Crop-like option for TFR with `fmin` and `fmax`

I’ve found out during classes where I teach mne that .crop-like operations are usually much more intuitive for students than having to dive into the data array and index it. With tfr they usually want to, for example selecte a time range and frequency range and then average (or not) and use in further computations/visualisations. If you agree it would be useful, there are two options:

adding fmin and fmax to TFR.crop()

This would allow for example for

tfr.crop(tmin=0.2, tmax=0.4, fmin=5, fmax=7)

@larsoner mentioned that he does not like this solution very much because .crop() sounds more time-related and it would stay to mean “crop in time” if we instead add another method for freq-cropping (Eric, please correct me if I misrepresent). I actually like putting it all into .crop() since I much prefer using kwargs anyway for clarity. Also crop sounds to me equally well suited for time, space or frequency (like trim, for example, but I’m not a native so my argument is weaker here 😃 ). I also think it would be easier for new mne users (like my students) to digest - one method for selection instead of multiple methods depending on the dimension type.

adding another method for frequency cropping

Another method that would work like .crop() but for frequency dimension. I don’t have a good idea for the method name in this case but just assuming it would be called .frop() the example from previous section would be:

tfr.crop(tmin=0.2, tmax=0.4).frop(fmin=5, fmax=7)

obviously I’m not suggesting to actually call it .frop() 😃

[larsoner]

Yes we should just override it in the TFR classes.

[larsoner]

Thinking about it more, I’m fine with crop. If we think of a “crop” operation as “trimming the current raw / average / epoch / source data for all channels/sources”, then you can crop in freq just as easily as time, and it’s different from other forms of subselection we have, such as trimming the set of epochs with epochs[10:20], subselecting source vertices with stc.in_label, or the set of channels with epochs.pick_types, etc.