mne.compute_rank¶

mne.
compute_rank
(inst, rank=None, scalings=None, info=None, tol='auto', proj=True, tol_kind='absolute', verbose=None)[source]¶ Compute the rank of data or noise covariance.
This function will normalize the rows of the data (typically channels or vertices) such that nonzero singular values should be close to one.
 Parameters
 instinstance of
Raw
,Epochs
, orCovariance
Raw measurements to compute the rank from or the covariance.
 rank
None
dict
 ‘info’  ‘full’ This controls the rank computation that can be read from the measurement info or estimated from the data. See
Notes
ofmne.compute_rank()
for details.The default is None. scalings
dict
None
(defaultNone
) Defaults to
dict(mag=1e15, grad=1e13, eeg=1e6)
. These defaults will scale different channel types to comparable values. infoinstance of
Info
None
The measurement info used to compute the covariance. It is only necessary if inst is a Covariance object (since this does not provide
inst.info
). tol
float
 ‘auto’ Tolerance for singular values to consider nonzero in calculating the rank. The singular values are calculated in this method such that independent data are expected to have singular value around one. Can be ‘auto’ to use the same thresholding as
scipy.linalg.orth()
. projbool
If True, all projs in
inst
andinfo
will be applied or considered whenrank=None
orrank='info'
. tol_kind
str
Can be: “absolute” (default) or “relative”. Only used if
tol
is a float, because whentol
is a string the mode is implicitly relative. After applying the chosen scale factors / normalization to the data, the singular values are computed, and the rank is then taken as:'absolute'
The number of singular values
s
greater thantol
. This mode can fail if your data do not adhere to typical data scalings.
'relative'
The number of singular values
s
greater thantol * s.max()
. This mode can fail if you have one or more large components in the data (e.g., artifacts).
New in version 0.21.0.
 verbosebool,
str
,int
, orNone
If not None, override default verbose level (see
mne.verbose()
and Logging documentation for more).
 instinstance of
 Returns
 rank
dict
Estimated rank of the data for each channel type. To get the total rank, you can use
sum(rank.values())
.
 rank
Notes
The
rank
parameter can be:None
(default)Rank will be estimated from the data after proper scaling of different channel types.
'info'
Rank is inferred from
info
. If data have been processed with Maxwell filtering, the Maxwell filtering header is used. Otherwise, the channel counts themselves are used. In both cases, the number of projectors is subtracted from the (effective) number of channels in the data. For example, if Maxwell filtering reduces the rank to 68, with two projectors the returned value will be 68.'full'
Rank is assumed to be full, i.e. equal to the number of good channels. If a
Covariance
is passed, this can make sense if it has been (possibly improperly) regularized without taking into account the true data rank.
New in version 0.18.