* BinnedSpikeTrain optional CSC format

* BinnedSpikeTrain.binarize() 'copy' arg is back

…#404)

* accelerated pmat_neighbors function
* cluster_matrix_entries with chunking

…386)

Co-authored-by: Michael Denker <[email protected]>

* Changed to pytest and pytest-cov for running tests with coverage

* Changed mpiexec test and added coveralls

Co-authored-by: Cristiano Köhler <[email protected]>

Co-authored-by: Cristiano Köhler <[email protected]>

* Fix complexity annotation dtype

* Adapt test name to new object name

* Combine redundant type casts

Co-authored-by: kleinjohann <[email protected]>

* bug fixed dealing with units

* fixed same unit error in trial shuffling

* added Bielefeld fim.so and adapted filtering and added window param to fpgrowth

* removed max_occ test

* further unit stuff

* debugging for new fim version

* enabled multithreading in fpgrowth

* less verbose in spade

* set equal number of bins in bin shuffling wrt spade

* added tolerance to binning in spade everywhere

* Added accelerated FIM algorithm sources by Florian Porrmann.

* Enh/accelerated spade build (#82)

* Added cibuildwheel action

* Added Python requirements to wheel build

* Build only on 64bit machines, otherwise overflow

* Removed Windows for testing, as vc is not available

* Removed MacOS for testing, as -fopenmp is not available

* Removed pp- (pypy) builds since they lack C.

* Fixed removing pp- (pypy) builds since they lack C.

* Put Macos back in.

* Windows Hack

* Remove vcpython alltogether, ignore 2.7 Python

* Removed extra compile option, which breaks on Windows

* Removed more extra compile options, which breaks on Windows

* Try C++ instead of Gnu++.

* Try C++ instead of Gnu++ Windows style argument.

* Remove linux build while testing windows.

* Remove libraries.

* Differentiate Windows and Linux.

* Added missing import.

* Last mile: MacOS

* Remove openMP lib

* Remove openMP lib

* Add openMP lib

* More brew installs

* Mac is called mac on github

* Make sure C is reinstalled.

* Multilib

* Next try, new options

* Ignore warning about void type

* Update newsest fim package

* Revert "Ignore warning about void type"

This reverts commit 3ff6b62

* Revert to prior fim, new compiler argument.

* Revert "Update newsest fim package"

This reverts commit f321f77

* Definitely, gnu++17, but new try.

* Try C++

* Warning message

* llvm maybe?

* Added apple in source

* Small fixes for MacOS, but not comprehensive

* Limit to Windows and Linux for now

* Remove MacOS entry

* Fix fix from mindlessness

* Testrun

* Trying to include fim.so, despite its renaming by wheels

* Added newest version of original module

* Reverted previous breaking change commited by accident.

* Reverted package name from testing.

* Test focal as CI build

* Test bionic as CI build

* Understand installation issue on CI -- is importing elephant importing the installed version?

* Spelling error only

* Try to make sure travis loads the installed elephant, not the cwd.

* One step further -- which version will nosetests use?

* Switch to pytest as of PR #413

* Added authors of new FIM module and reference in new docs.

* Added authors of new FIM module and reference in new docs.

* Small text clarifications.

* Test if entry for fim.so/pyd in MANIFEST is now redundant.

* Update elephant/spade.py

Co-authored-by: Alexander Kleinjohann <[email protected]>

* Update elephant/spade.py

Co-authored-by: Alexander Kleinjohann <[email protected]>

* Added SPADE tutorial

* Prevent wheel building on every push, and limit scipy version workaround

* Pushed tutorial, removed file added in error

* New attempt to make mybinder install requirements.

* New attempt, dropping viziphant.

