Entropy Visit Residuals Persist After Markov Finite-Window Corrections in Macrosystems

Authors: Sterling Geisel, QBist Lab; Dr. Hideo Tanaka, QBist Lab

Abstract

Polski and Skrebnev argue that entropy is not a static logarithm of quantum levels near a moving thermodynamic energy. They replace that picture with visit configurations counted over an observation interval. This paper applies one Pudding Theory Postulate: Temporal Softening. The application is narrow. It does not add observer causation, intent, or field bias to the source paper. It asks whether the time window has empirical structure after ordinary finite-sample and autocorrelation effects have been removed. The resulting prediction is a residual entropy-per-visit convergence term in systems whose transition architecture is not captured by an iid or fitted finite-order Markov null. The test is preregistered: fixed bins, fixed window schedule, control-derived null corrections, and a residual threshold. If the residual vanishes under that protocol, the Postulate is falsified in this domain. The claim is modest, measurable, and conditional on the visit ontology.

Source Synopsis

Polski and Skrebnev examine the standard statistical-mechanical statement that entropy is the logarithm of the number of quantum states of a physical system. Their target is the textbook move in which energy levels relevant to a macrostate are treated as concentrated in a narrow interval around the thermodynamic energy \(E\). If probabilities in that interval are equal, one obtains \(S=\ln W\), where \(W\) is the number of states.

They reject the physical basis of this construction. The spectrum of the Hamiltonian is fixed by the system. It does not move when the thermodynamic energy changes with temperature. Their criticism is therefore direct: “Accordingly, entropy cannot be equal to the logarithm of the number of states that do not exist.”

The paper then derives entropy from the canonical probability. Substituting the macroscopic energy into the canonical distribution gives

\[ \ln \rho(E)=-(E-F)/kT . \]

Using \(E-F=TS\), the authors obtain

\[ -\ln \rho(E)=S/k . \]

They treat entropy first as a dimensionless logarithmic quantity, with Boltzmann’s constant restored for thermodynamic units.

The constructive part of the paper replaces static state counting with temporal occurrence. A system is said to visit an energy state \(E_n\). During an observation time, the total number of visits is \(N\), and the number of visits to state \(n\) is \(\nu_n\). A visit configuration can be realized in

\[ P=\frac{N!}{\nu_1!\nu_2!\cdots\nu_l!}. \]

The configuration with maximal \(P\) yields the canonical distribution, and the authors write the entropy as

\[ S=\frac{\ln P_{\max}}{N}=-\ln \rho(E). \]

They attribute the underlying transitions to subquantum processes. That mechanism is not operationally specified in the paper. The measurable content is the visit sequence and its convergence toward the canonical distribution.

Postulate Lens

The applied Postulate is Temporal Softening.

The fit is restricted to finite-time convergence. Temporal Softening is not used here as a general claim that time alters every physical law. It is used as the weaker statement that a macroscopic quantity may become empirically rigid only after enough fluctuations and transitions have accumulated. In Polski and Skrebnev, entropy is not assigned to a frozen shell of levels. It is assigned to the maximal combinatorics of visits over an observation interval. The interval is therefore part of the construction, not only a measurement inconvenience.

This is the point where the Postulate does work. If it were removed, the paper would still predict ordinary estimator bias, estimator variance, and mixing-time effects. Those are standard stochastic-process facts. Pudding Theory cannot claim them as new physics. The distinct commitment is smaller: after a conventional finite-window mixing null has been specified and subtracted, a residual temporal term should remain when the visit generator contains slow structure not represented by the null.

That statement is falsifiable. It is also risky. If the residual is absent, then Temporal Softening adds nothing to this source domain. The Postulate does not rescue the source paper’s subquantum terminology. It only turns the visit ontology into a finite-window residual test.

Pudding Theory Prediction

Let an experimental platform generate a resolved sequence of energy-bin visits \(x_t\) under fixed Hamiltonian, fixed bath temperature, and fixed binning. Define the empirical visit counts \(\nu_n(T_{\mathrm{obs}})\) for each observation window \(T_{\mathrm{obs}}\), and compute

\[ \hat S(T_{\mathrm{obs}})=\frac{\ln P_{\max}(T_{\mathrm{obs}})}{N(T_{\mathrm{obs}})} . \]

The long-window value \(S_\infty\) is estimated from the longest stable record or from an independently calibrated canonical distribution. The null model is not vague equilibrium. It is a fitted finite-order Markov or bath-coupled Hamiltonian simulation, chosen before analysis, with the same energy bins, same sampling cadence, same transition rate matrix uncertainty, and same finite-window estimator. This null gives a predicted bias \(b_{\mathrm{null}}(T_j)\) and standard deviation \(\sigma_{\mathrm{null}}(T_j)\) for each scheduled window.

Pudding Theory predicts that some systems will show

\[ R_j=\frac{\hat S(T_j)-S_\infty-b_{\mathrm{null}}(T_j)} {\sigma_{\mathrm{null}}(T_j)} \]

with coherent structure across \(T_j\), rather than independent scatter around zero. The predicted form is not a universal constant. A single dominant slow mode gives a signed exponential tail. A hierarchy of hidden modes gives a stretched or multi-scale tail. Fast-mixing systems should give no residual once the null correction is applied.

A clean protocol uses mesoscopic systems with countable transitions: trapped-ion thermalization, superconducting qubit arrays coupled to engineered baths, optically tracked colloids, or a classical stochastic analog with known transition rules. The window schedule is geometric, for example

\[ T_j/\tau_1 = 1,2,4,8,16,32,64,128 , \]

where \(\tau_1\) is the measured one-step decorrelation time. Bins are fixed before analysis by equal-energy intervals or by instrument-resolved pointer states. No adaptive rebinding is permitted after looking at residuals.

