marked by error = 0 ' + CSIe)
# only with 1-sigma errors the chi2 should be close to one
if (chi2 - 1) > 10:
print('chi^2 => a bad model or too small error estimates!')
elif 1 < (chi2 - 1) < 10:
print('chi^2 => should be closer to 1 ; Is this a good model; good errors?')
elif 0.2 < (chi2 - 1) < 1:
print('chi^2 => looks quite good; satisfied or try again to get it closer 1?')
elif -0.2 < (chi2 - 1) < 0.2:
print('chi^2 => good!!! not to say its excellent')
elif -0.5 < (chi2 - 1) < -0.2:
print('chi^2 => seems to be overfitted,\n to much parameters or to large error estimates.')
else:
print('overfitting!!!!\n to much parameters or to large error estimates')
else:
if self._output:
print(CSIy +
'No Errors or zeros in Error!! Values are weighted equally! chi2 might not be closed to 1! '
+ CSIe)
else:
if self._output:
print(f'No cov matrix and no errors with {self._fitmethod} or dependent parameters.')
if isinstance(self._output, str):
if self._output.startswith('last'):
return self.lastfit
elif self._output.startswith('best'):
return best, best_err
if self._output:
print('_________' +
CSIg + 'fit successfully converged. We are done here !!' + CSIe +
'__________')
return
def getBayesSampler(self):
"""
Returns Bayes sampler after Bayesian fit for further analysis.
First do a fit with method='bayes' then the sampler can be retrieved.
Returns
-------
emcee sampler
Examples
--------
Access the chain and make a corner plot
First install corner for the corner plot *pip install corner*
See https://corner.readthedocs.io/en/latest/index.html ::
import jscatter as js
import numpy as np
import matplotlib.pyplot as plt
import corner
diffusion=lambda A,D,t,elastic,wavevector=0:A*np.exp(-wavevector**2*D*t)+elastic
i5=js.dL(js.examples.datapath+'/iqt_1hho.dat')[[5,6]]
i5.makeErrPlot(title='diffusion model residual plot')
# get better starting values
i5.fit(model=diffusion,freepar={'D':[0.2],'A':1}, fixpar={'elastic':0.0},
mapNames= {'t':'X','wavevector':'q'},
condition=lambda a:a.X>0.01, method='lm')
# do the emcee Bayes fit
i5.fit(model=diffusion,freepar={'D':i5.D,'A':i5.A}, fixpar={'elastic':0.0},
mapNames= {'t':'X','wavevector':'q'},
condition=lambda a:a.X<80, method='bayes', tolerance=50, bayesnsteps=1000)
# get sampler chain and examine results removing burn in time 2*tau
tau = i5.getBayesSampler().get_autocorr_time(tol=50)
flat_samples = i5.getBayesSampler().get_chain(discard=int(2*tau.max()), thin=1, flat=True)
labels = i5.getBayesSampler().parlabels
plt.ion()
fig = corner.corner(flat_samples, labels=labels,quantiles=[0.16, 0.5, 0.84],show_titles=True,title_fmt='.3f')
plt.show()
# fig.savefig(js.examples.imagepath+'/bayescorner.jpg')
.. image:: ../../examples/images/bayescorner.jpg
:width: 50 %
:align: center
:alt: bayescorner
"""
if self._bayes_sampler is not None:
return self._bayes_sampler
else:
raise UserWarning('First do a fit with method="bayes" ')
def estimateError(self, method='lm', output=True):
r"""
Estimate error using a refined fit (``method='lm'``) if no error is present.
As default .fit method is used with ``method='lm'`` to result in an error.
A previous fit determines the starting parameters to refine the previous fit.
It needs min. around 2*number_freepar function evaluations for a good :math:`\chi^2` minimum.
Errors are found as attributes ``.parname_err``.
See .fit for error determination.
Parameters
----------
method : fit method, default 'lm'
Fit method to use.
The default 'lm' results in error bars.
Some other methods may not deliver errors, but can be used to
restart a fit with the result of a previous fit.
output : bool
Suppress output
Examples
--------
::
import jscatter as js
import numpy as np
data=js.dL(js.examples.datapath+'/iqt_1hho.dat')[::3]
diffusion=lambda t,wavevector,A,D,b:A*np.exp(-wavevector**2*D*t)+b
data.fit(model=diffusion ,
freepar={'D':0.1,'A':1},
fixpar={'b':0.} ,
mapNames= {'t':'X','wavevector':'q'},method='Powell')
data.estimateError()
"""
# test if error exists in lastfit, then we don't need to do it again
if hasattr(self, 'lastfit') and hasattr(self.lastfit, next(iter(self._freepar.keys()))+'_err'):
print('Error exists and is not changed.')
else:
freepar = {}
# create freepar with last fit result as start parameter
try:
for k, v0 in self._freepar.items():
v = getattr(self, k)
freepar.update({k: v[0] if len(v) == 1 else v})
except AttributeError:
raise AttributeError('First do a fit , then estimate error')
# do the fit with same settings as previous fit
self.fit(model=self.model,
freepar=freepar,
fixpar=self._fixpar,
mapNames=self._mapNames,
method=method,
condition=None, # condition was merged into xslice in previous fit, so here we don't use it
xslice=self._xslice,
output=output)
return
def refineFit(self, method='lm', **kw):
"""
Refined fit with starting values from previous fit.
A previous fit determines the starting parameters to refine the previous fit e.g. using a different method.
Parameters
----------
method : fit method, default 'lm'
Fit method to use.
**kw : kwargs
Additional kwargs for fit.
Keywords 'model', 'freepar', 'fixpar', 'condition', 'xslice' are ignored and used from previous fit.
Examples
--------
::
import jscatter as js
import numpy as np
data=js.dL(js.examples.datapath+'/iqt_1hho.dat')[::3]
diffusion=lambda t,wavevector,A,D,b:A*np.exp(-wavevector**2*D*t)+b
data.fit(model=diffusion ,
freepar={'D':0.1,'A':1},
fixpar={'b':0.} ,
mapNames= {'t':'X','wavevector':'q'},method='Powell')
data.refineFit()
"""
freepar = {}
# create freepar with last fit result as start parameter
try:
for k, v0 in self._freepar.items():
v = getattr(self, k)
freepar.update({k: v[0] if len(v) == 1 else v})
except AttributeError:
raise AttributeError('First do a fit , then refine')
# do the fit with same settings as previous fit
_ = kw.pop('model', 0)
_ = kw.pop('freepar', 0)
_ = kw.pop('fixpar', 0)
_ = kw.pop('mapNames', 0)
_ = kw.pop('condition' ,0)
_ = kw.pop('xslice', 0)
self.fit(model=self.model,
freepar=freepar,
fixpar=self._fixpar,
mapNames=self._mapNames,
method=method,
condition=None, # condition was merged into xslice in previous fit, so here we don't use it
xslice=self._xslice,
**kw)
return
# placeholders for errPlot functions
def simulate(self, **kwargs):
r"""
Simulate model results (showing errPlot but without fitting).
Simulate model and show in errPlot if one is open (use .makeErrPlot).
Parameters can be set as in fit (using fixpar,freepar dict) or like calling a function (parameter=value).
Parameters
----------
model : function
The model to evaluate. See .fit.
If not given the last used fit model is tried.
mapNames : dict, required
Map parameter names from model to attribute names in data e.g. {'t':'X','wavevector':'q'}
args, kwargs :
Keyword arguments to pass to model or for plotting (same as in fit method).
Returns
-------
ModelValues : dataList or dataArray
Examples
--------
::
import numpy as np
import jscatter as js
i5=js.dL(js.examples.datapath+'/iqt_1hho.dat')
diffusion=lambda A,D,t,elastic,wavevector=0:(A-elastic)*np.exp(-wavevector**2*D*t)+elastic
diffusion2=lambda A,D,t,elastic,wavevector=0:A*np.exp(-wavevector**2*D*t)+elastic + 1
# like .fit
sim = i5.simulate(model=diffusion,freepar={'D':0.2,'A':1}, fixpar={'elastic':0.0},
mapNames= {'t':'X','wavevector':'q'}, condition=lambda a:a.X>0.01 )
# like calling a function
sim = i5.simulate(model=diffusion2, elastic=0.0, A=1, D=0.2, mapNames= {'t':'X','wavevector':'q'})
i5.makeErrPlot()
i5.fit(model=diffusion,freepar={'D':0.2,'A':1}, fixpar={'elastic':0.0},
mapNames= {'t':'X','wavevector':'q'}, condition=lambda a:a.X>0.01 )
simulatedValues = i5.simulate(D=i5.lastfit.D*2)
simulatedValues1 = i5.simulate(elastic=0.1)
"""
# remove unwanted
kwargs.pop('method', None)
kwargs.pop('condition', None)
# move all fixpar and freepar to one if existing
fixpar = {}
if self._fixpar:
fixpar.update(self._fixpar)
fixpar.update(kwargs.pop('fixpar', {}))
fixpar.update(kwargs.pop('freepar', {}))
mapNames = kwargs.pop('mapNames', self._mapNames)
try:
model = kwargs.pop('model')
except KeyError:
try:
model = self.model
except AttributeError:
raise AttributeError('Missing model')
if model:
# get code for code varnames
try:
# is wrapped in decorator as e.g.sas.smear but we want to look here at the wrapped code
_code = model.__wrapped__.__code__
except AttributeError:
# seems to be normal function
_code = model.__code__
else:
raise AttributeError('No model in dataList or as parameter')
modelParameterNames = _code.co_varnames[:_code.co_argcount]
# take all needed parameters and add it to fixpar
for mPN in modelParameterNames:
val = kwargs.pop(mPN, None)
if val is not None:
fixpar.update({mPN: val})
if not fixpar:
raise AttributeError('No parameters in dataList found')
if not mapNames:
raise AttributeError('No mapNames in dataList found')
kwargs.update(fixpar=fixpar)
kwargs.update(model=model)
kwargs.update(mapNames=mapNames)
# This calls fit in debug mode and plots
# switch to debug=3 which returns modelValues
kwargs.update(debug=3)
try:
oldfixpar = copy.deepcopy(self._fixpar)
oldfreepar = copy.deepcopy(self._freepar)
oldmodel = copy.deepcopy(self.model)
oldmapNames = copy.deepcopy(self._mapNames)
except AttributeError:
oldfixpar = None
_mV = self.fit(**kwargs)
if self.errplot is not None and self.errplot.is_open():
self.showlastErrPlot(modelValues=_mV)
if isinstance(oldfixpar, dict):
# restore old if present
self._fixpar = oldfixpar
self._freepar = oldfreepar
self.model = oldmodel
self._mapNames = oldmapNames
return _mV
def makeNewErrPlot(self, **kwargs):
"""dummy"""
pass
def makeErrPlot(self, **kwargs):
"""dummy"""
pass
def detachErrPlot(self):
"""dummy"""
pass
def killErrPlot(self, **kwargs):
"""dummy"""
pass
def savelastErrPlot(self, **kwargs):
"""dummy"""
pass
def errPlot(self, *args, **kwargs):
"""dummy"""
pass
def showlastErrPlot(self, **kwargs):
"""dummy"""
pass
def errPlotTitle(self, **kwargs):
"""dummy"""
pass
##################################################################################
# dataList including errPlot functions
# noinspection PyIncorrectDocstring
[docs]
class dataList(dataListBase):
def __init__(self, objekt=None,
block=None,
usecols=None,
delimiter=None,
takeline=None,
index=slice(None),
replace=None,
skiplines=None,
ignore='#',
XYeYeX=None,
lines2parameter=None,
encoding=None):
super().__init__(objekt, block, usecols, delimiter, takeline, index, replace, skiplines, ignore, XYeYeX,
lines2parameter, encoding)
# self._errplot = None # in base
self._errplotshowfixpar = None
self._errplottitle = ''
self._errplotLegendPosition = None
self._errplottype = None
self._errplotxscale = None
self._errplotyscale = None
[docs]
def makeNewErrPlot(self, **kwargs):
"""
Creates a NEW ErrPlot without destroying the last. See makeErrPlot for details.