* Avoid recursive elephant installation by viziphant in postBuild

* Removed unit test that is fragile as it depends on the implementation of surrogate methods

* Add viziphant to RTD environment

* Typo in tutorial

* Add viziphant to travis doc tests

Co-authored-by: pbouss <[email protected]>
Co-authored-by: stellalessandra <[email protected]>
Co-authored-by: Alessandra Stella <[email protected]>
Co-authored-by: Alexander Kleinjohann <[email protected]>

* First test to move viziphant to the environment.yml

* Back to start

* Update Python

* Remove mpi4py

* Remove pip

* Add mpi4py, viziphant

* Hopefully working version

* Remove viziphant again

* Added deprecated fake_neo objects from neo to allow neo 0.10.0+.

* Fix imports to Python 3.

* Another attempt to fix imports to Python 3.

* Attempt relative import.

* Cross-test with UE analysis test suite.

* Next attempt....

* Remove viziphant to prevent elephant install and new attempt.

* Remove viziphant to prevent elephant install and new attempt.

* Try with a new version number

* Stop trying to fix neo 0.10.0, must be separate.

* Reintroduce viziphant in conda environment.

* Introduce concurrent requirement for neo in environment and use pip install .

* Final beautification measures

* bug fixed dealing with units

* fixed same unit error in trial shuffling

* added Bielefeld fim.so and adapted filtering and added window param to fpgrowth

* removed max_occ test

* further unit stuff

* debugging for new fim version

* enabled multithreading in fpgrowth

* less verbose in spade

* added correction factor to rate estimation

* only spiketrain with more than one spikes are corrected

* started rearranging the spike train generation: equilibrium rp. and class structure

* enable option equilibrium /ordinary

* annotations

* Fixed bugs for zero rate

* added test of boundary correction

* PEP8

* revert wrongly merged master

* separate from firing rate correction

* simplified recovered rate test

* finished restructuring in classes

* Added lognormal and inv-gaussian processes

* Updated Tests, added Log-Normal, Inverse Gaussian

* PEP8

* added test for invgauss and lognorm

* PEP8

* updated hardcoded SPADE tests

* update hardcoded UE test

* update hardcoded UE test

* update hard-coded spike train synchrony test

* PEP8

* combined Poisson and PPD into one function

* simplified compound poisson process, single interaction process

* deleted unused hidden functions

* added CV test

* added test for CV and first spike & added more comments

* PEP8

* simplified bib-tex entry

* started working in Robins review

* fix breaked test for Joint-ISI dithering

* continued working on Robins review

* fixed bug - in getting first spike function

* finished working on Robins review, improved first-spike function

* changed to return None for option in init

Co-authored-by: stellalessandra <[email protected]>
Co-authored-by: Alessandra Stella <[email protected]>

This is a fix related to issue #424: 
t_start was lost while transposing lfp, because t_start was not set in the redefinition of the new neo.analogSignal. Fixed by transposing the neo.Analogsignal directly. @ojoenlanuca , @rgutzen

* Replaced custom test build with unittest.main() in 
 - test_cell_assembly_detection.py
 - test_cubic.py
 - test_gpfa.py
 - test_icsd.py 
 - test_spade.py
 - test_spike_train_surrogates.py

* Fixed deprecation warnings in test_cell_assembly_detection.py:
 - numpy.testing.utils -> numpy.testing
 - data -> binned_spiketrain
 - maxlag -> max_lag
 - min_occ -> min_occurrences

* Fixed deprecation warning (ximax -> max_iteractions) in test_cubic.py

* Fixed typo in test_cubic.py

* removed redundant blank lines at the end of code blocks

The plots generated in the Granger causality tutorial had incorrect axis ranges. This was fixed by setting x- and y-axis bounds in both plots.

- Fixed the figures in granger_causality.ipynb tutorial

* Updated zscore to change the units to dimensionless when performing in-place operations, and also to return the original object instead of a new AnalogSignal object pointing to the original array.
* Changed to raise ValueError when the in-place operation is not valid.
* New AnalogSignal object is created using Neo function.
* Updated unit tests to check for same/different objects when doing in-place operation in zscore.
* Changed test_zscore_single_inplace_int as the function now raises ValueError.

