I was told that early tests with -st 10 in the Neuromag maxfilter command line tool inserted jumps in the data
Do you have any data you could share that show this problem? I’ve always suspected it’s an issue and have a method that fixes it (overlap-add processing) but haven’t published it (and thus haven’t implemented it in MNE) mostly on account of not having good cases to showcase the issue…
But coming back to your data, if I do:
import mne
import matplotlib.pyplot as plt
fig, axes = plt.subplots(
2, 4, figsize=(10, 4), sharex=True, sharey='row',
constrained_layout=True, squeeze=False)
raws = dict()
raws['orig'] = raw = mne.io.read_raw_fif('run01_anon.fif', allow_maxshield='yes').load_data()
raws['sss'] = mne.preprocessing.maxwell_filter(raw)
raws['tsss_10'] = mne.preprocessing.maxwell_filter(raw, st_duration=10)
raws['tsss_full'] = mne.preprocessing.maxwell_filter(raw, st_duration=raw.times[-1])
raws['tsss_full_only'] = mne.preprocessing.maxwell_filter(raw, st_duration=raw.times[-1], st_only=True)
for ri, (key, r) in enumerate(raws.items()):
r.plot_psd(ax=axes[:, ri])
axes[0, ri].set_title(key)
I get:
I don’t see huge differences here – am I missing something?
In any case, feel free to try st_duration=10 locally to see if it helps…