从η到Tsallis q
From η to Tsallis q
摘要
论文 III 建立了η作为非平衡参数的四条件满足性，给出convention-corrected coupling κ和四制式分类。论文 IV 解决了Tsallis统计力学的三十年悬案：q从第一性原理是什么？答案是精确插值族：q = 1 + 1/K，Boltzmann分布是K→∞的数学极限，DD resolvent是K=1的特例。
一、非平衡参数的四条件空位
平衡统计力学有一套完整的工具箱：配分函数、自由能极小化、涨落-耗散定理。一旦离开平衡，大部分工具失效。近三十年最重要的进展是Jarzynski等式和Crooks涨落定理，但这些是特定场景下的结果，不是动力学理论——它们告诉你"平均而言自由能差是对的"，不告诉你"每一步泄漏多少"。
热力学家手里有许多描述非平衡行为的量，但当我们检查哪个量同时满足四个操作性条件时，会发现一个空位：
无量纲（可跨系统比较）、跨尺度稳定（在不同粗粒化尺度下给出近似相同的值）、动力学内生（来自系统的驱动-响应结构，不是外加的统计量）、定量可测（有明确的提取协议和误差估计）。
熵产生率σ满足后两条但有量纲、依赖系统尺寸；Lyapunov指数满足前两条但测量轨道不稳定性而非耗散；FDR violation中的有效温度T_eff依赖所选观测量，不是系统内禀性质。没有一个已有量同时满足全部四条。η是填补这个空位的候选量。
二、η的Convention-Corrected Coupling与四制式
不同文献采用不同符号约定，同一个物理系统的η值可能因写法不同而翻转正负号。论文 III 定义了convention-corrected coupling κ，消除这种依赖：
κ := −s_r · Cov(Δf, Δr) / Var(Δf)
κ的值域将系统分为四类。Null制式（κ ≈ 0）：f和r的涨落几乎无关，恢复通道没有预测驱动通道的能力，Ornstein-Uhlenbeck过程属于此类。吸收制式（0 < κ < 1）：恢复通道部分对冲了驱动通道的涨落，η = 1 − κ有明确的物理含义；Brusselator化学振荡器属于此类，η ≈ 0.19–0.20。过屏蔽制式（κ > 1）：恢复力超过驱动涨落，Schlögl双稳态模型全程属于此类。反补偿制式（κ < 0）：f和r不是在补偿而是在协同，通常指示分解本身有问题。
三系统benchmark panel（OU/Brusselator/Schlögl）构成一组判别三联组，可以用来识别任何新系统属于哪个制式。Brusselator的η在整个振荡区间内为[0.14, 0.34]，均值0.20，与ZFCρ DP递推的η ≈ 0.20数值吻合——来自两个在数学和物理上毫不相关的系统。
三、Tsallis q三十年的自由参数
1988年，Tsallis提出了Boltzmann-Gibbs统计的推广，核心对象是q-exponential：
e_q(u) = [1 + (1−q)u]^{1/(1−q)}
当q = 1时回到标准指数函数。在随后三十年，q在大量物理系统中被成功拟合：LHC强子横动量谱q ≈ 1.11–1.15，太阳风q ≈ 1.71，宇宙射线q ≈ 1.2，引力系统q ≈ 2.0，黑洞热力学持续应用至今。
但q始终是自由参数。每个系统拟合出不同的q值，没有人从微观机制推导出q应该是多少。Boltzmann统计的e−βE没有自由参数——给定温度，分布完全确定。Tsallis多了一个参数q，拟合力更强了，但解释力弱了。三十年来最被批评的一点：q为什么是这个值？
四、精确插值族：q = 1 + 1/K
论文 IV 给出了q的精确插值族身份。取q = 1 + 1/K（K > 0），代入Tsallis q-exponential：
e_q(−x) = (1 + x/K)^{−K}
这是一个精确的代数恒等式。这个族的极限行为揭示了所有结构：
K = 1（q = 2）：(1+x)−1 = DD resolvent。ZFCρ DP递推的自然核，"一步反馈，不细分"。
K → ∞（q → 1）：lim_{K→∞} (1+x/K)−K = e−x = Boltzmann分布。
这不是一个近似。这是指数函数的数学定义本身。Boltzmann分布是"无穷细分反馈"的极限：一步反馈被细分为无穷多个微步，每个微步的反馈量趋零，总效果是平滑的指数衰减。微积分的连续极限恰好是这个操作的数学实现。
因此，整个Tsallis q族是一个由K参数化的精确插值族，连接了两端：DD resolvent（离散、单步、一层反馈）和Boltzmann（连续、无穷细分、无限层反馈）。K的物理含义是有效反馈阶数。
五、通道平均屏蔽律猜想
在插值族身份之外，论文 IV 提出一个结构性猜想：在f/r双通道系统中，
q = Ω_eff / n_ch
其中Ω_eff是有效壳层深度（对应ZFCρ中的素因子个数Ω），n_ch是通道数。双通道时n_ch = 2，q = Ω_eff/2。
这个猜想有精确的端点锚定：Ω_eff = 2时q = 1（Boltzmann端），Ω_eff = 4时q = 2（DD resolvent端）。它给出了q ≥ 1的结构原因：至少需要n_ch个素因子才能在各通道之间分配屏蔽层。它还给出了实验q值的物理解读：太阳风q ≈ 1.7对应Ω_eff ≈ 3.4，靠近Boltzmann端但有非平凡的层深度；引力系统q ≈ 2.0对应Ω_eff = 4，DD resolvent端。
通道平均屏蔽律是结构性驱动的猜想，不是定理。插值族身份本身是精确的代数结果，不需要猜想状态。
Abstract
Paper III establishes η's satisfaction of four operational criteria for a nonequilibrium parameter, introducing convention-corrected coupling κ and a four-regime classification. Paper IV resolves a thirty-year puzzle in Tsallis statistical mechanics: what is q from first principles? The answer is an exact interpolation family: q = 1 + 1/K, with the Boltzmann distribution as the K → ∞ mathematical limit and the DD resolvent as the K = 1 special case.
