import osimport matplotlib.pyplot as pltimport numpy as npimport pandas as pdimport seaborn as snsimport mnefrom mne.preprocessing import find_bad_channels_maxwellsample_data_folder = mne.datasets.sample.data_path()sample_data_raw_file = os.path.join(    sample_data_folder, "MEG", "sample", "sample_audvis_raw.fif")raw = mne.io.read_raw_fif(sample_data_raw_file, verbose=False)raw.crop(tmax=60)

fine_cal_file = os.path.join(sample_data_folder, "SSS", "sss_cal_mgh.dat")crosstalk_file = os.path.join(sample_data_folder, "SSS", "ct_sparse_mgh.fif")

raw.info["bads"] = []raw_check = raw.copy()auto_noisy_chs, auto_flat_chs, auto_scores = find_bad_channels_maxwell(    raw_check,    cross_talk=crosstalk_file,    calibration=fine_cal_file,    return_scores=True,    verbose=True,)print(auto_noisy_chs)  # we should find them!print(auto_flat_chs)  # none for this dataset

Applying low-pass filter with 40.0 Hz cutoff frequency ...Reading 0 ... 36037  =      0.000 ...    60.000 secs...Filtering raw data in 1 contiguous segmentSetting up low-pass filter at 40 HzFIR filter parameters---------------------Designing a one-pass, zero-phase, non-causal lowpass filter:- Windowed time-domain design (firwin) method- Hamming window with 0.0194 passband ripple and 53 dB stopband attenuation- Upper passband edge: 40.00 Hz- Upper transition bandwidth: 10.00 Hz (-6 dB cutoff frequency: 45.00 Hz)- Filter length: 199 samples (0.331 s)Scanning for bad channels in 12 intervals (5.0 s) ...    No bad MEG channels    Processing 204 gradiometers and 102 magnetometers    Using fine calibration sss_cal_mgh.dat        Adjusting non-orthogonal EX and EY        Adjusted coil orientations by (μ ± σ): 0.5° ± 0.4° (max: 2.1°)    Automatic origin fit: head of radius 91.2 mm    Using origin -4.2, 16.4, 51.8 mm in the head frame        Interval   1:    0.000 -    4.998        Interval   2:    5.000 -    9.998        Interval   3:   10.000 -   14.998        Interval   4:   15.000 -   19.998        Interval   5:   20.000 -   24.998        Interval   6:   24.999 -   29.998        Interval   7:   29.999 -   34.997        Interval   8:   34.999 -   39.997        Interval   9:   39.999 -   44.997        Interval  10:   44.999 -   49.997        Interval  11:   49.999 -   54.997        Interval  12:   54.999 -   60.000    Static bad channels:  ['MEG 2443']    Static flat channels: [][done]['MEG 2443'][]

bads = raw.info["bads"] + auto_noisy_chs + auto_flat_chsraw.info["bads"] = bads

# Only select the data for gradiometer channels.ch_type = "grad"ch_subset = auto_scores["ch_types"] == ch_typech_names = auto_scores["ch_names"][ch_subset]scores = auto_scores["scores_noisy"][ch_subset]limits = auto_scores["limits_noisy"][ch_subset]bins = auto_scores["bins"]  # The the windows that were evaluated.# We will label each segment by its start and stop time, with up to 3# digits before and 3 digits after the decimal place (1 ms precision).bin_labels = [f"{start:3.3f} – {stop:3.3f}" for start, stop in bins]# We store the data in a Pandas DataFrame. The seaborn heatmap function# we will call below will then be able to automatically assign the correct# labels to all axes.data_to_plot = pd.DataFrame(    data=scores,    columns=pd.Index(bin_labels, name="Time (s)"),    index=pd.Index(ch_names, name="Channel"),)# First, plot the "raw" scores.fig, ax = plt.subplots(1, 2, figsize=(12, 8), layout="constrained")fig.suptitle(    f"Automated noisy channel detection: {ch_type}", fontsize=16, fontweight="bold")sns.heatmap(data=data_to_plot, cmap="Reds", cbar_kws=dict(label="Score"), ax=ax[0])[    ax[0].axvline(x, ls="dashed", lw=0.25, dashes=(25, 15), color="gray")    for x in range(1, len(bins))]ax[0].set_title("All Scores", fontweight="bold")# Now, adjust the color range to highlight segments that exceeded the limit.sns.heatmap(    data=data_to_plot,    vmin=np.nanmin(limits),  # bads in input data have NaN limits    cmap="Reds",    cbar_kws=dict(label="Score"),    ax=ax[1],)[    ax[1].axvline(x, ls="dashed", lw=0.25, dashes=(25, 15), color="gray")    for x in range(1, len(bins))]ax[1].set_title("Scores > Limit", fontweight="bold")

raw.info["bads"] += ["MEG 2313"]  # from manual inspection

raw_sss = mne.preprocessing.maxwell_filter(    raw, cross_talk=crosstalk_file, calibration=fine_cal_file, verbose=True)

Maxwell filtering raw data    Bad MEG channels being reconstructed: ['MEG 2443', 'MEG 2313']    Processing 204 gradiometers and 102 magnetometers    Using fine calibration sss_cal_mgh.dat        Adjusting non-orthogonal EX and EY        Adjusted coil orientations by (μ ± σ): 0.5° ± 0.4° (max: 2.1°)    Automatic origin fit: head of radius 91.2 mm    Using origin -4.2, 16.4, 51.8 mm in the head frame    Loading raw data from disk        Using 87/95 harmonic components for    0.000  (72/80 in, 15/15 out)    Processing    6 data chunks of (at least) 10.0 s with 0.0 s overlap and boxcar windowing    The final 0.0033299202193064646 s will be lumped into the final window        Using 87/95 harmonic components for    0.000  (72/80 in, 15/15 out)[done]

raw.pick(["meg"]).plot(duration=2, butterfly=True)raw_sss.pick(["meg"]).plot(duration=2, butterfly=True)

chpi_fif_file = os.path.join(    mne.datasets.testing.data_path(), "SSS", "test_move_anon_raw.fif")raw = mne.io.read_raw_fif(chpi_fif_file, allow_maxshield="yes")# time-resolved information on active HPI coils# if all hpi were inactive n_active is a zero-arrayn_active = mne.chpi.get_active_chpi(raw)print(f"Average number of coils active during recording: {n_active.mean()}")

Opening raw data file /home/circleci/mne_data/MNE-testing-data/SSS/test_move_anon_raw.fif...    Read a total of 12 projection items:        mag.fif : PCA-v1 (1 x 306)  idle        mag.fif : PCA-v2 (1 x 306)  idle        mag.fif : PCA-v3 (1 x 306)  idle        mag.fif : PCA-v4 (1 x 306)  idle        mag.fif : PCA-v5 (1 x 306)  idle        mag.fif : PCA-v6 (1 x 306)  idle        mag.fif : PCA-v7 (1 x 306)  idle        grad.fif : PCA-v1 (1 x 306)  idle        grad.fif : PCA-v2 (1 x 306)  idle        grad.fif : PCA-v3 (1 x 306)  idle        grad.fif : PCA-v4 (1 x 306)  idle        Average EEG reference (1 x 60)  idle    Range : 10800 ... 31199 =      9.000 ...    25.999 secsReady.Using 5 HPI coils: 83 143 203 263 323 HzAverage number of coils active during recording: 5.0

head_pos_file = os.path.join(    mne.datasets.testing.data_path(), "SSS", "test_move_anon_raw.pos")head_pos = mne.chpi.read_head_pos(head_pos_file)mne.viz.plot_head_positions(head_pos, mode="traces")