PSTACK-COMPRESS: Experimental Code v2

The code implements PSTACK-COMPRESS (Algorithm 1 from the paper): an online, linear-time algorithm for lossless compression of rate-independent data streams based on the Preisach extremum stack.

The extremum stack Π_n records the alternating local maxima and minima of a discrete sequence that survive the Preisach wiping-out rule. The paper proves that Π_n is the Kolmogorov-minimal sufficient statistic for the class R of all computable rate-independent functionals, and that PSTACK-COMPRESS is asymptotically optimal: no online algorithm can produce a shorter exact representation for any R-query. The dataset includes: pstack_dam_experiments.py — full implementation of PSTACK-COMPRESS, eight synthetic signal generators (monotone, slow sine, trend+noise, white noise, piecewise/SCADA-like, ECG-like, financial random walk, adversarial worst-case), a self-test suite verifying the Definition 4 stack invariant at every time step, and benchmark routines reproducing all numerical results reported in Sections 7.1–7.4 of the paper (compression ratios, adversarial lower bound verification, white noise k/n distribution, speedup of online vs. naive algorithm).