# Strange errors with calculating source estimates in a frequency band **URL:** \ **Category:** Mailing List Archive (read-only)\ **Tags:** list-archive\ **Created:** [July 31, 2014, 6:14am UTC](https://mne.discourse.group/t/strange-errors-with-calculating-source-estimates-in-a-frequency-band/767 "2014-07-31T06:14:38Z")\ **Posts on this page:** 7\ **Page:** 1
**Author:** ![system](https://global.discourse-cdn.com/free1/uploads/mne/original/1X/85cc6bd2b69cb698a166dc6d880fb550510d0144.jpeg) [@system](https://mne.discourse.group/u/system)\ **Post date:** [July 31, 2014, 6:14am UTC](https://mne.discourse.group/t/strange-errors-with-calculating-source-estimates-in-a-frequency-band/767/1 "2014-07-31T06:14:38Z")
Hi everyone! I have been trying to code a program that would enable to plot user selectable time windows of averaged source estimates in a certain frequency band in real time with FieldTrip Buffer. I'm now trying to simulate that in offline mode with looping through the epochs.get\_data() and then building an Epoch from a single epoch with EpochsArray (because that's the way the real time sytem works). I end up with a couple of odd errors using compute\_source\_psd\_epochs depending on how long my time window is. If my time window is shorter than 10 seconds, my error is this: /usr/lib/python2.7/dist-packages/IPython/utils/py3compat.pyc in execfile(fname, \*where)     173 else:     174 filename = fname --\> 175 \_\_builtin\_\_.execfile(filename, \*where) /home/silfves1/mne-python/examples/omatfunktiot/jatk\_off.py in \()     380     381 --\> 382 stc = compute\_source\_psd\_epochs(epochs=ep, inverse\_operator=inverse\_operator, lambda2=lambda2, method=method, fmin=15.0, fmax=25.0, pick\_ori=None, pca=True, bandwidth=bandwidth, n\_jobs=10, return\_generator=False)     383     384 /home/silfves1/.local/lib/python2.7/site-packages/mne-0.8.git-py2.7.egg/mne/minimum\_norm/time\_frequency.pyc in compute\_source\_psd\_epochs(epochs, inverse\_operator, lambda2, method, fmin, fmax, pick\_ori, label, nave, pca, inv\_split, bandwidth, adaptive, low\_bias, return\_generator, n\_jobs, verbose, pick\_normal) /home/silfves1/.local/lib/python2.7/site-packages/mne-0.8.git-py2.7.egg/mne/utils.pyc in verbose(function, \*args, \*\*kwargs)     549 return ret     550 else: --\> 551 ret = function(\*args, \*\*kwargs)     552 return ret     553 /home/silfves1/.local/lib/python2.7/site-packages/mne-0.8.git-py2.7.egg/mne/minimum\_norm/time\_frequency.pyc in compute\_source\_psd\_epochs(epochs, inverse\_operator, lambda2, method, fmin, fmax, pick\_ori, label, nave, pca, inv\_split, bandwidth, adaptive, low\_bias, return\_generator, n\_jobs, verbose, pick\_normal)     669 # return a list     670 stcs = list() --\> 671 for stc in stcs\_gen:     672 stcs.append(stc)     673 /home/silfves1/.local/lib/python2.7/site-packages/mne-0.8.git-py2.7.egg/mne/minimum\_norm/time\_frequency.pyc in \_compute\_source\_psd\_epochs(epochs, inverse\_operator, lambda2, method, fmin, fmax, pick\_ori, label, nave, pca, inv\_split, bandwidth, adaptive, low\_bias, n\_jobs, verbose)     519     520 dpss, eigvals = dpss\_windows(n\_times, half\_nbw, n\_tapers\_max, --\> 521 low\_bias=low\_bias)     522 n\_tapers = len(dpss)     523 /home/silfves1/.local/lib/python2.7/site-packages/mne-0.8.git-py2.7.egg/mne/time\_frequency/multitaper.pyc in dpss\_windows(N, half\_nbw, Kmax, low\_bias, interp\_from, interp\_kind)     202 # only calculate the highest Kmax eigenvalues     203 w = linalg.eigvals\_banded(ab, select='i', --\> 204 select\_range=(N - Kmax, N - 1))     205 w = w[::-1]     206 /usr/lib/python2.7/dist-packages/scipy/linalg/decomp.pyc in