Co-authored-by: Cristiano Köhler <[email protected]>

…ain (#425)

* Update README.md

added zenodo batch with DOI

* Update README.md

Co-authored-by: Michael Denker <[email protected]>

Co-authored-by: Michael Denker <[email protected]>

* fix for bug related to pytest 7.0.0, fixed version to 6.2.5 thus fixing errors when running pytest with  `--import-mode=importlib`

* Update .travis.yml

Co-authored-by: Michael Denker <[email protected]>

New Feature: multitaper psd estimate. Tests include a comparison against nitime multitaper.

This reverts commit d0d3920 to fix an issue where the authors of #417 where incorrectly identified. #417 will be re-merged with correct author attributions.

New Feature: multitaper psd estimate. Tests include a comparison against nitime multitaper.

Co-authored-by: ackurth <[email protected]>
Co-authored-by: Regimantas Jurkus <[email protected]>
Co-authored-by: Aitor Morales-Gregorio <[email protected]>
Co-authored-by: kleinjohann <[email protected]>
Co-authored-by: pbouss <[email protected]>
Co-authored-by: Danylo Ulianych <[email protected]>
Co-authored-by: Alexander Kleinjohann <[email protected]>
Co-authored-by: Andrew Davison <[email protected]>
Co-authored-by: Anno Christopher Kurth <[email protected]>

added `discretise_spiketimes(spiketrains, sampling_rate)` to conversion

…ant (#447)

* Handle SpikeTrainList as input for BinnedSpikeTrain

* Add corresponding regression test

* PEP 8; avoid universal try-except clause

* Fixed SpikeTrainList issue in spike train synchrony class

* Fixed SpikeTrainList issue in SPADE

* Pep8 issue

* This removes support of RrcordingChannelGroup in CSD methods.

* Changed requirements to Neo >=0.10.0

* Fix spiketrain lists in instantaneous rate.

* First fix for unittests missing fake_neo objects

* Fixed all GetAllObjsTestCase unit tests

* Started work on Events test case

* Fixed Events test case

* Fixed Epochs and Spiketrain tests

* Starting to fix ExtractNeoAttrs

* 7 more failing tests for ExtractNeoAttrs

* All neo_tools tests fixed for Neo 0.10

* Enabled dataset generators from neo 0.10.0

* This PR adds SpikeTrainList as output of `get_all_spiketrains()``

* Use SpikeTrainList, work around SpikeTrainList.extend() bug.

* Fix pandas_bridge tests to work without fake_neo()

* PEP8 Line length corrections

* More PEP8 finishes

* Yet more PEP8 finishes

* Last PEP8 finish

* Update elephant/neo_tools.py

* added CI.yml containing workflows

* added badge for CI tests to README.md

* restructured requirements, environment.yml refactored to environment-docs.yml and environment-tests.yml.

* adapted runners according to new new .yml files

* update requirements in environment-*.yml to neo>=0.10.0

* initial draft for .zenodo.json, collected some authors for release 0.11.0 from commits since release 0.10.0

* added keywords

* changed license id

* corrected creator entry

* added authors

* corrected entry

* added entry

* Apply suggestions from code review

added keyword, corrected affiliation for INM

Co-authored-by: Michael Denker <[email protected]>

* corrected INM affiliation in authors.rst and for all entries in .zenodo.json

* added grants by id

* added grant agreement no. 604102 (Human Brain Project, HBP) to .zenodo.json and acknowledgments.rst.

* added "title" field

* added orcid Pa. Da.

* reformat

Co-authored-by: Michael Denker <[email protected]>

* added Zenodo DOIs to docs, "how to cite elephant"

* added links for DOIs

* fix: added spade notebook to toctree

* added description for DOIs

* deleted extended description of different DOIs

* added "doi:" to DOI links

* typos

* reformulate citation, fixed typo

…414)

Co-authored-by: stellalessandra <[email protected]>

* added CI.yml containing workflows

* added badge for CI tests to README.md

* restructured requirements, environment.yml refactored to environment-docs.yml and environment-tests.yml.

