Coherent Comparison as Information Cost: A Cost-First Ledger Framework for Discrete Dynamics
Sebastian Pardo-Guerra, Megan Simons, Anil Thapa, Jonathan Washburn
TL;DR
This paper develops a cost-first framework for discrete dynamics by treating recognition as ratio-based comparison; coherence of sequential comparisons yields a unique reciprocal cost $J(x)=\tfrac{1}{2}(x+x^{-1})-1$, which then drives a minimal ledger encoding of recognition events. Under deterministic, minimally-structured updates and explicit conservation and locality assumptions, recognition events induce atomic ticks and balanced postings; quantization further yields a discrete posting arithmetic and a time-aggregated no-arbitrage principle. The authors show that time-aggregated cycle closure implies path-independence and, via the discrete Poincaré lemma, the existence and uniqueness of a scalar potential on each connected component. On hypercube graphs $Q_d$, atomicity imposes a lower bound $T\ge 2^d$ for visiting all vertices, with a Gray-code realization at $d=3$; conditional arguments (involving topological linking and synchronization) further select $d=3$. Taken together, the work connects ratio-based divergences, conservative graph flows, and discrete potential theory within a coherent information-theoretic ledger framework, offering a principled basis for discrete dynamics with potential links to physical and informational structures.
Abstract
We develop an information-theoretic framework for discrete dynamics grounded in a comparison-cost functional on ratios. Given two quantities compared via their ratio \(x=a/b\), we assign a cost \(F(x)\) measuring deviation from equilibrium (\(x=1\)). Requiring coherent composition under multiplicative chaining imposes a d'Alembert functional equation; together with normalization (\(F(1)=0\)) and quadratic calibration at unity, this yields a unique reciprocal cost functional (proved in a companion paper): \[ J(x) = \tfrac{1}{2}\bigl(x + x^{-1}\bigr) - 1. \] This cost exhibits reciprocity \(J(x)=J(x^{-1})\), vanishes only at \(x=1\), and diverges at boundary regimes \(x\to 0^+\) and \(x\to\infty\), excluding ``nothingness'' configurations. Using \(J\) as input, we introduce a discrete ledger as a minimal lossless encoding of recognition events on directed graphs. Under deterministic update semantics and minimality (no intra-tick ordering metadata), we derive atomic ticks (at most one event per tick). Explicit structural assumptions (conservation, no sources/sinks, pairwise locality, quantization in \(δ\mathbb{Z}\)) force balanced double-entry postings and discrete ledger units. To obtain scalar potentials on graphs with cycles while retaining single-edge impulses per tick, we impose time-aggregated cycle closure (no-arbitrage/clearing over finite windows). Under this hypothesis, cycle closure is equivalent to path-independence, and the cleared cumulative flow admits a unique scalar potential on each connected component (up to additive constant), via a discrete Poincaré lemma. On hypercube graphs \(Q_d\), atomicity imposes a \(2^d\)-tick minimal period, with explicit Gray-code realization at \(d=3\). The framework connects ratio-based divergences, conservative graph flows, and discrete potential theory through a coherence-forced cost structure.