eigvals\_banded(a\_band, lower, overwrite\_a\_band, select, select\_range)     698 return eig\_banded(a\_band, lower=lower, eigvals\_only=1,     699 overwrite\_a\_band=overwrite\_a\_band, select=select, --\> 700 select\_range=select\_range)     701     702 \_double\_precision = ['i','l','d'] /usr/lib/python2.7/dist-packages/scipy/linalg/decomp.pyc in eig\_banded(a\_band, lower, eigvals\_only, overwrite\_a\_band, select, select\_range, max\_ev)     490 vl, vu, il, iu = 0.0, 0.0, min(select\_range), max(select\_range)     491 if min(il, iu) \< 0 or max(il, iu) \>= a1.shape[1]: --\> 492 raise ValueError('select\_range out of bounds')     493 max\_ev = iu - il + 1     494 else: # 1, 'v', 'value' ValueError: select\_range out of bounds However, if my time window is 10 seconds, here's the error then: error Traceback (most recent call last) /usr/lib/python2.7/dist-packages/IPython/utils/py3compat.pyc in execfile(fname, \*where)     173 else:     174 filename = fname --\> 175 \_\_builtin\_\_.execfile(filename, \*where) /home/silfves1/mne-python/examples/omatfunktiot/jatk\_off.py in \()     380     381 --\> 382 stc = compute\_source\_psd\_epochs(epochs=ep, inverse\_operator=inverse\_operator, lambda2=lambda2, method=method, fmin=15.0, fmax=25.0, pick\_ori=None, pca=True, bandwidth=bandwidth, n\_jobs=10, return\_generator=False)     383     384 /home/silfves1/.local/lib/python2.7/site-packages/mne-0.8.git-py2.7.egg/mne/minimum\_norm/time\_frequency.pyc in compute\_source\_psd\_epochs(epochs, inverse\_operator, lambda2, method, fmin, fmax, pick\_ori, label, nave, pca, inv\_split, bandwidth, adaptive, low\_bias, return\_generator, n\_jobs, verbose, pick\_normal) /home/silfves1/.local/lib/python2.7/site-packages/mne-0.8.git-py2.7.egg/mne/utils.pyc in verbose(function, \*args, \*\*kwargs)     549 return ret     550 else: --\> 551 ret = function(\*args, \*\*kwargs)     552 return ret     553 /home/silfves1/.local/lib/python2.7/site-packages/mne-0.8.git-py2.7.egg/mne/minimum\_norm/time\_frequency.pyc in compute\_source\_psd\_epochs(epochs, inverse\_operator, lambda2, method, fmin, fmax, pick\_ori, label, nave, pca, inv\_split, bandwidth, adaptive, low\_bias, return\_generator, n\_jobs, verbose, pick\_normal)     669 # return a list     670 stcs = list() --\> 671 for stc in stcs\_gen:     672 stcs.append(stc)     673 /home/silfves1/.local/lib/python2.7/site-packages/mne-0.8.git-py2.7.egg/mne/minimum\_norm/time\_frequency.pyc in \_compute\_source\_psd\_epochs(epochs, inverse\_operator, lambda2, method, fmin, fmax, pick\_ori, label, nave, pca, inv\_split, bandwidth, adaptive, low\_bias, n\_jobs, verbose)     545     546 # compute tapered spectra in sensor space --\> 547 x\_mt, freqs = \_mt\_spectra(data, dpss, sfreq)     548     549 if k == 0: /home/silfves1/.local/lib/python2.7/site-packages/mne-0.8.git-py2.7.egg/mne/time\_frequency/multitaper.pyc in \_mt\_spectra(x, dpss, sfreq, n\_fft)     439 # remove mean (do not use in-place subtraction as it may modify input x)     440 x = x - np.mean(x, axis=-1)[:, np.newaxis] --\> 441 x\_mt = fftpack.fft(x[:, np.newaxis, :] \* dpss, n=n\_fft)     442     443 # only keep positive frequencies /usr/lib/python2.7/dist-packages/scipy/fftpack/basic.pyc in fft(x, n, axis, overwrite\_x)     213     214 if axis == -1 or axis == len(tmp.shape) - 1: --\> 215 return work\_function(tmp,n,1,0,overwrite\_x)     216     217 tmp = swapaxes(tmp, axis, -1) error: failed in converting 1st argument `x' of \_fftpack.zrfft to C/Fortran array I don't know, why these error come up. Do some of you have any ideas to help? Thank you so much! Best regards Sebastian -------------- next part -------------- An HTML attachment was scrubbed... URL: [http://mail.nmr.mgh.harvard.edu/pipermail/mne\_analysis/attachments/20140731/55d6286b/attachment.html](http://mail.nmr.mgh.harvard.edu/pipermail/mne_analysis/attachments/20140731/55d6286b/attachment.html) ---
