Back to the main page.
Bug 561 - not able to plot (topo/singleTFR nonlinear (time/freq) axis
| Status | CLOSED WONTFIX |
| Reported | 2011-04-06 11:51:00 +0200 |
| Modified | 2011-09-14 14:33:48 +0200 |
| Product: | FieldTrip |
| Component: | core |
| Version: | unspecified |
| Hardware: | PC |
| Operating System: | Windows |
| Importance: | P1 normal |
| Assigned to: | Stephen Whitmarsh |
| URL: | |
| Tags: | |
| Depends on: | |
| Blocks: | |
| See also: |
Stephen Whitmarsh - 2011-04-06 11:51:18 +0200
> Hi John, > > Well, first of all your axes *are* not linear. You can clearly see it if > you plot e.g. plot(sub03freq_pitch{1}.time) or > plot(sub03freq_pitch{2}.freq). If you replace your time/freq axis with > something linear (e.g. [1:66], or 1:14 respectively - multiply with > samplerate etc. to get a true one) it does work. I will bring it to the > development group later this week since i wonder why it shouldn't work > with nonlinear axis. > > Perhaps the way you use the wavelets is not optimal. Given the timewindow > you have (1.3 seconds) there are only certain frequencies you can estimate > exactly. Now these are not integers. Perhaps some padding will help (to > 1.5 seconds e.g.). There are also plenty of nan's because of the > timewindow you use. see attachement. > > Considering the time axis - i dont think your samplerate permits the small > timesteps and therefor runs into rounding issues making it nonlinear. That > might be your main problem. > > hope this helps, > Stephen > > script used to make figure: > > sub03freq_pitch{1}.time = [1:1:66]; > sub03freq_pitch{1}.freq = [1:1:17]; > ft_singleplotTFR([],sub03freq_pitch{1}); > > > > > >> Hi Stephen, >> >> I apologize for not getting back to you sooner, and I very much appreciate >> the offer to help. Below, are links to data before and after I ran it >> through the code in the .m file that is attached. The ft_mulitplottfr >> works, but gives a warning about axes not being linear. The singleplottfr >> will not plot anything and warns about masknans. >> >> I would appreciate any insight you could provide. >> >> Before ft_freqanalysis: http://dl.dropbox.com/u/10557416/sub03.mat >> After ft_freqanalysis: http://dl.dropbox.com/u/10557416/sub03_freq_col.mat >> >> -John Moses >> >> On Mar 30, 2011, at 3:08 AM, Stephen Whitmarsh wrote: >> >>> Dear John, >>> Good to hear from you! >>> Could you either send me either the preprocessed or frequency data, as >>> well as the script you use to plot? So that I can reproduce the >>> bug/error. >>> >>> Best, >>> Stephen >>> >>> From: John Christopher Moses [mailto:jcmoses83@gmail.com] >>> Sent: Wednesday, March 30, 2011 8:07 AM >>> To: Stephen Whitmarsh >>> Cc: Raquel Bibi >>> Subject: Re: [FieldTrip Toolkit] Thanks and documentation >>> >>> Hi Stephen, >>> >>> I hope all is going well. I was wondering if you could help me out a >>> little bit. Raquel and I are working on the frequency analysis and we >>> are getting an error when we try to plot it. The error reads "Warning: >>> (one of the) axis are not evenly spaced -> nans cannot be masked out -> >>> cfg.masknans is set to 'no';" The code I used to do the wavelets is >>> below. If you could be of any help I would greatly appreciate it. >>> >>> -John Moses >>> >>> current = what('MANUAL_REJ/'); >>> current = current.mat; >>> for i = 1:length(current) >>> currentv(i, 1) = strrep(current(i), '.mat', ''); >>> end >>> >>> for s=1:length(current) >>> clearvars -except current s currentv >>> >>> load(['MANUAL_REJ/', current{s}]); >>> >>> % compute averages for pitch and timbre >>> for istim = 1:3 >>> cfg=[]; >>> cfg.lpfilter = 'yes'; >>> cfg.lpfreq = 40; >>> [data] = ft_preprocessing(cfg,data); >>> >>> cfg=[]; >>> cfg.keeptrials = 'yes'; >>> cfg.output = 'pow'; >>> cfg.foi = 8:2:40; >>> cfg.toi = -0.3:0.02:1.0; >>> cfg.method = 'wavelet'; >>> >>> cfg.trials = find((data.trialinfo(:,3) == 1 | >>> data.trialinfo(:,3) == 2 ) & data.trialinfo(:,4) == istim); >>> eval([currentv{s} 'freq_pitch{istim} = ft_freqanalysis(cfg, >>> data)']); >>> >>> cfg.trials = find((data.trialinfo(:,3) == 3 | >>> data.trialinfo(:,3) == 4 ) & data.trialinfo(:,4) == istim); >>> eval([currentv{s} 'freq_timbre{istim} = ft_freqanalysis(cfg, >>> data)']); >>> >>> end >>> >>> save(['FREQ/', currentv{s} '_freq_col'],[currentv{s} 'freq_*']); >>> >>> end >>>
Stephen Whitmarsh - 2011-04-08 18:24:51 +0200
You knew something more about that right?
Jan-Mathijs Schoffelen - 2011-06-22 16:03:31 +0200
I don't know anymore what this is about. @ Stephen, is this still a 'bug' to be dealt with?