Back to the main page.

Bug 1344 - Problem in ft_denoise_pca, returned 'time' field might be wrong

Reported 2012-02-24 15:14:00 +0100
Modified 2013-01-17 10:44:16 +0100
Product: FieldTrip
Component: core
Version: unspecified
Hardware: PC
Operating System: Windows
Importance: P3 normal
Assigned to: Roemer van der Meij
Depends on:
See also:

Panagiotis Tsiatsis - 2012-02-24 15:14:42 +0100

In the data structure that is returned by ft_denoise_pca, it seems that the "time" field is not being retained from the input argument structure; it rather seems to be reset to start from 0.

Jan-Mathijs Schoffelen - 2012-02-26 14:51:36 +0100

Hi Panagiotis, I cannot confirm this one. Could you upload a very short (i.e. <10 trial) data structure along with the configuration structure, so that I can try and reproduce this?

Panagiotis Tsiatsis - 2012-02-26 16:35:51 +0100

Created attachment 232 Test Dataset plus Testing Script - run as it is

Panagiotis Tsiatsis - 2012-02-26 16:36:05 +0100

Hi Jan-Mathijs, I guess that submitting two related bugs at once at the end of a working day is not a good idea, due to dependencies and mental fatigue. So all apologies, I was wrong in saying that the time structure is reset to zero, this is not true. There is still a problem though as the time structure seems to be altered somehow when compared to the original timings. In my case there is a shift of 0.7 ms - it is not that I am so picky but this would not allow me to append/merge data later on cause the timings would not agree. I attach you a small file of 10 trials as requested and a script that should show you were the problem is. Note that the same code is used (as you will in the code) for ft_denoise_synthetic and (also has been tested with ICA in the past) and this problem does not appear in those cases, so unless I am doing something stupid, it must be a ft_denoise_pca specific problem. IN the code that I am sending you, you can set the flag DENOISE_METHOD to either DEF_DENOISE_SYNTHETIC or DEF_DENOISE_PCA to test both ft_denoise_pca and ft_denoise_pca with the same code. Thanks again!

Jan-Mathijs Schoffelen - 2012-02-29 13:58:42 +0100

reassigned to Roemer, who will have a look at it.

Roemer van der Meij - 2012-02-29 17:22:18 +0100

Hi Panagiotis, Thanks for the report! It should be fixed now. Both denoise_PCA and and synthetic should give the same time-axis now. The changes should be on the FTP server within the next few days. (we worked on many bugs today and we still need pass it through testing on the center here). Could you confirm whether it is fixed in a few days? Regards, Roemer

Roemer van der Meij - 2013-01-17 10:44:16 +0100

bug closing time