some entries in the dictionaries that glmsingle.py generates may be of type list, and it's not clear that this is desirable, since h5py cannot save ragged nested lists (errors out)
for example, in this dictionary, pcregressors and rrbadness are of type list. we should check whether this is necessary, and if so, fix the h5py saving to prevent bugs if the users generate ragged nested lists (e.g. pcregressors could be a list of arrays of different shape).
whenever h5py attempts to save a field that is an np array with type Object, there will be an error in create_dataset. this relates to lines such as the following, which will not cause issues if the entries in the list are the same shape. but, that will not always be the case.
some entries in the dictionaries that glmsingle.py generates may be of type list, and it's not clear that this is desirable, since h5py cannot save ragged nested lists (errors out)
for example, in this dictionary, pcregressors and rrbadness are of type list. we should check whether this is necessary, and if so, fix the h5py saving to prevent bugs if the users generate ragged nested lists (e.g. pcregressors could be a list of arrays of different shape).
outdict = { 'HRFindex': HRFindex, 'HRFindexrun': HRFindexrun, 'glmbadness': glmbadness, 'pcvoxels': pcvoxels, 'pcnum': pcnum, 'xvaltrend': xvaltrend, 'noisepool': noisepool, 'pcregressors': pcregressors, 'betasmd': modelmd, 'R2': R2, 'R2run': R2run, 'rrbadness': rrbadness, 'FRACvalue': FRACvalue, 'scaleoffset': scaleoffset, 'meanvol': meanvol }
whenever h5py attempts to save a field that is an np array with type Object, there will be an error in create_dataset. this relates to lines such as the following, which will not cause issues if the entries in the list are the same shape. but, that will not always be the case.