Alternating stable schedules made two of 36 fixed SEEG readers unstable during simulated 100 ms gaps

Two individually stable observation schedules gave an unstable alternating product in two of 36 fixed SEEG readers. The readers used nine public rest records from OpenNeuro ds006065 and ds005670, version 1.0.0. All gaps were simulated and lasted 100 ms. The strongest example had fixed-schedule spectral radii of 0.9786 and 0.8344. Their two-cycle product had radius 2.3777. A guard limited the gap map norm to 0.95 without the next gap time. A proof bounds state differences for any nonnegative observed intervals under the fixed model and history assumptions. The guard changed two readers. It stopped large state growth in replay. It increased mean endpoint phase error from 61.21 to 62.29 degrees in a fixed 200 ms control. This is a numerical filter result on a reused clinical cohort. It does not establish true neural phase or stimulation benefit. The package contains code, exact decimal-matrix certificates, native voltage extracts, source checks, results, figures, and reproduction instructions.

stereoelectroencephalography · switched systems · state estimation · stability · signal processing · simulated gaps