The prediction is therefore not “finite samples matter.” They do. The prediction is that the visit construction exposes temporal structure not exhausted by the preregistered mixing null.

Falsifiable Observable

The distinguishing observable is the preregistered residual vector \(R_j\) after subtracting finite-sample bias and autocorrelation corrections from the declared null model. The binning rule, sampling cadence, window schedule, null class, bootstrap method, and threshold must be fixed before data collection. If the preregistered residual vector \(R_j\) were measured to satisfy \(\max_j |R_j|\le 1\) with no fitted slow component exceeding the 95 percent null envelope in three independent platforms, this Postulate would be falsified. That outcome would show that Temporal Softening contributes no observable term beyond conventional finite-window stochastic mechanics in this domain.

Editorial Dialogue

Tanaka critique, verbatim:

Temporal Softening is a stretch unless Sterling explicitly strips it to finite-time convergence. Which postulate is doing the work here? If the observer, intent, Lumina, proximity, and field bias are absent, then the remaining claim is ordinary mixing time plus finite-sample entropy estimation. Would removing Temporal Softening change the prediction? If not, drop the postulate or rename the lens as a non-PT analogy. The draft does well by refusing observer-causal inflation, but it still lets PT take credit for a prediction that follows from stochastic process theory alone.

The falsifiability sentence is not yet operational. What measurement would distinguish this from the null? “Exactly zero within experimental precision” is not a usable falsifier because finite samples always leave estimator variance and bias. Define the null model first. Is it iid sampling, a fitted Markov chain, a known bath-coupled Hamiltonian simulation, or conventional equilibrium statistical mechanics with finite-window corrections? Name the observable that would falsify Temporal Softening here. It should be a residual after subtracting the predicted finite-sample and autocorrelation corrections, with a preregistered threshold, binning rule, and window schedule.

The unsupported risk is mostly citation and framing. Are all references verifiable in the source bibliography? The arXiv paper, the 2025 JHEPGC citation, and the 2017 DOI need resolution checks before authorship is asserted. The single quoted sentence appears fair-use safe, but quote count should be logged. The claim that Landau and Lifshitz “remain operational” is plausible, but it needs a page or section reference. The source-paper dependence is high. If the source paper had not been published, would this prediction follow from PT alone?

Not yet. Replace the falsifier with a preregistered residual test against an explicit finite-sample mixing null.

Sterling response:

Accepted. The Postulate has been stripped to finite-time convergence. The prediction is no longer the existence of finite-window bias. That belongs to stochastic process theory. The claimed Pudding Theory term is the residual after a declared finite-window mixing null has done its work.

The falsifier is now operational. The observable is \(R_j\). It is calculated only after subtracting null bias and scaling by null variance. The protocol fixes bins, windows, sampling cadence, and threshold before data collection.

The source dependence is also explicit. Without Polski and Skrebnev’s visit construction, Pudding Theory alone would not generate this paper. It would only say that temporal accumulation can matter. The source paper supplies the entropy variable. The Postulate supplies a risky residual claim.

Citation scope has been narrowed. All supporting references are from the source paper’s own bibliography. The Landau and Lifshitz reference is cited as the source paper’s standard comparator, specifically its entropy discussion in Chapter I, Section 7.

Discussion

This paper does not decide the dispute between textbook state counting and visit configurations. Landau and Lifshitz give the standard statistical-physics construction in Chapter I, Section 7, and Polski and Skrebnev challenge the physical interpretation of that construction. The present claim is not that the challenge succeeds. It is that their time-indexed formulation creates a measurable object.

The weakest point remains the source paper’s subquantum language. It names a cause for visits but does not define an instrument that isolates that cause. This working paper therefore treats the generator as a black box. The test requires only transition records and a null model strong enough to absorb ordinary finite-window behavior.

A null result would matter. If \(R_j\) remains inside the preregistered envelope across platforms, Temporal Softening has no role here. A positive result would not prove the source ontology. It would show only that entropy-per-visit convergence contains temporal structure not captured by the selected null. Further work would then have to decide whether the residual is hidden Markov structure, bath miscalibration, coarse-graining error, or new dynamics.

References

[1] Maria Polski and Vladimir Skrebnev, “Quantum states of macrosystems and entropy,” arXiv:2602.06060v1, 2026. DOI: doi:10.48550/arXiv.2602.06060.

[2] Sterling Geisel, “Pudding Theory: A Topological Theory of Information Fields,” QBist Lab, September 10, 2025.

[3] L. D. Landau and E. M. Lifshitz, Statistical Physics, 3rd ed., Part 1, Butterworth-Heinemann, Oxford, 1996. See Chapter I, Section 7.

[4] M. V. Polski and V. A. Skrebnev, “An Alternative Derivation of Canonical Distribution as a Result of Irreversible Processes in Macrosystems,” European Journal of Physics 38, 025101, 2017. DOI: doi:10.1088/1361-6404/38/2/025101.

[5] V. A. Skrebnev and M. V. Polski, “Subquantum Processes as the True Cause of the Canonical Distribution and the Entropy,” Journal of High Energy Physics, Gravitation and Cosmology 11, 938-942, 2025. DOI: doi:10.4236/jhepgc.2025.113060.

[6] L. Boltzmann, “Uber die Beziehung zwischen dem zweiten Hauptsatze der mechanischen Warmtheorie und der Wahrscheinlichkeitsrechnung den Satzen uber das Warmegleichgewicht,” Wiener Berichte 76, 373-435, 1877.