The Principle of Nonuniformity Three Structural Quantities, a Master Equation Graded in Four Layers, and the Correct Classification of Aggregation Limits

Abstract

Abstract: The macroscopic phenomenological apparatus of open flow-through systems — an income-minus-expenditure master equation, a gradient-flow relaxation, and a quasi-potential landscape — is usually posed as a set of postulates. This paper assembles that apparatus into a diagnostic framework and draws its boundary of validity. The organizing proposition is retained as a first principle for open systems, the Principle of Nonuniformity: the state of an open system departs structurally from uniformity along both a cross-sectional and a temporal axis, a departure supplied continuously by work and paid for by non-negative internal entropy production. Where an ingredient of the apparatus cannot be derived it is graded rather than asserted, and in this version the one ingredient that carries the apparatus is graded in four layers, of which two are rigorous. V14 makes one regrading, one correction and four completions. Each is argued where it occurs. The regrading, which is the change that matters. V13 recorded the master equation of Section 4.2 as the single load-bearing phenomenological item in the whole chain, layer B rather than layer A, and recorded that everything inheriting from it inherits that status. That registration was right in spirit and too coarse in its object. Splitting the generator about the passive baseline gives the income-minus-expenditure shape as an identity, so what is carried is not the equation but one step inside it, the linearity of the dissipation term; and that step admits a one-sided bound, so what propagates downstream is an inequality rather than an unlabelled reduction. Sections 2.4 and 4.2 accordingly grade the equation in four layers instead of one. The consequence that matters here falls in Section 4.5: the minimum maintenance power acquires a lower bound that does not pass through the stationary relation and therefore escapes the circularity that section itself identifies. The correction. The adiabatic condition was attached to the wrong object, and the failure it reports at the fossil state has a different cause. V13 wrote that an adiabatic condition must accompany the master equation because the exchange coefficient is a function of the active fraction. It need not. The stock is a pure number and the coefficient appears only in the income term, not inside the quantity being differentiated, so a state-dependent coefficient adds no term to the equation. What the condition governs is the adiabatic elimination of Section 4.4 and the pure-parameter status of the thresholds that follow from it. The second half of the correction is sharper. The reported failure near the fossil state is not an approximation breaking down; it is the assumption that the exchange coefficient factors into an efficiency and a fluctuation temperature, and that assumption is incompatible with the fossil state being a fixed point. Both cannot hold. This paper keeps the fixed point and bounds the coefficient, which is a negative result about one of the framework's own layer-B assumptions and carries more information than the limit-ordering repair it replaces (Sections 3.6 and 4.6). Completion one. The size coordinate acquires a nonlinear segment, and the empty cell of Section 9.2 is filled by the equations of this paper (Section 4.7). The causal chain of Section 8.5 has concentration at both of its ends, so within the size coordinate it is a feedback and not a one-way link. Written into the income term it is a gain factor, and the stock equation ceases to be linear in the stock. The folds satisfy a tangency condition containing no parameterization; on the simplest two-parameter family the bistability condition, the fold positions and three exact identities are closed, and one of the identities contains no parameter at all, which makes it a test of the family rather than of the mechanism. V13 recorded the jump-type size cell as an external input rather than a consequence of its own dynamics. That record is withdrawn. Completion two. The stock side acquires its noise, its boundary classification, a stationary law and a quasi-potential (new Section 4.9). V13 noted that a zero-mean shock enters only the variance and then never wrote the variance, while Section 4.6 stratifies early warning by variance and lagged autocorrelation. The multiplicative form of the three expenditures carries over to the noise, giving a square-root diffusion whose behaviour at the origin is settled by a Feller classification rather than by inspection of the drift: the two branches are separately invariant exactly when the income rate is at least the stock-side noise intensity, and below that the branch label can flip along a trajectory, which Section 3.4 says it cannot. The size coordinate thereby acquires barrier heights, well depths and a Maxwell point of its own. Completion three. The field form and the ensemble-versus-trajectory reading