zackeskin / pycausality Goto Github PK

Currently PyCausality requires a number of dependencies (e.g. Nose, SciPy) which are in fact only used in testing. It is not uncommon to provide only the functional code when building for PyPi deployment, so consider removing testing-related code if this is not required for core functionality. The problem becomes clear when using PyCausality in non-local environments such as AWS Lambda or mobile, where environment management and storage issues become more difficult.

(Similarly, Matplotlib is required only for one method, so consider separating this out.)

There was an error reported when I run "TE.nonlinear_TE(pdf_estimator = 'kernel', n_shuffles=100)".

self.add.results({'TE_XY': np.array(TEs)[:,0],

IndexError: too many indices for array

I'm trying to reproduce Example 3 in Schreiber 2000's seminal article using PyCausality here.

I wonder if it is possible at all to reproduce such results with PyCausality and get the same TE numbers to reproduce the figure in question?

When running TE.nonlinearTE() or TE.linearTE(), if one of these has been called already in the TE object scope, then the add_result() method will fail.

This has been observed using windows but, from memory, is likely also to happen even without applying windowing.

For now, creating two objects (e.g. l_TE and nl_TE) is a simple workaround.