* adapted runners according to new new .yml files

* re-add environment.yml for Travis, to be removed together with travis.yml

* update requirements in environment-*.yml to neo>=0.10.0

* updated .yml files to python version 3.7

* removed python 3.6 from docs 'install.rst', added python version requirement to `setup.py`

* rename travis conda 3.6 extras,opencl runner

* updated build_wheels workflow to exclude python 3.6

* removed redundant line

* changed structure of installation for docs runner

* added extras to docs runner

* added trigger for github actions CI: run tests after merged PR

* added classifiers to setup.py (python >=3.7), removed classifiers for python 2

* add python 3.10

* update cibuildwheel version to 2.3.1

…5.0 (#441)

* updated UE tutorial and requirements to use nixio 1.5.0 (#86)

* updated links for datasets used in tutorials asset and UE

* use 'read only' to assure backwards compatibility

* reset outputs of notebook

* remove permalinks from tutorials asset and UE

* hotfix fixed jinja2 version to 3.0.3

* changed URLs to elephant-data master

* added function and environment variable to specifiy elephant-data download URLs

* fixed errors

* changed ASSET tutorial to work with nixio==1.5.0b3

* added docstrings

* use return of download_elephant_data in tutorials

* rename variable, update docstrings

* fixed return of download_elephant_data

* added version specific URL (depending on elephant version)

* added in 'v' for versioning , e.g. v0.10.0

* changed check to see if version is available

* Enh/data utils (#92)

* added handling for SSL verification error
* update ASSET and UE tutorials

* Enh/data utils (#93)

* allow download without verified certificate

* rename data_utils to datasets, and download_elephant_data to download_datasets

* fixed UE tutorial

* fixed pep8

* removed download.py

* fixed deprecation warnings, replaced homogeneous_poisson_process with StationaryPoissonProcess

* added import for scipy.signal

* fixed size of output for instantaneous_rate, edited unittests accordingly

* test to check length of outputs

* added unittest to check consistency of rate times (should be multiple of sampling period), to get consistent ouput sizes, cutting of wings is done by fft convolve for center_kernel=True

* fixed pep8, added comments, restructure

* added include endpoint option

* added notes to docstrings, reformat docstrings and code, removed endpoint option

* fix docstrings

* restructured unit tests, added regression tests

* fixed pep8

* fixed indentation

* fixed wrong return

* update copyright year to 2022

* fix PyPI description display and updated copyright year in `LICENSE.txt`

* Bumped version number

* clear outputs of gpfa notebook

* fix pep8

* fix further pep8

* fixed CodeFactor issues

* fixed further CodeFactor issues

* removed maintainers_guide.rst

* remove travis.yml

* added release notes for release 0.11.0

* Update spike_train_correlation.py

Improvement of Spike Time Tiling Coefficient (STTC) calculation by utilizing numpy function.

Details:
- use numpy.where to find overlapping tiles in run_T
- some comments elaborated/changed
- changed name of time_A to time_max for maximum time covered by tiles

This enhancement was inspired by Thierry Nieus

* removed compilation of fim for darwin

* compile only on Linux Windows

* added example to docstring, removed reference to annotations
* fixed doctest for conversion

* removed compilation of fim for darwin

* changed precision in asset to double, added 2 github action runners for macOS

* removed viziphant from environment-tests.yml

* Update setup.py

Co-authored-by: Michael Denker <[email protected]>

* bumped version number to 0.11.1

* added doi for elephant 0.11.0 to citation

* fixed typo in version

* added initial draft of release note to release_notes.rst

* fixed link to all versions of elephant

* updated release notes

* edited section to cite specific elephant version

* update title in .zenodo.json

* added a test that compares the victor_purpura_distance results to the results of the oroginal matlab implementation

Co-authored-by: Oliver Kloß <[email protected]>

Co-authored-by: Cristiano Köhler <[email protected]>

…he environment (#485)

* added coverage to MPI runner, removed MPI from pip added function get_opencl_capability() to utils, to detect openCL devices

* updated docstring, changed return logic

* added unittest, regression Issue #481

Co-authored-by: Cristiano Köhler <[email protected]>