Parameters
----------
**kwargs
Keyword arguments passed to makeErrPlot.
"""
self.detachErrPlot()
self.makeErrPlot(**kwargs)
[docs]
def makeErrPlot(self, title=None, **kwargs):
"""
Creates a GracePlot for intermediate output from fit with residuals.
ErrPlot is updated only if consecutive steps need more than 2 seconds.
The plot can be accessed later as ``.errplot`` .
Parameters
----------
title : string
Title of plot.
residuals : string
Plot type of residuals (=y-f(x,...)).
- 'absolut' or 'a' absolute residuals
- 'relative' or 'r' relative =residuals/y
- 'x2' or 'x' residuals/eY with chi2 =sum((residuals/eY)**2)
showfixpar : boolean default: True
Show the fixed parameters in errplot.
yscale,xscale : 'n','l' for 'normal', 'logarithmic'
Y scale, log or normal (linear)
fitlinecolor : int, [int,int,int]
Color for fit lines (or line style as in plot).
If not given same color as data.
legpos : 'll', 'ur', 'ul', 'lr', [rx,ry]
Legend position shortcut in viewport coordinates.
Shortcuts for lower left, upper right, upper left, lower right
or relative viewport coordinates as [0.2,0.2].
headless : bool, 'agr', 'png', 'jpg', 'svg', 'pnm', 'pdf'
Use errPlot in headless mode (NO-Gui).
True saves to lastErrPlot.agr with regular updates (all 2 seconds).
A file type changes to specified file type as printed.
size : [float, float]
Plot size in inch.
Examples
--------
ErrPlot with fitted data (points), model function (lines) and the difference between both
in an additional plot to highlight differences.
.. image:: ../../examples/images/4sinErrPlot.jpg
:align: center
:width: 50 %
:alt: 4sinErrPlot
"""
headless = kwargs.pop('headless', None)
if headless not in [None, 'agr', 'png', 'jpg', 'svg', 'pnm', 'pdf']:
headless = None
size = kwargs.pop('size', None)
if isinstance(size, numbers.Number):
size = [size, size]
if self._errplot is not None and self._errplot.is_open():
pass
else:
# we need to make a new errPlot with some attributes describing type and legend and scaling
if size:
self._errplot = openplot(size=size)
else:
self._errplot = openplot()
self._errplottitle = ''
self._errplotLegendPosition = None
self._errplotyscale = 'normal'
self._errplotxscale = 'normal'
# headless is set always when new opened
self._errplot._headlessformat = headless
yscale = kwargs.pop('yscale', [None])
if yscale[0] in ['n', 'l']:
self._errplotyscale = yscale
xscale = kwargs.pop('xscale', [None])
if xscale[0] in ['n', 'l']:
self._errplotxscale = xscale
legpos = kwargs.pop('legpos', None)
if legpos in ['ll', 'ur', 'ul', 'lr'] or isinstance(legpos, (list, tuple)):
self._errplotLegendPosition = legpos
# errplot layout
residuals = kwargs.pop('residuals', [None])
if residuals[0] in ['r', 'a', 'x']:
if residuals[0] == 'r':
self._errplottype = 'relative'
elif residuals[0] == 'x' and not np.any([ey is None for ey in self.eY]):
self._errplottype = 'x2'
else:
self._errplottype = 'absolute'
if title is not None:
self._errplottitle = str(title)
self._errplot.Multi(2, 1)
self._errplot[0].Title(self._errplottitle)
self._errplot[0].SetView(0.1, 0.255, 0.95, 0.9)
self._errplot[1].SetView(0.1, 0.1, 0.95, 0.25)
self._errplot[0].Yaxis(label='Y values')
self._errplot[0].Xaxis(label='')
self._errplot[1].Xaxis(label='X values')
if 'fitlinecolor' in kwargs:
self._errplot[0].fitlinecolor = kwargs['fitlinecolor']
del kwargs['fitlinecolor']
if self._errplottype == 'relative':
self._errplot[1].Yaxis(label='residuals/Y')
elif self._errplottype == 'x2':
self._errplot[1].Yaxis(label='residual/eY')
else:
self._errplot[1].Yaxis(label='residuals')
# set scaling
self._errplot[0].Yaxis(scale=self._errplotyscale)
self._errplot[0].Xaxis(scale=self._errplotxscale)
self._errplot[1].Xaxis(scale=self._errplotxscale)
self._errplotshowfixpar = kwargs.pop('showfixpar', True)
self._errplot[0].Clear()
self._errplot[1].Clear()
@property
def errplot(self):
"""
Errplot handle
"""
return self._errplot
[docs]
def detachErrPlot(self):
"""
Detaches ErrPlot without killing it and returns a reference to it.
"""
if self._errplot:
errplot = self._errplot
self._errplot = None
return errplot
[docs]
def errPlotTitle(self, title):
if self._errplot:
self._errplot[0].Title(title)
[docs]
def killErrPlot(self, filename=None):
"""
Kills ErrPlot
If filename given the plot is saved.
"""
if self._errplot:
self.savelastErrPlot(filename)
self._errplot.Exit()
self._errplot = None
[docs]
def savelastErrPlot(self, filename, format=None, size=(3.4, 2.4), dpi=300, **kwargs):
"""
Saves errplot to file with filename.
See graceplot.save
"""
if not self._errplot.is_open():
self.showlastErrPlot(**kwargs)
if filename is not None and isinstance(filename, str):
self._errplot.Save(filename, format=format, size=size, dpi=dpi)
[docs]
def errPlot(self, *args, **kwargs):
"""
Plot into an existing ErrPlot. See Graceplot.plot for details.
"""
if self._errplot and self._errplot.is_open():
self._errplot[0].Plot(*args, **kwargs)
self._errplot[0].Legend()
else:
raise AttributeError('There is no errPlot to plot into')
# noinspection PyBroadException
[docs]
def showlastErrPlot(self, title=None, modelValues=None, **kwargs):
"""
Shows last ErrPlot as created by makeErrPlot with last fit result.
Same arguments as in makeErrPlot.
Additional keyword arguments are passed as in modelValues and simulate changes in the parameters.
Without parameters the last fit is retrieved.
"""
self.makeErrPlot(title=title, **kwargs)
# remove makeErrPlot specific kwargs
if 'fitlinecolor' in kwargs: del kwargs['fitlinecolor']
if 'yscale' in kwargs: del kwargs['yscale']
if 'xscale' in kwargs: del kwargs['xscale']
if 'residuals' in kwargs: del kwargs['residuals']
if 'size' in kwargs: del kwargs['size']
if modelValues is None:
# calculate modelValues if not given
modelValues = self.modelValues(**kwargs)
# generate some useful output from fit parameters
outlist = r''
outerror = r''
for name in sorted(self._freepar):
# here we need the names from modelValues
values = np.atleast_1d(modelValues.dlattr(name))
outlist += r'%-8s=[' % name + r''.join([r' %.4G' % val for val in values]) + r']\n'
try:
outerror += r'%-6s=[' % (name + '_e') + \
r''.join([r' %.3G' % val for val in getattr(self, name + '_err')]) + r']\n'
except:
pass
if self._errplotshowfixpar:
outlist += r'-----fixed-----\n'
for name, values in sorted(self._fixpar.items()):
try:
outlist += r'%-8s=[' % name + r''.join([r' %.4G' % val for val in values]) + r']\n'
except:
outlist += r'%-8s=[%.4G]\n' % (name, values)
# plot the data that contribute to the fit
for XYeY, xslice, c in zip(self, self._xslice, range(1, 1 + len(self.X))):
if hasattr(XYeY, 'eY'):
self._errplot[0].Plot(XYeY.X[xslice], XYeY.Y[xslice], XYeY.eY[xslice],
symbol=[-1, 0.3, c], line=0, comment=outerror)
else:
self._errplot[0].Plot(XYeY.X[xslice], XYeY.Y[xslice],
symbol=[-1, 0.3, c], line=0, comment=outerror)
outerror = ''
# plot modelValues and residuals
residual = []
error = []
# if X axis is changed in kwargs we don't plot residuals
showresiduals = not next((k for k, v in self._mapNames.items() if v == 'X')) in kwargs
for mXX, mYY, XX, YY, eYY, xslice, c in zip(modelValues.X, modelValues.Y,
self.X, self.Y, self.eY, self._xslice, range(1, 1 + len(self.X))):
if hasattr(self._errplot[0], 'fitlinecolor'):
if isinstance(self._errplot[0].fitlinecolor, numbers.Integral):
cc = [1, 1, self._errplot[0].fitlinecolor]
else:
cc = self._errplot[0].fitlinecolor
else:
cc = [1, 1, c]
self._errplot[0].Plot(mXX, mYY, symbol=0, line=cc, legend=outlist, comment=outerror)
# only first get nonempty outlist
outlist = ''
outerror = ''
if showresiduals:
# residuals type
residual.append(YY[xslice] - mYY)
error.append(residual[-1])
if self._errplottype == 'relative':
residual[-1] = (residual[-1] / YY[xslice])
elif self._errplottype == 'x2' and self._nozeroerror:
residual[-1] = (residual[-1] / eYY[xslice])
self._errplot[1].Plot(XX[xslice], residual[-1], symbol=0, line=[1, 1, c], legend=outlist,
comment='r %s' % c)
if self._nozeroerror:
error[-1] = error[-1] / eYY[xslice]
if not showresiduals:
self._errplot[0].Subtitle(r'No residuals as X is changed for simulation.')