**Author:** ![system](https://global.discourse-cdn.com/free1/uploads/mne/original/1X/85cc6bd2b69cb698a166dc6d880fb550510d0144.jpeg) [@system](https://mne.discourse.group/u/system)\ **Post date:** [July 31, 2014, 9:22am UTC](https://mne.discourse.group/t/strange-errors-with-calculating-source-estimates-in-a-frequency-band/767/2 "2014-07-31T09:22:58Z")
Hi Sebastian, It seems to me that you are somehow iterating over the epochs and stc objects incorrectly. You do not need to use EpochsArray because you can iterate over the Epochs object directly (same with RtEpochs I think). For stc, if you have a single epoch, can you check if your iterations are correct? A small code snippet / gist that reproduces this issue would be great. Mainak ---
**Author:** ![system](https://global.discourse-cdn.com/free1/uploads/mne/original/1X/85cc6bd2b69cb698a166dc6d880fb550510d0144.jpeg) [@system](https://mne.discourse.group/u/system)\ **Post date:** [July 31, 2014, 10:37am UTC](https://mne.discourse.group/t/strange-errors-with-calculating-source-estimates-in-a-frequency-band/767/3 "2014-07-31T10:37:39Z")
This next snippet is from the offline version I've been testing: for ii, data in enumerate(epochs.get\_data()):     # make the 2D epoch to 3D     data = data[newaxis, ...]     ep = EpochsArray(data=data, info=raw.info, events=events, tmin=0,event\_id=event\_id, reject=reject)     stc = compute\_source\_psd\_epochs(epochs=ep, inverse\_operator=inverse\_operator, lambda2=lambda2,method=method, fmin=15.0, fmax=25.0, pick\_ori=None, pca=True, bandwidth=bandwidth, n\_jobs=10, return\_generator=False) and the error comes in the last command. What is the best way to iterate over rt\_epochs? rt.epochs.next()? I'm now doing just for ii, data in enumerate(rt\_epochs): Sebastian ---
**Author:** ![system](https://global.discourse-cdn.com/free1/uploads/mne/original/1X/85cc6bd2b69cb698a166dc6d880fb550510d0144.jpeg) [@system](https://mne.discourse.group/u/system)\ **Post date:** [July 31, 2014, 8:31pm UTC](https://mne.discourse.group/t/strange-errors-with-calculating-source-estimates-in-a-frequency-band/767/4 "2014-07-31T20:31:54Z")
What are you using for `events`? It should have the same number of entries as the number of epochs (1), I suspect it might not. (We should probably add a check in mne-python that this is the case...) Eric ---
**Author:** ![system](https://global.discourse-cdn.com/free1/uploads/mne/original/1X/85cc6bd2b69cb698a166dc6d880fb550510d0144.jpeg) [@system](https://mne.discourse.group/u/system)\ **Post date:** [July 31, 2014, 10:20pm UTC](https://mne.discourse.group/t/strange-errors-with-calculating-source-estimates-in-a-frequency-band/767/5 "2014-07-31T22:20:35Z")
Also, can you try something like this? for ii, ep in enumerate(epochs):     stc = compute\_source\_psd\_epochs(epochs=ep ...) which would not require EpochsArray. Mainak ---
