Dpss

Namespace: UTILSLIB  ·  Library: Utilities Library

#include <utils/dpss.h>

class UTILSLIB::Dpss

Discrete Prolate Spheroidal Sequences (Slepian tapers).

Computes DPSS tapers via the tridiagonal eigenvalue formulation. These tapers are used by the multitaper spectral estimator to achieve optimal spectral concentration within a given half-bandwidth.

src/examples/ex_dsp_spectral/main.cpp

const DpssResult dpss = Dpss::compute(64, 2.5, 4); // N = 64 samples, NW = 2.5, 4 tapers

---

Static Methods

compute(nSamples, halfBandwidth, nTapers, sym, lowBias)

Compute DPSS tapers for a given window length and half-bandwidth.

Parameters:

• nSamples : int Window length N in samples.

• halfBandwidth : double Half-bandwidth parameter (NW); typical values: 2, 3, 4.

• nTapers : int Number of tapers to return; -1 → floor(2*halfBandwidth - 1).

• sym : bool false: periodic tapers for spectral analysis, i.e. the first N samples of the length N + 1 tapers (scipy dpss sym=False, used by MNE-Python).

• lowBias : bool Keep only tapers with concentration > 0.9, at least the first (MNE-Python low_bias).

Returns:

• DpssResult — DpssResult with matTapers (nTapers × N) and vecEigenvalues.

---

Example

• Full source: src/examples/ex_dsp_spectral/main.cpp
• Build target: ex_dsp_spectral (runs under CTest)

Authors of this file

• Christoph Dinh <christoph.dinh@mne-cpp.org>