return
error = np.hstack(error)
chi2 = sum(error ** 2) / self._dof
try:
factor = 5
residual = np.hstack(residual)
ymin = residual.mean() - residual.std() * factor
ymax = residual.mean() + residual.std() * factor
self._errplot[1].Yaxis(ymin=ymin, ymax=ymax, scale='n')
except:
pass
self._errplot[0].Legend(charsize=0.7, position=self._errplotLegendPosition)
try:
modelname = 'Model ' + str(self.model.__name__)
except:
modelname = ''
self._errplot[0].Subtitle(modelname +
r'; chi\S2\N=' + f'{chi2:.4g}; ' +
f'DOF = {self._lenerror-self._len_p}; ' +
f'parameters {self._len_p}; '+
f' f(X) eval {self.numberOfModelEvaluations}',size=0.9)
if self._errplot.headless:
self.savelastErrPlot(filename='lastErrPlot', format=self._errplot._headlessformat)
##################################################################################
# this will generate automatic attributes
def gen_XYZ(cls, name, ixyz):
"""
generate property with name name that returns column ixyz
cls needs to be accessible as class[ixyz]
Parameters
----------
cls : class with column structure
name : name of the property
ixyz : index of column to return
Returns
-------
array
"""
def get(cls):
if not hasattr(cls, ixyz):
raise AttributeError('dataArray has no attribute ', name)
if not isinstance(getattr(cls, ixyz), numbers.Integral):
raise AttributeError('dataArray. ' + ixyz, 'needs to be integer.')
if cls.ndim == 1:
return cls.view(np.ndarray)
elif cls.ndim > 1:
return cls[getattr(cls, ixyz)].view(np.ndarray)
def set(cls, val):
if not hasattr(cls, ixyz):
raise AttributeError('dataArray has no attribute ', name)
if cls.ndim == 1:
cls[:] = val
elif cls.ndim > 1:
cls[getattr(cls, ixyz), :] = val
# noinspection PyBroadException
def delete(cls):
try:
delattr(cls, ixyz)
except:
pass
docu = """this delivers attributes of dataArray class"""
setattr(cls.__class__, name, property(get, set, delete, doc=docu))
# noinspection PyIncorrectDocstring,PyDefaultArgument,PyMethodOverriding
class dataArrayBase(np.ndarray):
# noinspection PyShadowingBuiltins
def __new__(cls, input=None,
dtype=None,
filename=None,
block=None,
index=0,
usecols=None,
skiplines=None,
replace=None,
ignore='#',
delimiter=None,
takeline=None,
lines2parameter=None,
XYeYeX=None,
encoding=None):
r"""
dataArray (ndarray subclass) with attributes for fitting, plotting, filter.
- A subclass of numpy ndarrays with attributes to add parameters describing the data.
- Allows fitting, plotting, filtering, prune and more.
- .X, .Y, .eY link to specified columns.
- Numpy array functionality is preserved.
- dataArray creation parameters (below) mainly determine how a file is read from file.
- .Y are used as function values at coordinates [.X,.Z,.W] in fitting.
Parameters
----------
input : string, ndarray
Object to create a dataArray from.
- Filenames with extension '.gz' are decompressed (gzip).
- filenames with asterisk like exda=dataList(objekt='aa12*') as input for multiple files.
- An in-memory stream for text I/O (io.StringIO).
dtype : data type
dtype of final dataArray, see numpy.ndarray
index : int, default 0
Index of the dataset in the given input to select one from multiple.
block : string,slice (or slice indices), default None
Indicates separation of dataArray in file if multiple are present.
- None : Auto detection of blocks according to change between datalines and non-datalines.
A new dataArray is created if data and attributes are present.
- string : If block is found at beginning of line a new dataArray is created and appended.
block can be something like "next" or the first parameter name of a new block as block='Temp'
- slice or slice indices :
block=slice(2,100,3) slices the filelines in file as lines[i:j:k] .
If only indices are given these are converted to slice.
XYeYeX : list integers, default=[0,1,2,None,None,None]
Columns for X, Y, eY, eX, Z, eZ, W, eW.
Change later with eg. setColumnIndex(3,5,32).
Values in dataArray can be changed by dataArray.X=[list of length X ].
usecols : list of integer
Use only given columns and ignore others (after skiplines).
ignore : string, default '#'
Ignore lines starting with string e.g. '#'.
For more complex lines to ignore use skiplines.
replace : dictionary of [string,regular expression object]:string
String replacement in each read line as {'old':'new',...} (done after takeline).
- This is done prior to determining line type and can be used to convert strings to number or remove
bad characters {',':'.'}.
- If dict key is a regular expression object (e.g. rH=re.compile('H\d+') ),it is replaced by string.
See Python module re for syntax.
skiplines : function returning bool, list of string or single string
Skip if line meets condition. Function gets the list of words in a data line.
Examples ::
# words with exact match
skiplines = lambda w: any(wi in w for wi in ['',' ','NAN',''*****])
skiplines = lambda w: any(float(wi)>3.1411 for wi in w)
skiplines = lambda w: len(w)==1
# skip explicitly empty lines
skiplines = lambda w: len(w)==0
If a list is given, the lambda function is generated automatically as in above example.
If single string is given, it is tested if string is a substring of a word ( 'abc' in '123abc456')
delimiter : string
Separator between data fields in a line, default any whitespace.
E.g. '\\t' tabulator
takeline : string,list of string, function
Filter lines to be included (all lines) e.g. to select line starting with 'ATOM'.
Should be combined with: replace (replace starting word by number {'ATOM':1} to be detected as data)
and usecols to select the needed columns.
Examples (function gets words in line):
- lambda words: any(w in words for w in ['ATOM','CA']) # one of both words somewhere in line
- lambda w: (w[0]=='ATOM') & (w[2]=='CA') # starts with 'ATOM' and third is 'CA'
For word or list of words first example is generated automatically.
lines2parameter : list of integer
List of lines i which to prepend with 'line_i' to be found as parameter line_i.
Used to mark lines with parameters without name (only numbers in a line as in .pdh files in the header).
E.g. to skip the first lines.
XYeYeX : list integers, default=[0,1,2,None,None,None]
Sets columns for X, Y, eY, eX, Z, eZ, W, eW.
This is ignored for dataList and dataArray objects as these have defined columns.
Change later by: data.setColumnIndex .
encoding : None, 'utf-8', 'cp1252', 'ascii'
The encoding of the files read. By default the system default encoding is used.
Others python2.7='ascii', python3='utf-8',
Windows_english='cp1252', Windows_german='cp1251'
Returns
-------
dataArray
Notes
-----
- Attributes to avoid (they are in the name space of numpy ndarrays):
T,mean,max,min,... These names are substitute by appended '_' (underscore) if found in read data.
Get a complete list by "dir(np.array(0))".
- Avoid attribute names including special math characters as " ** + - / & ".
Any char that can be interpreted as a function (eg datalist.up-down)
will be interpreted from python as : updown=datalist.up operator(minus) down
and result in: AttributeError.
To get the values use getattr(dataList,'up-down') or avoid usage of these characters.
- If an attribute 'columnname' exists with a string containing columnnames separated by semicolon
the corresponding columns can be accessed in 2 ways ( columnname='wavevector; Iqt' ):
- attribute with prepended underscore '_'+'name' => data._Iqt
- columnname string used as index => data['Iqt']
From the names all char like "+-*/()[]()|§$%&#><°^, " are deleted.
The columnname string is saved with the data and is restored when rereading the data.
This is intended for reading and not writing.
**Data access/change** ::
exa=js.dA('afile.dat')
exa.columnname='t; iqt; e+iqt' # if not given in read file
exa.eY=exa.Y*0.05 # default for X, Y is column 0,1; see XYeYeX or .setColumnIndex ; read+write
exa[-1]=exa[1]**4 # direct indexing of columns; read+write
exa[-1,::2]=exa[1,::2]*4 # direct indexing of columns; read+write; each second is used (see numpy)
eq1=exa[2]*exa[0]*4 # read+write
iq2=exa._iqt*4 # access by underscore name; only read
eq3=exa._eiqt*exa._t*4 # read
iq4=exa['iqt']*4 # access like dictionary; only read
eq5=exa['eiqt']*exa['t']*4 # read
aa=np.r_[[np.r_[1:100],np.r_[1:100]**2]] #load from numpy array
daa=js.dA(aa) # with shape
daa.Y=daa.Y*2 # change Y values; same as daa[1]
dbb=js.zeros((4,12)) # empty dataArray
dbb.X=np.r_[1:13] # set X
dbb.Y=np.r_[1:13]**0.5 # set Y
dbb[2]=dbb.X*5
dbb[3]=0.5 # set 4th column
dbb.a=0.2345
dbb.setColumnIndex(ix=2,iy=1,iey=None) # change column index for X,Y, end no eY
Selecting ::
ndbb=dbb[:,dbb.X>20] # only X>20
ndbb=dbb[:,dbb.X>dbb.Y/dbb.a] # only X>Y/a
**Read/write** ::
import jscatter as js
#load data into dataArray from ASCII file, here load the third datablock from the file.
daa=js.dA('./exampleData/iqt_1hho.dat',index=2)
dbb=js.ones((4,12))
dbb.ones=11111
dbb.save('folder/ones.dat')
dbb.save('folder/ones.dat.gz') # gziped file
**Rules for reading of ASCII files**
"""
# read from input to get data
if isinstance(input, str):
# if a filename is given read it
if os.path.isfile(input):
input = _parsefile(input, block=block, usecols=usecols, skiplines=skiplines, replace=replace,
ignore=ignore, delimiter=delimiter, takeline=takeline,
lines2parameter=lines2parameter, encoding=encoding)
if not input:
raise IOError('nothing read from file.')
else:
raise NameError('file does not exist . ',input)
elif isinstance(input, dict) and 'val' in input:
# was output of _read
input = [input]
index = 0
elif input is None:
# creates empty dataArray
return zeros(0)
elif all([isinstance(zz, str) for zz in input]) and len(input) > 0:
# a list with lines e.g. from a file or something else
# just interpret it in _read
input = _parsefile(input, block=block, usecols=usecols, skiplines=skiplines, replace=replace, ignore=ignore,
delimiter=delimiter, takeline=takeline, lines2parameter=lines2parameter,
encoding=encoding)
# create dataArray dependent on input
if hasattr(input, '_isdataArray'):
# for completeness if it was a dataArray
return input
elif isinstance(input, np.ndarray):
# create dataArray from numpy array
if dtype is None:
dtype = input.dtype
else:
dtype = np.dtype(dtype)
# Input array is an already formed ndarray instance
# We cast to be our class type taking the view (no copy of data!)
data = np.asanyarray(input, dtype=dtype).view(cls)
data.comment = []
elif isinstance(input, list):
# create dataArray from a given list like the output from _read; default
# file already read by _read, so we need to search for internal links like @name
input = _searchForLinks(input)
# check dtype of original data
if dtype is None:
dtype = input[int(index)]['val'].dtype
else:
dtype = np.dtype(dtype)
# now create the dataArray as cls and create attributes from para
data = np.asanyarray(input[int(index)]['val'], dtype=dtype).view(cls)
data.comment = input[int(index)]['com']
# filter for case-insensitive 'xyeyex'
para = {(k.lower() if k.lower() == 'xyeyex' else k): v for k, v in input[int(index)]['para'].items()}
data.setattr(para)
data.raw_data = input[index:index]
data._orgcommmentline = input[int(index)]['_original_comline']
else:
raise Exception('nothing useful found to create dataArray')
# set/recover column indices and defines ._ix,._iy,._iey and X,Y,EY...
# if first in XYeYeX was char we tak it from comments
data.getfromcomment(attrname='xyeyex', ignorecase=True)
if XYeYeX is None:
XYeYeX = getattr(data, 'xyeyex', [0, 1, 2])
# XYeYeX = (0, 1, 2) # default values
try:
delattr(data, 'xyeyex')
except AttributeError:
pass
data.setColumnIndex(XYeYeX)
# generate columnname if existent in comments
data.getfromcomment('columnname')
data._isdataArray = True
data._asdataList = None
return data
# add docstring from _read
__new__.__doc__ += _parsefile.__doc__
# add docstring from __new__ to class docstring to show this in help
__doc__ = __new__.__doc__
def __array_finalize__(self, obj):
"""
Finalize our dataArray to have attributes and updated parameters.
