Rest SEEG pairwise covariance passed 18 PSD tests with common average reference and three scaled MAD voltage masks without guards. Seven Pearson matrices failed.

We tested two rest SEEG segments from each of nine adults in OpenNeuro ds003688, version 1.0.7. Each segment had 50 seconds of data in the 1–40 Hz band. The main test used common average reference and voltage masks at three scaled median absolute deviations (MAD). It used no mask guard. All 18 pairwise covariance matrices passed the positive semidefinite (PSD) test. The secondary Pearson test failed for seven matrices from four people. These failures needed 0.0418–0.5845% identity mixing to meet the PSD condition. All 319 available shared sample estimates passed. They retained fewer samples. Tests of other references, mask thresholds, and guards gave 11 covariance failures in 324 conditions. Five conditions had no shared samples. These results apply to this fixed cohort. The voltage masks are not confirmed artifact labels. The package includes compact signals, source versions, code, results, a proof, and repeat instructions. All nine fresh source ranges gave exact signal matches. Separate estimators and a full offline repeat confirmed the numerical results.

SEEG · pairwise covariance · Pearson correlation · positive semidefinite matrices · sample deletion · voltage masks · signal processing