signal.hilbert incorrect axis handling for 2d input (Trac #1093)

[scipy-gitbot]

Original ticket http://projects.scipy.org/scipy/ticket/1093 on 2010-01-17 by @josef-pkt, assigned to unknown.

signal.hilbert returns incorrect results if the input matrix is 2d.
reported by Ariel Rokim,
see thread http://mail.scipy.org/pipermail/scipy-dev/2010-January/013686.html

The problem is that fft is done with axis=0, ifft is done with default axis -1

Proposed fix and enhancement:

• add correct axis statement to all function calls
• make axis a keyword argument
• set default axis=-1 to be consistent with usage for fft, this changes the API compared to the description in current docstring. However, since 2d is currently not working correctly it wouldn't break any existing code, since it would be incorrect already.

script with rewritten function and test cases by Ariel are attached

[scipy-gitbot]

Attachment added by @josef-pkt on 2010-01-17: try_signalhilbert.py

[scipy-gitbot]

@josef-pkt wrote on 2010-01-17

Note: test in script and docstring for changed function needs some cosmetic changes and adjustments.

[scipy-gitbot]

@josef-pkt wrote on 2010-01-18

committed in 8348983

not sure whether N argument makes sense but added regression tests for it

I keep this open for now, in case someone wants to review and comment e.g. on axis change

[scipy-gitbot]

@josef-pkt wrote on 2010-02-03

no comment, I'm closing it