Here we look in __dict__ if we have new dynamical created attributes
and inherit them to slices or whatever
remember ndarray has no __dict__ and all in it belongs to dataArrays
"""
if obj is None:
return
if np.ndim(obj) == 0:
# don't attach attrib
return
# copy the columnIndices from obj
self.setColumnIndex(obj)
self._isdataArray = True
self._asdataList = None # do not copy the dataList representation
if hasattr(obj, '__dict__'):
for attribut in obj.attr + ['_orgcommentline']:
# noinspection PyBroadException
try:
if attribut not in protectedNames:
self.__dict__[attribut] = getattr(obj, attribut)
except:
pass
def __array_wrap__(self, out_arr, context=None):
# When a numpy ufunc is called on a subclass of ndarray, the __array_wrap__ method is called
# to transform the result into a new instance of the subclass.
# By default, __array_wrap__ will call __array_finalize__ and the attributes will be inherited.
#
# some ufunc return context as 3-element tuple: (name, argument, domain) of the ufunc.
#
# out_array is the return of ufunc(self)
if np.ndim(out_arr) == 0:
# for zero dim we return array
return out_arr
x = np.ndarray.__array_wrap__(self, out_arr, context)
return x
def __reduce__(self):
"""
Needed to pickle dataArray including the defined attributes .
"""
# from https://stackoverflow.com/questions/26598109/
# preserve-custom-attributes-when-pickling-subclass-of-numpy-array
# Get the parent's __reduce__ tuple
pickled_state = super().__reduce__()
# Create our own tuple to pass to __setstate__ with added __dict__
new_state = pickled_state[2] + (self.__dict__.copy(),)
# Return a tuple that replaces the parent's __setstate__ tuple with our own
return pickled_state[0], pickled_state[1], new_state
def __setstate__(self, state):
"""
Needed to unpickle dataArray including the defined attributes.
"""
self.__dict__.update(state[-1]) # Set the stored __dict__ attribute
# Call the parent's __setstate__ with the other tuple elements.
super().__setstate__(state[0:-1])
# regenerate automatic attributes using protectedIndicesNames
self.setColumnIndex(self)
@property
def name(self):
"""
name, mainly the filename of read data files.
"""
return getattr(self, '@name')
# noinspection PyBroadException
def setColumnIndex(self, *args, **kwargs):
"""
Set the column index where to find X,Y,Z,W, eY, eX, eZ
Parameters
----------
ix,iy,iey,iex,iz,iez,iw,iew : integer, None
Set column index, where to find X, Y, eY.
- Default from initialisation is ``ix,iy,iey,iex,iz,iez,iw,iew=0,1,2,None,None,None,None,None``.
(Usability wins iey=2!!)
- If dataArray is given the ColumnIndex is copied, everything else is ignored.
Negative indices stay as they are if the number of columns is different.
- If list [0,1,3] is given these are used as [ix,iy,iey,iex,iz,iez,iw,iew].
Remember that negative indices always are counted from back.
Notes
-----
- *integer*: column index, values outside range are treated like *None*.
- *None*,'-': reset as not used e.g.iex=None -> no errors for x
- Anything else is ignored.
- Take care that negative integers as -1 is counted from the back.
- For array.ndim=1 -> ix=0 and others=None as default.
Examples
--------
::
data.setColumnIndex(ix=2,iy=3,iey=0,iex=None)
# remove y error in (only needed if 3rd column present)
data.setColumnIndex(iey=None)
# add Z, W column for 3D data
data.setColumnIndex(ix=0, iz=1, iw=2, iy=3)
"""
# ixiyiey contains the indices where to find columns
ixiyiey = [''] * len(protectedIndicesNames)
if args:
if hasattr(args[0], '_isdataArray'):
# copy the ColumnIndex from objekt in ix
ixiyiey = (getattr(args[0], pIN) if hasattr(args[0], pIN) else None for pIN in protectedIndicesNames)
elif isinstance(args[0], (tuple, list)):
# if a list is given as argument
ixiyiey = ([_w2i(a) for a in args[0]] + [''] * len(protectedIndicesNames))[:len(protectedIndicesNames)]
elif isinstance(args[0], str):
ixiyiey = [_w2i(w) for w in args[0].split()]
ixiyiey = (ixiyiey + [''] * len(protectedIndicesNames))[:len(protectedIndicesNames)]
else:
ixiyiey = [kwargs[ix[1:]] if ix[1:] in kwargs else '' for ix in protectedIndicesNames]
if self.ndim == 1:
# in this case icol<self.shape[0]
ixiyiey = [0] + [None] * (len(protectedIndicesNames) - 1)
for icol, name, icolname in zip(ixiyiey,
protectedNames,
protectedIndicesNames):
if isinstance(icol, numbers.Integral):
if icol < self.shape[0]: # accept only if within number of columns
setattr(self, icolname, icol)
gen_XYZ(self, name, icolname)
else:
try:
delattr(self, name)
except:
pass
elif icol is None or icol == '-':
try:
delattr(self, name)
except:
pass
@property
def columnIndex(self):
"""
Return defined column indices.
"""
ci = {name: getattr(self, idx, None) for name, idx in zip(protectedNames, protectedIndicesNames)}
return ci
# noinspection PyBroadException
def __deepcopy__(self, memo):
cls = self.__class__
# deepcopy of the ndarray
result = cls(copy.deepcopy(self.array, memo))
# add to memo
memo[id(self)] = result
# copy attributes .attr has only the correct attributes and no private stuff
for k in self.attr + protectedIndicesNames:
try:
setattr(result, k, copy.deepcopy(getattr(self, k), memo))
except:
pass
# copy ColumnIndex
result.setColumnIndex(self)
return result
def nakedCopy(self):
"""
Deepcopy without attributes, thus only the data.
"""
cls = self.__class__
return cls(copy.deepcopy(self.array))
def copy(self):
"""
Deepcopy of dataArray
To make a normal shallow copy use copy.copy
"""
return copy.deepcopy(self)
def __getattribute__(self, attribute):
return np.ndarray.__getattribute__(self, attribute)
def __getattr__(self, attribute):
"""x.__getattr__('name') <==> x.name
if operator char like + - * / in attribute name
use getattr(dataArray,'attribute') to get the value
"""
# ----for _access
if attribute not in protectedNames + protectedIndicesNames + ['_isdataArray', '_isdataList']:
if attribute[0] == '_' and attribute[1] != '_' and hasattr(self, 'columnname'):
if self.ndim<2:
raise IndexError('String indexing not allowed for ndim < 2.')
columnnames = _deletechars(self.columnname, '+-*/()[]()|§$%&#><°^, ').split(';')
if attribute[1:] in columnnames:
return self[columnnames.index(attribute[1:])].view(np.ndarray)
return np.ndarray.__getattribute__(self, attribute)
def setattr(self, objekt, prepend='', keyadd='_'):
"""
Set (copy) attributes from objekt.
Parameters
----------
objekt : objekt or dictionary
Can be a dictionary of names:value pairs like {'name':[1,2,3,7,9]}
If object is dataArray the attributes from dataArray.attr are copied
prepend : string, default ''
Prepend this string to all attribute names.
keyadd : char, default='_'
If reserved attributes (T, mean, ..) are found the name is 'T'+keyadd
"""
if hasattr(objekt, '_isdataArray'):
for attribut in objekt.attr:
try:
setattr(self, prepend + attribut, getattr(objekt, attribut))
except AttributeError:
self.comment.append('mapped ' + attribut + ' to ' + attribut + keyadd)
setattr(self, prepend + attribut + keyadd, getattr(objekt, attribut))
elif isinstance(objekt, dict):
for key in objekt:
try:
setattr(self, prepend + key, objekt[key])
except AttributeError:
self.comment.append('mapped ' + key + ' to ' + key + keyadd)
setattr(self, prepend + key + keyadd, objekt[key])
def __getitem__(self, idx):
# the following allows to use columnnames as items
# use protectedNames or columnname
if isinstance(idx, str) or ((type(idx) is tuple) and isinstance(idx[0], str)):
if self.ndim<2:
raise IndexError('String indexing not allowed for ndim < 2.')
if isinstance(idx, str):
idx0, idx1 = idx, ()
else:
idx0, idx1 = idx[0], idx[1:]
if idx0 in protectedNames:
try:
idx0 = getattr(self, '_i' + idx0.lower())
except AttributeError:
raise AttributeError('Given column not valid for this dataArray. Check columnIndex.')
else:
try:
columnnames = _deletechars(self.columnname, '+-*/()[]()|§$%&#><°^, ').split(';')
idx0 = columnnames.index(idx0)
except (IndexError, ValueError):
raise IndexError('Given string not in .columnname.')
idx = (idx0,) + idx1
return super(dataArrayBase, self).__getitem__(idx).view(np.ndarray)
result = super(dataArrayBase, self).__getitem__(idx)
try:
# slice columnname like idx[0]
columnname=self.columnname.split(';')[idx if isinstance(idx, numbers.Integral) else idx[0]]
if not isinstance(columnname, str):
columnname = ';'.join(columnname)
setattr(result, 'columnname', columnname)
except:
try:
del result.columnname
except:
pass
return result
@property
def array(self):
"""
Strip attributes and return a simple ndarray.
"""
return self.view(np.ndarray)
@inheritDocstringFrom(np.ndarray)
def argmin(self, axis=None, out=None):
return self.array.argmin(axis=axis, out=out)
@inheritDocstringFrom(np.ndarray)
def argmax(self, axis=None, out=None):
return self.array.argmax(axis=axis, out=out)
def prune(self, lower=None, upper=None, number=None, kind='lin', col='X', weight='eY',
keep=None, type='mean', fillvalue=None):
r"""
Reduce number of values between upper and lower limits by selection or averaging in intervals.
Reduces dataArrays size. New values may be determined as next value, average in intervals,
sum in intervals or by selection.
Parameters
----------
lower : float
Lower bound
upper : float
Upper bound
number : int
Number of points between [lower,upper] resulting in number intervals.
kind : 'log','lin', array, default 'lin'
Determines how new points were distributed.
- explicit list/array of new values as [1,2,3,4,5].
Interval borders were chosen in center between consecutive values.
Outside border values are symmetric to inside.
*number*, *upper*, *lower* are ignored.
The value in column specified by *col* is the average found in the interval.
The explicit values given in kind, can be set after using prune for the column given in col.
- 'log' closest values in log distribution with *number* points in [lower,upper]
- 'lin' closest values in lin distribution with *number* points in [lower,upper]
- If *number* is None all points between [lower,upper] are used.
type : {'next','mean','error','mean+error','sum'} default 'mean'
How to determine the value for a new point.
