This is a slight modification of np.linalg.matrix_rank. The tolerance performs poorly for some matrices Here the tolerance is boosted by

General linear model (for t contrasts). :Parameters: - *Y* --- (J x Q) numpy array (dependent variable) - *X* --- (J x B) design ma

New version of reml 2025-05-18 which uses np.array instead of np.matrix to avoid NumPy linear algebra deprecation warnings

One-sample t test. :Parameters: - *Y* --- (J x Q) data array (J responses, Q nodes) - *y0* --- optional Q-component datum array (d

Cluster-level inference. Probability that 1D Gaussian fields would produce an upcrossing of extent *k* when thresholded at *u*. .. warn

Compute the D'Agostino-Pearson K2 test statistic at a single node. This code is modified from "DagosPtest.m" by Antonio Trujillo-Ortiz available

Bonferroni correction. When fields are very rough a Bonferroni correction might be less severe than the RFT threshold. This function yields Bonf

Modified from: https://stackoverflow.com/questions/38544460/how-to-generate-permutations-without-generating-repeating-results-but-with-a-fix

Plot a condfidence interval.

Set-level inference. Probability that 1D Gaussian fields would produce at least *c* upcrossings with a minimum extent of *k* when thresholde