are written out (new Section 4.10). Table 16 already assigns a dissipation slot to the stock field and Section 8 already builds a screening length on it, while Section 4 writes only spatially aggregated scalars. The field form is given here with a stated caveat that its gradient term is leading order. And the master equation is declared to be an ensemble relation, with its trajectory-side companion written down, an interface this paper had left implicit. Completion four. An identifiability table (new Section 4.11). Section 3.6 performs one honest demotion, from a zero-parameter test to an over-determined one. That demotion was never performed for the remaining coefficients, and the calibration routes are scattered across three sections. The table collects them and states which coefficients a single experiment fixes, which are fixed only jointly, and which have no protocol anywhere — the last group being the three coefficients of the driving equation, out of which both the loop gain and the bistability parameter are built. What V13 settled, retained here unchanged. V13 made fifteen corrections and five completions to V12. They arrived in two rounds — the first from an internal audit, the second from a cross-check against the phase-transition classification and the cross-scale synchronization treatment — and are listed here in one sequence, since only V12 was ever deposited. Two of the corrections repair an internal inconsistency in the framework’s own statement of what it measures, and two more change a definition and a prediction. Each is argued where it occurs. One. Two different baselines were called the zero of one quantity. Section 2 defined structure as departure from the constrained maximum-entropy baseline, while Section 3 defines the stock as departure from the passive baseline. For a coupled system these are different laws, so the framework had two zeros for the quantity it claims to carry. The two offices are separated: the maximum-entropy baseline defines the domain, that is, when structure exists at all; the passive baseline defines the measured quantity. The two coincide on the uncoupled side and only there (Sections 2.1 and 3.4). Two. The mixed coordinate carries a domain, and its identification with the stationary size law is asymptotic rather than exact. The map φ(x) = x/s_c + (1 − r)·ln x is strictly monotone only for r ≤ 1, so the cell exists on r ∈ (0,1). And the caption of V12’s Table 13 stated that requiring a constant hazard in that coordinate gives a survival function which is exactly the law Section 4.7 derives — but Section 4.7 derives a mass function. A law whose survival function has that form and a law whose density has it are different laws; the two hazard rates agree to first order and separate at second (Sections 4.7 and 10.2). Three. The closed form of the sharp-replacement kernel was printed at zero delay. The second derivative of the log-recovery curve under a sharp threshold is −λ²e(−λ(s−t))/(1−e(−λ(s−t)))², which depends on the delay; V12 printed the expression with the delay removed and gave no numbers beside it (Section 7.5). Four. The open-loop gain and its value at full ignition are written apart. V12 defined the gain as a function of the active fraction and then used the same symbol for the pure parameter combination in the same paragraph (Section 4.4). Five. The count of independent blocks is dimension-dependent. V12 wrote that the ratio of system size to screening length is the number of roughly independent blocks; in d dimensions that number is the d-th power of the ratio. The paper is careful about dimension in Section 8.2 and was not here (Section 10.4). Six. The three-coefficient regression is exactly identified, not over-determined. Three regression coefficients determining three physical quantities is a bijection. Over-determination is available, but only if the exchange coefficient is measured elsewhere, and that qualification is now stated (Section 7.3). Seven. The minimum maintenance power cannot be evaluated from an exchange coefficient measured at the same operating point. Substituting the stationary relation for the coefficient into the expression returns the applied work rate identically, because the two are one equation read twice. V12 recorded only the weaker circularity through the active fraction (Section 4.5). Eight. A1″ was used without definition. The uniqueness of the coordinate dichotomy was said to rest on A1″, a label this paper never defined. The axiom system is now stated by reference, with every label the body uses (Section 2.4). Nine. The body carried no citations. V12 listed thirty-four references and cited none of them inline. Every borrowed result is now attributed at the point of use. Ten. Table 2 was called a reconciliation of five scalars inside a section titled three structural quantities. The canonical reconciliation carries six scalars, the sixth being the ergodicity gap of the temporal treatment; the scope of this paper’s table is now stated against it (Section 3.1). Eleven. The register of predictio