- 'sum' : Sum in intervals.
The *col* column will show the average (=sum/numberofvalues).
The last column contains the number of summed values.
- 'mean' : mean values in interval;
Give `weight` to get a weigthed mean.
- 'mean+std' : Calc mean and add error columns as standard deviation in intervals.
Give `weight` to get a weigthed mean.
Can be used if no errors are present to generate errors as std in intervals.
For single or *keep* values the error is interpolated from neighbouring values.
! For less pruned data error may be bad defined if only a few points are averaged.
col : 'X','Y'....., or int, default 'X'
Column to prune along X,Y,Z or index of column.
weight : None, protectedNames as 'eY' or int
Column of errors=w for weight as 1/w² in 'mean' calculation.
Weight column gets finally new error :math:`(\sum_i(1/w_i^2))^{-0.5}`
- None is equal weight
- If weight not existing or contains zeros equal weights are used.
keep : list of int
List of indices to keep in any case e.g. *keep=np.r_[0:10,90:101]*.
Missing error values for *type='mean+error'* are interpolated.
fillvalue : float, 'remove', 'interp', default='remove'
Fillvalue for empty intervals.
- float: explicit value
- 'remove': removes empty interval
- 'interp': interpolate missing values by linear interpolation.
Returns
-------
dataArray with values pruned to *number* values.
Examples
--------
::
import jscatter as js
import numpy as np
x=np.r_[0:10:0.01]
data=js.dA(np.c_[x,np.sin(x)+0.2*np.random.randn(len(x)),x*0+0.2].T) # simulate data with error
p=js.grace()
p.plot(data,le='original',sy=[1,0.3,11])
p.plot(data.prune(lower=1,upper=5,number=100,type='mean+'),le='mean')
p.plot(data.prune(lower=5,upper=8,number=100,type='mean+',keep=np.r_[1:50]),le='mean+keep')
p.plot(data.prune(lower=1,upper=10,number=40,type='mean+',kind='log'),sy=[1,0.5,5],le='log')
p.plot(data.prune(lower=8).prune(number=10,col='Y'),sy=[1,0.5,7],le='Y prune')
p.legend(x=0,y=-1)
# p.save(js.examples.imagepath+'/prune.jpg')
.. image:: ../../examples/images/prune.jpg
:align: center
:width: 50 %
:alt: prune example
Notes
-----
Attention !!!!
- Dependent on the distribution of points a lower number of new points can result for fillvalue='remove'.
e.g. think of noisy data between 4 and 5 and a lin distribution from 1 to 10 of 9 points
as there are no data between 5 and 10 these will all result in 5 and be set to 5 to be unique.
- For asymmetric distribution of points in the intervals or at intervals at the edges
the pruned points might be different than naively expected,
specifically not being equidistant relative to neighboring points.
To force the points of *col* set these explicitly.
"""
# values to keep
if keep is not None:
keep = np.array([i in keep for i in range(len(self.X))], dtype=bool)
temp = self[:, ~keep].array
keep = self[:, keep].array
else:
temp = self.array
# isinstance to avoid Future warning from numpy about elementwise comparison if kind is numpy array
if number is not None and (isinstance(kind, str) and kind == 'all'):
kind = 'lin'
if col in protectedNames:
col = getattr(self, '_i' + col.lower())
val = temp[int(col)]
if weight in protectedNames:
if hasattr(self, '_i' + weight.lower()):
weight = getattr(self, '_i' + weight.lower())
else:
weight = None
if weight is None:
# then no weights err=1 as equal weight
wval = np.ones_like(temp[int(col)])
else:
if np.any(temp[int(weight)] == 0.):
print('Prune found zeros in weight, so it ignored weight.')
weight = None
wval = np.ones_like(temp[int(col)])
else:
wval = 1. / temp[int(weight)] ** 2
if isinstance(kind, (set, list, np.ndarray)):
# use explicit list and reset following
kind = np.unique(kind)
number = len(kind)
lower = None
upper = None
# determine min and max from values and use only these
valmin = np.max([np.min(val), lower]) if lower is not None else np.min(val)
valmax = np.min([np.max(val), upper]) if upper is not None else np.max(val)
temp = temp[:, (val >= valmin) & (val <= valmax)]
wval = wval[(val >= valmin) & (val <= valmax)]
# update values
val = temp[int(col)]
if number is None:
# only keep, upper and lower important
if keep is not None:
temp = np.c_[keep, temp]
temp = dataArray(temp)
temp.setattr(self)
temp.setColumnIndex(self)
return temp
# We have to find the new intervals pruneval
if isinstance(kind, (set, list, np.ndarray)):
# an explicitly given list of values in kind
diff = np.diff(kind) / 2
pruneval = np.r_[kind[0] - diff[0], kind[:-1] + diff, kind[-1] + diff[-1]]
elif kind[:3] == 'log':
# log distributed points
pruneval = loglist(valmin, valmax, number + 1)
else:
# lin distributed points as default
pruneval = np.r_[valmin:valmax:(number + 1) * 1j]
# calc the new values
if type[:4] == 'mean':
nn = self.shape[0]
# out is one smaller than selected as we look at the intervals
if type == 'mean':
out = np.zeros((nn, number))
else:
out = np.zeros((nn * 2, number)) # for error columns
# mark all intervals as filled, set to False if empty later
nonempty = np.ones(number, dtype=bool)
for i, low, upp in zip(range(number), pruneval[:-1], pruneval[1:]):
# weighted average
if i < number - 1:
select = (low <= val) & (val < upp)
else:
select = (low <= val) & (val <= upp)
if not np.any(select):
# marks empty intervals
nonempty[i] = False
continue
out[:nn, i] = (temp[:, select] * wval[select]).sum(axis=1) / wval[select].sum()
# error from error propagation for weight
wv = wval[select]
if weight is not None and len(wv) > 1:
out[weight, i] = np.sqrt(1 / (wv.sum() * (len(wv) - 1)))
if type != 'mean':
# is more than 'mean' => error need to be calculated with weight and attached
if len(wv) > 1:
out[nn:, i] = temp[:nn, select].std(axis=1)
# deal with empty intervals using fillvalue
if fillvalue == 'remove' or fillvalue is None:
out = out[:, nonempty]
elif isinstance(fillvalue, numbers.Number):
out[:, ~nonempty] = fillvalue
out[col, ~nonempty] = 0.5 * (pruneval[:-1] + pruneval[1:])[~nonempty]
elif fillvalue == 'interp':
out[col, ~nonempty] = 0.5 * (pruneval[:-1] + pruneval[1:])[~nonempty]
for i in range(len(out)):
if i != col:
out[i, ~nonempty] = np.interp(out[col, ~nonempty], out[col, nonempty], out[i, nonempty])
else:
pass
if keep is not None:
try:
out = np.c_[keep, out]
except ValueError:
# additional columns from errors are missing so add them, later will be interpolated
out = np.c_[np.r_[keep, np.zeros_like(keep)], out]
temp = dataArray(out)
temp.setattr(self)
temp.setColumnIndex(self)
# find indices of error=0 which could make trouble. These come from non-average as it was single number
if type != 'mean':
# interpolate from neighbours to get an error estimate
# keep values might get the error of the border
bzeros = (temp[nn, :] == 0)
for inn in range(nn, len(temp)):
temp[inn, bzeros] = np.interp(temp.X[bzeros], temp[col, ~bzeros], temp[inn, ~bzeros])
# set attributes that errors can be found
temp.setColumnIndex(
iex=(getattr(self, '_ix') + nn if (hasattr(self, 'X') and not hasattr(self, 'eX')) else ''),
iey=(getattr(self, '_iy') + nn if (hasattr(self, 'Y') and not hasattr(self, 'eY')) else ''),
iez=(getattr(self, '_iz') + nn if (hasattr(self, 'Z') and not hasattr(self, 'eZ')) else ''),
iew=(getattr(self, '_iw') + nn if (hasattr(self, 'W') and not hasattr(self, 'eW')) else ''))
return temp
elif type[:3] == 'sum':
nn = self.shape[0]
# we return the sum in intervals
# out is one smaller than selected number as we look at the intervals and we need one column for counting
out = np.zeros((nn + 1, number))
# mark all intervals as filled, set to False if empty later
nonempty = np.ones(number, dtype=bool)
for i, low, upp in zip(range(number), pruneval[:-1], pruneval[1:]):
# weighted average
if i < number - 1:
select = (low <= val) & (val < upp)
else:
select = (low <= val) & (val <= upp)
if not np.any(select):
# marks empty intervals
nonempty[i] = False
continue
out[:nn, i] = (temp[:, select]).sum(axis=1)
out[nn, i] = np.sum(select) # counting
out[int(col), i] = out[int(col), i] / out[nn, i]
# deal with empty intervals using fillvalue
if fillvalue == 'remove' or fillvalue is None:
out = out[:, nonempty]
elif isinstance(fillvalue, numbers.Number):
out[:, ~nonempty] = fillvalue
out[col, ~nonempty] = 0.5 * (pruneval[:-1] + pruneval[1:])[~nonempty]
elif fillvalue == 'interp':
out[col, ~nonempty] = 0.5 * (pruneval[:-1] + pruneval[1:])[~nonempty]
for i in range(len(out)):
if i != col:
out[i, ~nonempty] = np.interp(out[col, ~nonempty], out[col, nonempty], out[i, nonempty])
else:
pass
if keep is not None:
out = np.c_[keep, out]
temp = dataArray(out)
temp.setattr(self)
temp.setColumnIndex(self)
return temp
def interpolate(self, X, left=None, right=None, deg=1, col=None):
"""
Piecewise spline interpolated values for a column.
Parameters
----------
X : array,float
New values to interpolate in .X.
left : float
Value to return for `X < X[0]`, default is `Y[0]`.
right : float
Value to return for `X > X[-1]`, defaults is `Y[-1]`.
deg : str or int, optional default =1
Specifies the kind of interpolation as a string (‘linear’, ‘nearest’, ‘zero’, ‘slinear’, ‘quadratic’,
‘cubic’, ‘previous’, ‘next’, where ‘zero’, ‘slinear’, ‘quadratic’ and ‘cubic’ refer to a
spline interpolation of zeroth, first, second or third order; ‘previous’ and ‘next’
simply return the previous or next value of the point) or as an integer specifying the
order of the spline interpolator to use.
scipy.interpolate.interp1d is used with specified fill_value=(left,right) and kind=deg.
col : string, int
Which column to interpolate. Default is 'Y'.
Can be index number, columnname or in ['X','Y','eY',...].
Returns
-------
dataArray
Notes
-----
See numpy.interp. Sorts automatically along X.
"""
if X is None:
raise TypeError('X values missing.')
if col is None:
col = 'Y'
X = np.atleast_1d(X)
xsort = self.X.argsort()
# boundaries
if left is None:
left = self[col][xsort[0]]
if right is None:
right = self[col][xsort[-1]]
ifunc = scipy.interpolate.interp1d(self.X[xsort], self[col][xsort], kind=deg,
bounds_error=False, fill_value=(left, right))
result = dataArray(np.c_[X, ifunc(X)].T)
# copy attributes
result.setattr(self)
return result
def interp(self, X, left=None, right=None, col=None):
"""
Piecewise linear interpolated values for a column (faster).
