flip-observer-1024: synthetic classical channel-reversal observer calibration prototype (512 inputs, two noisy banks)

Abstract

flip-observer-1024 is a deterministic, offline Python prototype for an ordinary classical weighted readout under a hidden channel-order reversal. It uses 512 dimensionless synthetic input channels. The optional 1,024-slot layout is two noisy/dropout-affected banks observing the same 512 underlying signals—not 1,024 independent latent channels. Model. Target z(t)=wᵀx(t) is defined before any hidden reversal. Calibration compares only identity vs exact reversal using known orthogonal full-channel patterns. Flip means reversed channel order; P⁻¹=P. The scalar z is an ordinary weighted sum—not quantum superposition. No entanglement, membrane, or biological model. Evidence. All traces, noise, and dropout are SYNTHETIC and seed-controlled. Bundled results and visualization are from the default local sweep (108 held-out trials across 9 noise/dropout regimes). Limits. Software toy only: not a physical observer, communications product, medical device, quantum model, or claim of arbitrary-permutation recovery. v0.1.1 adds independent-review fixes (fully generated sample report with a report-sync test; scope note on identity vs exact reversal). Cite all versions via the concept DOI 10.5281/zenodo.23173449.