TransylvanianInstituteOfNeuroscience
commented
7 months ago Hello,
The superlet transform, as it is presented in the original papers, is defined to compute power, not a complex representation. However, the old version was computing the power a bit weirdly and was outputting a complex representation that the user then had to extract the power values from. Luckily, due to some interactions with a group that wanted to use superlets for radar applications, and needed a complex representation that spans the negative frequencies as well, I developed a version of the superlet transform that can take real or complex input and output a complex representation, correctly computing the complex geometric combination of the individual wavelet responses. It is implemented in MATLAB, for now, in this file Superlets/matlab-pure/nfaslt.m at main · TransylvanianInstituteOfNeuroscience/Superlets (github.com) https://github.com/TransylvanianInstituteOfNeuroscience/Superlets/blob/main/matlab-pure/nfaslt.m. If I understood correctly, this is exactly what you need. I will work with one of the python wizards in our lab this week to get out a python version of that file sometime this week.
Hope this helps, Harald
On Sat, 9 Dec 2023 at 02:03, VikaMan @.***> wrote:
Dear developers,
Thank you for your work! In my lab we're using superlet, and I also recently tested the last version of it, "superlets" function from superlet.py. I observed that this new version works a bit better with my data, however, its output is power, as far as I understand. Besides the power, for some analyses I also need the phase of oscillations, which I can get only from a complex output, but not power. So I wanted to ask, do you think it's possible to modify your superlets function to get complex values as its output? I tried to do that by myself, but I realized that geometric mean estimation used in this function might work incorrectly with negative values, i.e. the row 207 of superlet.py. Thank you in advance, I would be very grateful for any response!
— Reply to this email directly, view it on GitHub https://github.com/TransylvanianInstituteOfNeuroscience/Superlets/issues/3, or unsubscribe https://github.com/notifications/unsubscribe-auth/AR3QLFHSAPVBE4X3YJX4CGLYIOTCDAVCNFSM6AAAAABANK5J6CVHI2DSMVQWIX3LMV43ASLTON2WKOZSGAZTGNBUHEZDQNY . You are receiving this because you are subscribed to this thread.Message ID: @.***>
-- Harald Bârzan, PhD Transylvanian Institute of Neuroscience (TINS), Department of Experimental and Theoretical Neuroscience, Str. Ploiesti 33 https://maps.google.com/?q=Str.+Ploiesti+33&entry=gmail&source=g, 400157 Cluj-Napoca, Romania Email: @.***
vikamanyukhina
perevo
HaraldBarzan
Dear developers,
Thank you for your work! In my lab we're using superlet, and I also recently tested the last version of it, "superlets" function from superlet.py. I observed that this new version works a bit better with my data, however, its output is power, as far as I understand. Besides the power, for some analyses I also need the phase of oscillations, which I can get only from a complex output, but not power. So I wanted to ask, do you think it's possible to modify your superlets function to get complex values as its output? I tried to do that by myself, but I realized that geometric mean estimation used in this function might work incorrectly with negative values, i.e. the row 207 of superlet.py. Thank you in advance, I would be very grateful for any response!