Parameters
----------
X : array,float
Values to interpolate
left : float
Value to return for `X < X[0]`, default is `col[0]`.
right : float
Value to return for `X > X[-1]`, defaults is `col[-1]`
col : string, int
Which column to interpolate. Default is 'Y'.
Can be index number, columnname or in ['X','Y','eY',...].
Returns
-------
array 1D only interpolated values.
Notes
-----
See numpy.interp. Sorts automatically along X.
"""
if X is None:
raise TypeError('X values missing.')
if col is None:
col = 'Y'
X = np.atleast_1d(X)
xsort = self.X.argsort()
return np.interp(X, self.X[xsort], self[col][xsort], left=left, right=right)
def interpAll(self, X=None, left=None, right=None):
"""
Piecewise linear interpolated values of all columns.
Parameters
----------
X : array like
New values where to interpolate
left : float
Value to return for `X < X[0]`, default is `Y[0]`.
right : float
Value to return for `X > X[-1]`, defaults is `Y[-1]`.
Returns
-------
dataArray, here with X,Y,Z preserved and all attributes
Notes
-----
See numpy.interp. Sorts automatically along X.
"""
if X is None:
raise TypeError('X values missing.')
X = np.atleast_1d(X)
newself = zeros((self.shape[0], np.shape(X)[0]))
xsort = self.X.argsort()
columns = list(range(self.shape[0]))
# noinspection PyUnresolvedReferences
columns.pop(self._ix)
newself[self._ix] = X
for i in columns:
newself[i] = np.interp(X, self.X[xsort], self[i][xsort], left=left, right=right)
newself.setattr(self)
newself.setColumnIndex(self)
return newself
def polyfit(self, X=None, deg=1, function=None, efunction=None, col=None):
"""
Inter(extra)polated values for column using a polyfit.
Extrapolation is done by using a polynominal fit over all Y with weights eY.
To get the correct result the output needs to be evaluated by the inverse of function (if used).
Other columns can be used without weight.
Parameters
----------
X : arraylike
X values where to calculate Y
If None then X=self.X e.g. for smoothing/extrapolation.
deg : int
Degree of polynom used for interpolation see numpy.polyfit
function : function or lambda
Used prior to polyfit as polyfit( function(Y) )
efunction : function or lambda
Used prior to polyfit for eY as weights = efunction(eY)
efunction should be built according to error propagation.
col : string, int
Which column to interpolate. Default is 'Y'.
Can be index number, columnname or in ['X','Y','eY',...].
Returns
-------
dataArray
Examples
--------
Examples assumes exp decaying data (diffusion).
To fit a linear function in the exponent we use log before fitting of the polyfit.
Later we recover the data using the inverse function (exp).
::
import jscatter as js
import numpy as np
t = np.r_[0:100:10]
q=0.5
D=0.2
data = js.dA(np.c_[t, np.exp(-q**2*D*t)].T)
pf=data.polyfit(np.r_[0:100], deg=1, function=np.log)
p=js.grace()
p.plot(data,le='original')
p.plot(pf.X,np.exp(pf.Y),sy=[1,0.1,2],le='polyfit data')
p.yaxis(scale='log')
p.legend(x=10,y=0.1)
"""
if X is None:
raise TypeError('X values missing.')
if col is None:
col = 'Y'
X = np.atleast_1d(X)
if function is None:
function = lambda y: y
efunction = None
if efunction is None:
efunction = lambda ey: ey
if col == 'Y':
if hasattr(self, 'eY'):
poly = np.polyfit(x=self.X, y=function(self.Y), deg=deg, w=efunction(self.eY))
else:
poly = np.polyfit(self.X, function(self.Y), deg)
else:
poly = np.polyfit(self.X, function(self[col]), deg)
return dataArray(np.c_[X, np.poly1d(poly)(X)].T)
def regrid(self, xgrid=None, zgrid=None, wgrid=None, method='nearest', fill_value=0):
"""
Regrid multidimensional data to a regular grid with optional interpolation of .Y values for missing points.
E.g. 2D data (with .X .Z) are checked for missing points to get a regular grid (like image pixels)
and .Y values are interpolated. By default, the unique values in a dimension are used but can be
set by ?grid parameters. For 1D data use *.interpolate*
Parameters
----------
xgrid : array,int, None
New grid in x dimension. If None the unique values in .X are used.
For integer the xgrid with these number of points between [min(X),max(X)] is generated.
zgrid :array,int, None
New grid in z dimension (second dimension). If None the unique values in .Z are used.
For integer the zgrid with these number of points between [min(X),max(X)] is generated.
wgrid :array, int, None
New grid in w dimension (3rd dimension). If None the unique values in .W are used.
For integer the wgrid with these number of points between [min(X),max(X)] is generated.
wgrid<2 ignores wgrid.
method : float,'linear', 'nearest', 'cubic'
Filling value for new points as float or order of interpolation
between existing points.
See `griddata <https://docs.scipy.org/doc/scipy/reference/generated/scipy.interpolate.griddata.html>`_
fill_value
Value used to fill in for requested points outside of the convex
hull of the input points.
See `griddata <https://docs.scipy.org/doc/scipy/reference/generated/scipy.interpolate.griddata.html>`_
Returns
-------
dataArray
Notes
-----
Using unique values in any of the ?grids might create large arrays if close values differ by small values.
Examples
--------
Read sasImage with masked areas.
Reconstruct image by regrid which allows to interpolate the masked areas to get a full image.
In the example calda might also be the result of a fit.
::
import jscatter as js
cal = js.sasimagelib.sasImage(js.examples.datapath+'/calibration.tiff')
# asdataArray removing masked areas
calda = cal.asdataArray(masked=None)
# regrid with image sasImageshape
# masked areas will be filled with nearest values
Y= calda.regrid(calda.qx,calda.qy,0).Y
#reshape to sasImageshape
Y.reshape([calda.sasImageshape[1],calda.sasImageshape[0]])
# show image
fig=js.mpl.contourImage(aa.T,axis='pixel',origin='upper')
The example repeats the
`griddata example <https://docs.scipy.org/doc/scipy/reference/generated/scipy.interpolate.griddata.html>`_
::
import jscatter as js
import numpy as np
import matplotlib.pyplot as pyplot
import matplotlib.tri as tri
def func(x, y):
return x*(1-x)*np.cos(4*np.pi*x) * np.sin(4*np.pi*y**2)**2
# create random points in [0,1]
xz = np.random.rand(1000, 2)
v = func(xz[:,0], xz[:,1])
# create dataArray
data=js.dA(np.stack([xz[:,0], xz[:,1],v],axis=0),XYeYeX=[0, 2, None, None, 1, None])
fig0=js.mpl.scatter3d(data.X,data.Z,data.Y)
fig0.suptitle('original')
newdata=data.regrid(np.r_[0:1:100j],np.r_[0:1:200j],method='cubic')
fig1=js.mpl.surface(newdata.X,newdata.Z,newdata.Y)
fig1.suptitle('cubic')
pyplot.show(block=False)
"""
griddata = scipy.interpolate.griddata
if xgrid is None:
xgrid = np.unique(self.X)
elif isinstance(xgrid, numbers.Integral):
xgrid = np.linspace(self.X.min(), self.X.max(), xgrid)
if zgrid is None:
zgrid = np.unique(self.Z)
elif isinstance(zgrid, numbers.Integral):
zgrid = np.linspace(self.Z.min(), self.Z.max(), zgrid)
if wgrid is None:
try:
# this might fail if there is no .W
wgrid = np.unique(self.W)
except AttributeError:
wgrid = None
elif isinstance(wgrid, numbers.Integral):
if wgrid > 1:
wgrid = np.linspace(self.W.min(), self.W.max(), wgrid)
else:
wgrid = None
if wgrid is None:
# we have only 2D data
xx, zz = np.meshgrid(xgrid, zgrid)
xx = xx.flatten()
zz = zz.flatten()
# ww = None
f = griddata(points=self[[self._ix, self._iz]].T.array, values=self.Y,
xi=(xx, zz), method=method, fill_value=fill_value)
if hasattr(self, '_iey'):
# linear interpolate the errors if present
fe = griddata(points=self[[self._ix, self._iz]].T.array, values=self.eY,
xi=(xx, zz), method='nearest', fill_value=fill_value)
out = dA(np.stack([xx, zz, f, fe]), XYeYeX=[0, 2, 3, None, 1, None])
else:
out = dA(np.stack([xx, zz, f]), XYeYeX=[0, 2, None, None, 1, None])
else:
# we have 3D data
xx, zz, ww = np.meshgrid(xgrid, zgrid, wgrid)
xx = xx.flatten()
zz = zz.flatten()
ww = ww.flatten()
f = griddata(points=self[[self._ix, self._iz, self._iw]].T.array, values=self.Y,
xi=(xx, zz, ww), method=method, fill_value=fill_value)
if hasattr(self, '_iey'):
fe = griddata(points=self[[self._ix, self._iz, self._iw]].T.array, values=self.eY,
xi=(xx, zz, ww), method='nearest')
out = dA(np.stack([xx, zz, ww, f, fe]), XYeYeX=[0, 3, 4, None, 1, None, 2])
else:
out = dA(np.stack([xx, zz, ww, f]), XYeYeX=[0, 3, None, None, 1, None, 2])
out.setattr(self)
return out
# use the fit routines from dataList to be used in dataArray
def fit(self, model, freepar={}, fixpar={}, mapNames={}, xslice=slice(None), condition=None, output=True, **kw):
r"""
Least square fit to model that minimizes :math:`\chi^2` (uses scipy.optimize).
See :py:meth:`dataList.fit`, but only first parameter is used if more than one given.
