Age | Commit message (Collapse) | Author |
|
Conflicts:
pyspike/__init__.py
resolved
|
|
new function import_spike_trains_from_time_series that loads spike trains
from time series.
|
|
|
|
treatment of empty lines was incorrect. now empty spike trains are created
from empty lines in the txt file if parameter ignore_empty_lines=False is
given.
|
|
save_spike_trains_to_txt allows to save spike train data into txt files which
can then be loaded via load_spike_trains_from_txt again.
|
|
with the new interface, the previous functions for computing multivariate
profiles and distances are obsolete. This is now noted in the docs.
|
|
fixed ugly bugs in code for computing multi-variate spike sync profile and
multi-variate spike sync value.
|
|
Similar to the isi and spike distance functions, also the spike sync functions
now support the new generic interface.
|
|
spike_profile and spike_distance now have a generic interface that allows
to compute bi-variate and multi-variate results with the same function.
|
|
isi profile and distance functionc an now compute bi-variate and multi-variate
results. Therefore, it can be called with different "overloads".
|
|
Spike trains with single spikes now only get auxiliary spikes at the edges
for the SPIKE distance instead of real spikes before.
|
|
|
|
fixed bug within new edge correction (auxiliary spike was ignored in some
cases)
added regression test with 10000 random spike train sets
|
|
|
|
directionality development happens in separate branch.
|
|
Eero's improved edge correction now also implemented in the python backend.
|
|
|
|
Improvement following Eeros suggestions to use auxiliary spike at the edges
consistently with the corresponding corrected ISI intervals.
|
|
Signed-off-by: Igor Gnatenko <i.gnatenko.brain@gmail.com>
|
|
Signed-off-by: Igor Gnatenko <i.gnatenko.brain@gmail.com>
|
|
Signed-off-by: Igor Gnatenko <i.gnatenko.brain@gmail.com>
|
|
|
|
|
|
|
|
test case and fix for Issue #14. Spike-Sync function now correctly deal with
empty intervals as well.
|
|
add test case and fix for Issue #15
closes #15
|
|
+ spike delay asymmetry profile
+ spike delay dual profile
|
|
added first version of spike delay asymmetry functions. still incomplete and
untested.
|
|
fixes #11
|
|
|
|
|
|
|
|
Dividing the indices by 2 gives a float which cannot be used for slicing
the matrix. Changed the type of the indices to integer to ensure that
matrix slicing does not return error.
|
|
Users can now disable the warning messages produced when the cython backend is
not available by writing
spk.disable_backend_warning = True
in the beginning
|
|
|
|
This allows to use SpikeTrain objects to be used in many applications as if
they were arrays with spike times.
|
|
Conflicts:
setup.py
|
|
|
|
pwc and pwl function object return the average of the left and right limit
as function value at the exact spike times.
|
|
|
|
Added __call__ operators to PieceWiseConst and PieceWiseLin class for
obtaining function values at certain points in time.
|
|
|
|
|
|
|
|
|
|
Additional cython implementation for overall spike sync values. It is not
necessary to compute the profile anymore if only the spike sync value is
required. 3x performance gain.
|
|
isi- and spike distances over complete intervals are now computed without
obtaining the profile first. This gives more than x2 performance improvements.
|
|
|
|
|
|
dont compute the average profile in the function spike_sync_multi, but rather
compute the overall average distance directly
|