**Author:** ![system](https://global.discourse-cdn.com/free1/uploads/mne/original/1X/85cc6bd2b69cb698a166dc6d880fb550510d0144.jpeg) [@system](https://mne.discourse.group/u/system)\ **Post date:** [August 1, 2014, 7:06am UTC](https://mne.discourse.group/t/strange-errors-with-calculating-source-estimates-in-a-frequency-band/767/6 "2014-08-01T07:06:05Z")
Eric: I have tried to use the events with which I have determined the epochs which has 99 events. That didn't work so I tried to make another events with just one event with this:           # make the 2D epoch to 3D (because the EpochsArray has to be given 3D data     ep = ep[newaxis, ...]     new\_event\_list = np.empty([1, 3])     new\_event\_list[0] = [event\_list[ii][0], 0, 10] # put here the measurement recording start time.     # Make your own filepath to which to make the event     fname\_new\_event = 'FILEPATH'     # Write the event     write\_events(fname\_new\_event, new\_event\_list)     # Finally make the event     new\_events = read\_events(fname\_new\_event) and from here on I use the EpochsArray and compute\_source\_psd\_epochs. Mainak: I tried your suggestion and that doesn't work either: AttributeError Traceback (most recent call last) /usr/lib/python2.7/dist-packages/IPython/utils/py3compat.pyc in execfile(fname, \*where)     173 else:     174 filename = fname --\> 175 \_\_builtin\_\_.execfile(filename, \*where) /home/silfves1/mne-python/examples/omatfunktiot/jatk\_off.py in \()     345 # for ii, ep in enumerate(epochs.get\_data()):     346 for ii, ep in enumerate(epochs): --\> 347 stc = compute\_source\_psd\_epochs(epochs=ep, inverse\_operator=inverse\_operator, lambda2=lambda2, method=method, fmin=15.0, fmax=25.0, pick\_ori=None, pca=True, bandwidth=bandwidth, n\_jobs=10, return\_generator=False)     348     349 # make the 2D epoch to 3D /home/silfves1/.local/lib/python2.7/site-packages/mne-0.8.git-py2.7.egg/mne/minimum\_norm/time\_frequency.pyc in compute\_source\_psd\_epochs(epochs, inverse\_operator, lambda2, method, fmin, fmax, pick\_ori, label, nave, pca, inv\_split, bandwidth, adaptive, low\_bias, return\_generator, n\_jobs, verbose, pick\_normal) /home/silfves1/.local/lib/python2.7/site-packages/mne-0.8.git-py2.7.egg/mne/utils.pyc in verbose(function, \*args, \*\*kwargs)     549 return ret     550 else: --\> 551 ret = function(\*args, \*\*kwargs)     552 return ret     553 /home/silfves1/.local/lib/python2.7/site-packages/mne-0.8.git-py2.7.egg/mne/minimum\_norm/time\_frequency.pyc in compute\_source\_psd\_epochs(epochs, inverse\_operator, lambda2, method, fmin, fmax, pick\_ori, label, nave, pca, inv\_split, bandwidth, adaptive, low\_bias, return\_generator, n\_jobs, verbose, pick\_normal)     669 # return a list     670 stcs = list() --\> 671 for stc in stcs\_gen:     672 stcs.append(stc)     673 /home/silfves1/.local/lib/python2.7/site-packages/mne-0.8.git-py2.7.egg/mne/minimum\_norm/time\_frequency.pyc in \_compute\_source\_psd\_epochs(epochs, inverse\_operator, lambda2, method, fmin, fmax, pick\_ori, label, nave, pca, inv\_split, bandwidth, adaptive, low\_bias, n\_jobs, verbose)     483 # Pick the correct channels from the data     484 # --\> 485 sel = \_pick\_channels\_inverse\_operator(epochs.ch\_names, inv)     486 logger.info('Picked %d channels from the data' % len(sel))     487 logger.info('Computing inverse...') AttributeError: 'numpy.ndarray' object has no attribute 'ch\_names' because then the compute\_source\_psd\_epochs uses a numpy array as an epoch. Sebastian ---
**Author:** ![system](https://global.discourse-cdn.com/free1/uploads/mne/original/1X/85cc6bd2b69cb698a166dc6d880fb550510d0144.jpeg) [@system](https://mne.discourse.group/u/system)\ **Post date:** [August 1, 2014, 11:08am UTC](https://mne.discourse.group/t/strange-errors-with-calculating-source-estimates-in-a-frequency-band/767/7 "2014-08-01T11:08:00Z")
I had a look at the script together with Sebastian. This works: for ii, ep in enumerate(epochs):     stc = compute\_source\_psd\_epochs(epochs=ep ..., return\_generator=False) but setting return\_generator=True does not work. I suspect it returns an empty generator but maybe others have more insight. Mainak