"""
if self._asdataList is None:
self._asdataList = dataList(self)
free = {}
fix = {}
# use only first value if a list is given
for key, value in freepar.items():
free[key] = (value if isinstance(value, numbers.Number) else value[0])
for key, value in fixpar.items():
fix[key] = (value if isinstance(value, numbers.Number) else value[0])
if 'debug' in kw:
return self._asdataList.fit(model=model, freepar=free, fixpar=fix, mapNames=mapNames,
xslice=xslice, condition=condition, output=output, **kw)
res = self._asdataList.fit(model=model, freepar=free, fixpar=fix, mapNames=mapNames,
xslice=xslice, condition=condition, output=output, **kw)
if res == -1:
return res
# pick fit result
self.lastfit = self._asdataList.lastfit[0]
for attr in self._asdataList.lastfit.__dict__:
if attr[0] != '_':
temp = getattr(self._asdataList.lastfit, attr)
if attr in free:
# is first element in an attributelist
setattr(self.lastfit, attr, temp[0])
setattr(self, attr, temp[0])
elif '_err' in attr and attr[:-4] in free:
setattr(self.lastfit, attr, temp[0])
setattr(self, attr, temp[0])
elif attr in mapNames:
try:
setattr(self.lastfit, attr, temp[0])
except (IndexError, TypeError):
# ith might not b iterable for float
setattr(self.lastfit, attr, temp)
elif attr in fix:
setattr(self.lastfit, attr, fix[attr])
else:
setattr(self.lastfit, attr, temp)
return
def estimateError(self, output=True):
"""
Estimate error
See :py:meth:`dataList.estimateError`
Examples
--------
::
import jscatter as js
import numpy as np
data=js.dA(js.examples.datapath+'/iqt_1hho.dat',index=3)
diffusion=lambda t,wavevector,A,D,b:A*np.exp(-wavevector**2*D*t)+b
data.fit(diffusion ,{'D':0.1,'A':1},{'b':0.},{'t':'X','wavevector':'q'},method='Powell')
data.estimateError()
"""
# use estimate error from previous _asdataList
self._asdataList.estimateError(output=output)
# copy all refined fit attributes from _asdataList to self
self.lastfit = self._asdataList.lastfit[0]
for attr in self._asdataList.lastfit.__dict__:
if attr[0] != '_':
temp = getattr(self._asdataList.lastfit, attr)
if attr in self._asdataList._freepar:
# is first element in an attributelist
setattr(self.lastfit, attr, temp[0])
setattr(self, attr, temp[0])
elif '_err' in attr and attr[:-4] in self._asdataList._freepar:
setattr(self.lastfit, attr, temp[0])
setattr(self, attr, temp[0])
elif attr in self._asdataList._mapNames:
try:
setattr(self.lastfit, attr, temp[0])
except (IndexError, TypeError):
# ith might not b iterable for float
setattr(self.lastfit, attr, temp)
elif attr in self._asdataList._fixpar:
setattr(self.lastfit, attr, self._asdataList._fixpar[attr])
else:
setattr(self.lastfit, attr, temp)
return
def refineFit(self, method='lm', **kw):
"""
Refined fit with starting values from previous fit.
See :py:meth:`dataList.refineFit`
Examples
--------
::
import jscatter as js
import numpy as np
data=js.dA(js.examples.datapath+'/iqt_1hho.dat',index=3)
diffusion=lambda t,wavevector,A,D,b:A*np.exp(-wavevector**2*D*t)+b
data.fit(diffusion ,{'D':0.1,'A':1},{'b':0.},{'t':'X','wavevector':'q'},method='Powell')
data.refineFit()
"""
# use estimate error from previous _asdataList
self._asdataList.refineFit(method=method, **kw)
# copy all refined fit attributes from _asdataList to self
self.lastfit = self._asdataList.lastfit[0]
for attr in self._asdataList.lastfit.__dict__:
if attr[0] != '_':
temp = getattr(self._asdataList.lastfit, attr)
if attr in self._asdataList._freepar:
# is first element in an attributelist
setattr(self.lastfit, attr, temp[0])
setattr(self, attr, temp[0])
elif '_err' in attr and attr[:-4] in self._asdataList._freepar:
setattr(self.lastfit, attr, temp[0])
setattr(self, attr, temp[0])
elif attr in self._asdataList._mapNames:
try:
setattr(self.lastfit, attr, temp[0])
except (IndexError, TypeError):
# ith might not b iterable for float
setattr(self.lastfit, attr, temp)
elif attr in self._asdataList._fixpar:
setattr(self.lastfit, attr, self._asdataList._fixpar[attr])
else:
setattr(self.lastfit, attr, temp)
return
def setLimit(self, *args, **kwargs):
"""
Set upper and lower limits for parameters in least square fit.
See :py:meth:`dataList.setlimit`
"""
if self._asdataList is None:
self._asdataList = dataList(self)
self._asdataList.setLimit(*args, **kwargs)
setlimit = setLimit
def getBayesSampler(self, *args, **kwargs):
"""
Returns Bayes sampler after Bayesian fit.
First do a fit with method='bayes' then the sampler can be retrieved.
Returns
-------
emcee sampler
Examples
--------
See :py:func:`~jscatter.dataarray.dataList.getBayesSampler`
"""
if self._asdataList is None:
self._asdataList = dataList(self)
return self._asdataList.getBayesSampler(*args, **kwargs)
@property
@inheritDocstringFrom(dataList)
def hasLimit(self):
"""
Return existing limits.
See :py:meth:`dataList.has_limit`
"""
return self._asdataList.hasLimit
has_limit = hasLimit
@inheritDocstringFrom(dataList)
def setConstrain(self, *args):
"""
Set constrains for constrained minimization in fit.
Inequality constrains are accounted by an exterior penalty function increasing chi2.
Equality constrains should be incorporated in the model function
to reduce the number of parameters.
Parameters
----------
args : function or lambda function
Function that defines constrains by returning boolean with free and fixed parameters as input.
The constrain function should return True in the accepted region and return False otherwise.
Without function all constrains are removed.
Notes
-----
See dataList
"""
if self._asdataList is None:
self._asdataList = dataList(self)
self._asdataList.setConstrain(*args)
@property
@inheritDocstringFrom(dataList)
def hasConstrain(self):
"""
Return list with defined constrained source code.
"""
return self._asdataList.hasConstrain
@inheritDocstringFrom(dataList)
def modelValues(self, *args, **kwargs):
"""
Calculates modelValues after a fit allowing simulation with changed parameters.
See :py:meth:`dataList.modelValues`
"""
try:
return self._asdataList.modelValues(*args, **kwargs)[0]
except AttributeError:
print('First do a fit!!')
def extract_comm(self, iname=0, deletechars='', replace={}):
"""
Extracts not obvious attributes from comment and adds them to attributes.
The iname_th word is selected as attribute and all numbers are taken.
Parameters
----------
deletechars : string
Chars to delete
replace : dictionary of strings
Strings to replace {',':'.','as':'xx','r':'3.14',...}
iname : integer
Which string to use as attr name; in example 3 for 'wavelength'
Notes
-----
example : w [nm] 632 +- 2,5 wavelength
extract_comm(iname=3,replace={',':'.'})
result .wavelength=[632, 2.5]
"""
if isinstance(self.comment, str):
self.comment = [self.comment]
for line in self.comment:
words = _deletechars(line, deletechars)
for old, new in replace.items():
words = words.replace(old, new)
words = [_w2f(word) for word in words.split()]
number = [word for word in words if isinstance(word, numbers.Number)]
nonumber = [word for word in words if not isinstance(word, numbers.Number)]
if nonumber:
self.setattr({nonumber[iname]: number})
def getfromcomment(self, attrname, convert=None, ignorecase=False):
"""
Extract a non number parameter from comment with attrname in front
If multiple names start with parname first one is used.
Used comment line is deleted from comments.
Parameters
----------
attrname : string without spaces
Name of the parameter in first place
convert : function
Function to convert the remainder of the line to the desired attribut value. E.g. ::
# line "Frequency MHz 3.141 "
.getfromcomment('Frequency',convert=lambda a: float(a.split()[1]))
ignorecase : bool
Ignore attrname character case.
If True the lowercase attrname is used.
Notes
-----
A more complex example with unit conversion ::
f={'GHz':1e9,'MHz':1e6,'KHz':1e3,'Hz':1}
# line "Frequency MHz 3.141"
.getfromcomment('Frequency',convert=lambda a: float(a.split()[1]) * f.get(a.split()[0],1))
"""
if ignorecase:
for i, line in enumerate(self.comment):
words = line.split()
if len(words) > 0 and words[0].lower() == attrname.lower():
val = ' '.join(words[1:])
if convert is not None:
val = convert(val)
setattr(self, attrname.lower(), val)
del self.comment[i]
return
else:
for i, line in enumerate(self.comment):
words = line.split()
if len(words) > 0 and words[0] == attrname:
val = ' '.join(words[1:])
if convert is not None:
val = convert(val)
setattr(self, attrname, val)
del self.comment[i]
return
@property
def attr(self):
"""
Return data specific attributes as sorted list.
Returns
-------
list : string
"""
if hasattr(self, '__dict__'):
attrlist = [k for k in self.__dict__ if (k[0] != '_') and (k not in protectedNames + ['raw_data'])]
return sorted(attrlist)
else:
return []
# noinspection PyBroadException
def showattr(self, maxlength=None, exclude=['comment']):
"""
Show attributes with values as overview.
Parameters
----------
maxlength : int
Truncate string representation after maxlength char
exclude : list of str
List of attr names to exclude from result
"""
for attr in self.attr:
if attr not in exclude:
# print( '%25s = %s' %(attr,str(getattr(self,attr))[:maxlength]))
values = getattr(self, attr)
try:
valstr = shortprint(values.split('\n'))
print('{:>24} = {:}'.format(attr, valstr[0]))
for vstr in valstr[1:]:
print('{:>25} {:}'.format('', vstr))
except:
print('%24s = %s' % (attr, str(values)[:maxlength]))
def resumeAttrTxt(self, names=None, maxlength=None):
"""
Resume attributes in text form.
A list with the first element of each attr is converted to string.
Parameters
----------
names : iterable
Names in attributes to use
maxlength : integer
Max length of string
Returns
-------
string
"""
if names is None:
names = self.attr
ll = []
for name in names:
if name == 'comment' and len(getattr(self, name)) > 0:
# only the first one in short
ll.append(name + '=' + _deletechars(getattr(self, name)[0], ' ,.#*+-_"?§$%&/()=')[:10])
else:
# only first element
par = getattr(self, name)
try:
val = par if isinstance(par, numbers.Number) else par[0]
ll.insert(0, f'{name}={val:.2e}')
except (IndexError, TypeError, ValueError):
ll.insert(0, f'{name}={par}'[:20])
text = ' '.join(ll)
return text[:min(len(text), maxlength)]
def savetxt(self, name, fmt='%8.5e', exclude=[]):
"""
Saves dataArray in ASCII text file (optional gzipped).
- If name extension is '.gz' the file is compressed (gzip).
- For writing format see :py:meth:`.dataList.savetxt`
Parameters
----------
name : string, stringIO
Filename to write to or io.BytesIO.
exclude : list of str, default []
List of attribute names to exclude from being saved.
- `exclude = ['XYeYeX']` excludes
the specific lines to recover columnIndex ("XYeYeX 1 2 3 - - - - ").
- To exclude all attributes `exclude = data.attr`
fmt : string
Format specifier for float.
- passed to numpy.savetext with example for ndarray part:
- A single format (%10.5f), a sequence of formats
or a multi-format string, e.g. 'Iteration %d -- %10.5f', in which
- case `delimiter` is ignored.
"""
if isinstance(exclude, str):
exclude = [exclude]
if hasattr(name, 'writelines'):
# write to stringIO
name.writelines(_maketxt(self, name=None, fmt=fmt, exclude=exclude))
return
if os.path.splitext(name)[-1] == '.gz':
_open = gzip.open
else: # normal file
_open = open
with _open(name, 'wb') as f:
f.writelines(_maketxt(self, name=name, fmt=fmt, exclude=exclude))
return
savetext = savetxt
save = savetxt
def __repr__(self):
attr = self.attr[:6]
try:
attr.remove('comment')
except ValueError:
pass
# noinspection PyBroadException
try:
if isinstance(self.comment, list):
comment = self.comment[:2]
else:
comment = [self.comment]
except:
comment = []
desc = """dataArray->(X,Y,....=\n{0},\ncomment = {1}...,\nattributes = {2} ....,\nshape = {3} """
return desc.format(shortprint(self.array, 49, 3) + '......', [a[:70] for a in comment], attr, np.shape(self))
def concatenate(self, others, axis=1, isort=None, missing=None):
"""
Concatenates the dataArray[s] others to self !NOT IN PLACE!
and add attributes from others according to parameter ``missing`` .
Parameters
----------
others : dataArray, dataList, list of dataArray
Objects to concatenate with same shape as self.
axis : integer, None
Axis along to concatenate. Default concatenates along X.
None flattens the dataArray before merge. See numpy.concatenate.
isort : integer
Sort array along column isort =i
missing : None, 'error', 'drop', 'skip'. 'first', default=None
Determines how to deal with missing attributes.
- Insert none
- Raise AttributeError
- 'drop' attribute value for missing
- 'skip' attribute for all
- Use 'first' one
Returns
-------
dataArray with merged attributes and isorted
Notes
-----
See numpy.concatenate
"""
if not isinstance(others, list):
others = [others]
# new naked dataArray
data = dataArray(np.concatenate([self] + others, axis=axis))
# copy attributes
attribs = {a for one in [self]+others for a in one.attr}
for a in attribs:
if missing is None:
val = [getattr(one, a, None) for one in [self] + others]
elif missing in ['Error', 'error', 'e']:
# raise AttributeError if missing
val = [getattr(one, a) for one in [self] + others]
elif missing in ['skip', 's', 'Skip']:
try:
val = [getattr(one, a) for one in [self] + others]
except AttributeError:
val =[]
elif missing in ['first', 'First', 'f']:
if hasattr(self, a):
val = [getattr(self, a)]
else:
# drop
val = [getattr(one, a) for one in [self] + others if hasattr(one, a)]
# reduce the lists
if a == 'comment':
# single list of comments, removing empty
val = [line for lines in val for line in lines if line]
elif np.all([v == val[0] for v in val]):
# all equal
val = val[0]
setattr(data, a, val)
if isort is not None:
data.isort(col=isort)
return data
def addZeroColumns(self, n=1):
"""
Copy with n new zero columns at the end !!NOT in place!!
Column indices are converted to positive values to preserve assignment.
Parameters
----------
n : int
Number of columns to append
"""
newdA = dataArray(np.vstack((self.array, np.zeros((n, self.shape[1])))))
newdA.setattr(self)
# get Indices
piv = [getattr(self, _ix, None) for _ix in protectedIndicesNames]
# convert negative to positive to preserve assignment
newdA.setColumnIndex([p if p is None else (p if p >= 0 else self.shape[0] + p) for p in piv])
return newdA
def addColumn(self, n=1, values=0):
"""
Copy with new columns at the end populated by values !!NOT in place!!
Column indices (where to find .X, ..Y) are converted to positive values to preserve assignment.
Parameters
----------
n : int
Number of columns to append
values : float, list of float
Values to append in columns as data[-n:]=values
Examples
--------
::
import jscatter as js
i5 = js.dL(js.examples.datapath+'/iqt_1hho.dat')[0]
i55 = i5.addColumn(2,i5[1:])
"""
newdA = self.addZeroColumns(n) # copy self with new columns
newdA[-n:] = values
return newdA
def merge(self, others, axis=1, isort=None, missing=None):
"""
Merges dataArrays to self !!NOT in place!!
and add attributes from others according to parameter ``missing`` .
Parameters
----------
others : dataArray, dataList, list of dataArray
Objects to concatenate with same shape as self.
axis : integer
Axis along to concatenate. Default concatenates along X.
None flattens the dataArray before merge. See numpy.concatenate.
isort : integer
Sort array along column isort =i
missing : None, 'error', 'drop', 'skip'. 'first', default=None
Determines how to deal with missing attributes.
- Insert none
- Raise AttributeError
- 'drop' attribute value for missing
- 'skip' attribute for all
- Use 'first' one
Returns
-------
dataArray with merged attributes and isorted
"""
return self.concatenate(others, axis, isort, missing=missing)
def isort(self, col='X'):
"""
Sort along a column !!in place
Parameters
----------
col : 'X','Y','Z','eX','eY','eZ' or 0,1,2,...
Column to sort along
"""
self[:, :] = self[:, self[col].argsort()]
def where(self, condition):
"""
Copy with lines where condition is fulfilled.
Parameters
----------
condition : function
Function returning bool
Examples
--------
::
data.where(lambda a:a.X>1)
data.where(lambda a:(a.X**2>1) & (a.Y>0.05) )
"""
return self[:, condition(self)]
@inheritDocstringFrom(dataListBase)
def simulate(self, **kwargs):
if self._asdataList is None:
self._asdataList = dataList(self)
return self._asdataList.simulate(**kwargs)[0]
@inheritDocstringFrom(dataListBase)
def makeErrPlot(self, *args, **kwargs):
pass
@inheritDocstringFrom(dataListBase)
def makeNewErrPlot(self, *args, **kwargs):
pass
@inheritDocstringFrom(dataListBase)
def detachErrPlot(self, *args, **kwargs):
pass
@inheritDocstringFrom(dataListBase)
def errPlot(self, *args, **kwargs):
pass
@inheritDocstringFrom(dataListBase)
def savelastErrPlot(self, *args, **kwargs):
pass
@inheritDocstringFrom(dataListBase)
def showlastErrPlot(self, *args, **kwargs):
pass
@inheritDocstringFrom(dataListBase)
def killErrPlot(self, *args, **kwargs):
pass
@inheritDocstringFrom(dataListBase)
def errPlottitle(self, *args, **kwargs):
pass
# dataArray including errPlot functions
# noinspection PyIncorrectDocstring
[docs]
class dataArray(dataArrayBase):
[docs]
@inheritDocstringFrom(dataList)
def makeErrPlot(self, *args, **kwargs):
if self._asdataList is None:
self._asdataList = dataList(self)
if 'fitlinecolor' not in kwargs:
kwargs.update(fitlinecolor=4)
self._asdataList.makeErrPlot(*args, **kwargs)
@property
def errplot(self):
"""
Errplot handle
"""
return self._asdataList._errplot
[docs]
@inheritDocstringFrom(dataList)
def makeNewErrPlot(self, *args, **kwargs):
if self._asdataList is None:
self._asdataList = dataList(self)
if 'fitlinecolor' not in kwargs:
kwargs.update(fitlinecolor=4)
self._asdataList.makeNewErrPlot(*args, **kwargs)
[docs]
@inheritDocstringFrom(dataList)
def detachErrPlot(self, *args, **kwargs):
try:
self._asdataList.detachErrPlot(*args, **kwargs)
except:
pass
[docs]
@inheritDocstringFrom(dataList)
def errPlot(self, *args, **kwargs):
if self._asdataList is None:
self._asdataList = dataList(self)
self._asdataList.errPlot(*args, **kwargs)
[docs]
@inheritDocstringFrom(dataList)
def savelastErrPlot(self, *args, **kwargs):
if self._asdataList is None:
self._asdataList = dataList(self)
self._asdataList.savelastErrPlot(*args, **kwargs)
[docs]
@inheritDocstringFrom(dataList)
def showlastErrPlot(self, *args, **kwargs):
if self._asdataList is None:
print('first do a fit!!')
else:
if 'fitlinecolor' not in kwargs:
kwargs.update(fitlinecolor=4)
self._asdataList.showlastErrPlot(*args, **kwargs)
[docs]
@inheritDocstringFrom(dataList)
def killErrPlot(self, *args, **kwargs):
if self._asdataList is None:
print('first do a fit!!')
else:
self._asdataList.killErrPlot(*args, **kwargs)
[docs]
def errPlotTitle(self, *args, **kwargs):
if self._asdataList is None:
print('first do a fit!!')
else:
self._asdataList.errPlotTitle(*args, **kwargs)
[docs]
@inheritDocstringFrom(dataList)
def getChi2Trace(self):
if self._asdataList is None:
self._asdataList = dataList(self)
return self._asdataList.getChi2Trace()
# end dataArray main definitions ------------------------
# noinspection PyIncorrectDocstring
def zeros(*args, **kwargs):
"""
dataArray filled with zeros.
Parameters
----------
shape : integer or tuple of integer
Shape of the new array, e.g., (2, 3) or 2.
Returns
-------
dataArray
Examples
--------
::
js.zeros((3,20))
"""
zero = np.zeros(*args, **kwargs)
return dataArray(zero)
# noinspection PyIncorrectDocstring
def ones(*args, **kwargs):
"""
dataArray filled with ones.
Parameters
----------
shape : integer or tuple of integer
Shape of the new array, e.g., (2, 3) or 2.
Returns
-------
dataArray
Examples
--------
::
js.ones((3,20))
"""
one = np.ones(*args, **kwargs)
return dataArray(one)
# noinspection PyIncorrectDocstring
def fromFunction(function, X, *args, **kwargs):
"""
Evaluation of Y=function(X) for all X and returns a dataArray with X,Y
Parameters
----------
function or lambda
function to evaluate with first argument as X[i]
result is flattened (to be one dimensional)
X : array N x M
X array
function is evaluated along first dimension (N)
e.g np.linspace or np.logspace
*args,**kwargs : arguments passed to function
Returns
-------
dataArray with N x ndim(X)+ndim(function(X))
Examples
--------
::
import jscatter as js
result=js.fromFunction(lambda x,n:[1,x,x**(2*n),x**(3*n)],np.linspace(1,50),2)
#
X=(np.linspace(0,30).repeat(3).reshape(-1,3)*np.r_[1,2,3])
result=js.fromFunction(lambda x:[1,x[0],x[1]**2,x[2]**3],X)
#
ff=lambda x,n,m:[1,x[0],x[1]**(2*n),x[2]**(3*m)]
X=(np.linspace(0,30).repeat(3).reshape(-1,3)*np.r_[1,2,3])
result1=js.fromFunction(ff,X,3,2)
result2=js.fromFunction(ff,X,m=3,n=2)
result1.showattr()
result2.showattr()
"""
res = [np.r_[x, np.asarray(function(x, *args, **kwargs)).flatten()] for x in X]
result = dataArray(np.asarray(res).T)
result.setColumnIndex(0, len(np.atleast_1d(X[0])))
result.args = args
for key in kwargs:
setattr(result, key, kwargs[key])
if hasattr(function, 'func_name'):
result.function = str(function.func_name)
elif hasattr(function, '__name__'):
result.function = str(function.__name__)
return result
# create two shortcuts
dL = dataList
dA = dataArray
# this generates the same interface for grace as in mplot
# unfortunately both use the same names with small char at beginning for different objects
# using big letters solves this
from . import graceplot
if graceplot.GraceIsInstalled:
from .graceplot import GracePlot as openplot
else:
try:
from . import mpl
mpl.gf = 20
openplot = mpl.mplot
except ImportError:
# use the base classes with errPlot only as dummy functions
dataList = dataListBase
dataArray = dataArrayBase
print('No plot interface found')