I. A Four-Criterion Gap in Nonequilibrium Parameters
Equilibrium statistical mechanics has a complete toolkit: partition functions, free-energy minimization, the fluctuation-dissipation theorem. Away from equilibrium, most of these tools fail. The most important advances of the past thirty years — Jarzynski equality, Crooks fluctuation theorem — give results for specific scenarios, not dynamical theories. They say "on average the free-energy difference is correct," not "how much leaks per step."
Many quantities describe nonequilibrium behavior. But when we ask which simultaneously satisfies four operational criteria, we find a gap:
Dimensionless (comparable across systems), approximately cross-scale stable (similar values at different coarse-graining scales), dynamically endogenous (arising from the system's drive-response structure, not an externally imposed statistic), and operationally measurable (with an explicit extraction protocol and error estimate).
Entropy production rate σ satisfies the last two but carries dimensions and depends on system size. The Lyapunov exponent satisfies the first two but measures orbital instability, not dissipation. The effective temperature T_eff from FDR violation depends on which observable is chosen and is not an intrinsic system property. No existing quantity satisfies all four simultaneously. η is the candidate to fill this gap.
II. Convention-Corrected Coupling and Four Regimes
Different literature conventions use different sign choices; the same physical system's η value can flip sign depending purely on how the equations are written. Paper III defines the convention-corrected coupling κ to eliminate this dependence:
κ := −s_r · Cov(Δf, Δr) / Var(Δf)
κ's range of values classifies systems into four regimes. Null (κ ≈ 0): f and r fluctuations are nearly uncoupled; the restoring channel has no predictive power over the driving channel. The Ornstein-Uhlenbeck process belongs here. Absorptive (0 < κ < 1): the restoring channel partially offsets driving fluctuations; η = 1 − κ carries a clear physical meaning. The Brusselator chemical oscillator belongs here, with η ≈ 0.19–0.20. Overscreening (κ > 1): the restoring force exceeds the driving fluctuations; the Schlögl bistable model is here throughout its parameter space. Anti-compensatory (κ < 0): f and r co-amplify rather than compensate, usually indicating a misidentified decomposition.
The three-system benchmark panel (OU/Brusselator/Schlögl) forms a discriminating triad for classifying any new system. The Brusselator's η ranges over [0.14, 0.34] across the oscillatory regime, mean 0.20 — coinciding numerically with the ZFCρ DP recursion's η ≈ 0.20, despite the two systems being mathematically and physically unrelated.
III. Thirty Years of Free Parameters
In 1988, Tsallis proposed a generalization of Boltzmann-Gibbs statistics. The central object is the q-exponential, which recovers the standard exponential at q = 1 and generates power-law tails otherwise. Over three decades, q was successfully fitted across physical systems: LHC hadron transverse-momentum spectra yield q ≈ 1.11–1.15, solar wind q ≈ 1.71, cosmic rays q ≈ 1.2, gravitational systems q ≈ 2.0, black hole thermodynamics through today.
Yet q has always been a free parameter. Each system is fitted independently; no microscopic mechanism has predicted which q a given system should exhibit. The Boltzmann distribution e−βE has no free parameters — given temperature, the distribution is fully determined. Tsallis gains fitting power by adding q, but loses explanatory power. The longest-standing criticism: why is q the value it is?
IV. The Exact Interpolation Family: q = 1 + 1/K
Paper IV gives q its exact interpolation family identity. Set q = 1 + 1/K (K > 0) and substitute into the Tsallis q-exponential:
e_q(−x) = (1 + x/K)^{−K}
This is an exact algebraic identity. The limiting behavior of this family reveals its full structure.
At K = 1 (q = 2): (1+x)−1 = the DD resolvent. The natural kernel of the ZFCρ DP recursion: "one-step feedback, no subdivision."
As K → ∞ (q → 1): lim_{K→∞} (1+x/K)−K = e−x = the Boltzmann distribution.
This limit is not an approximation. It is the mathematical definition of the exponential function itself. The Boltzmann distribution is the limit of "infinitely subdivided feedback": one step of feedback divided into infinitely many micro-steps, each with vanishing feedback, whose total effect is smooth exponential decay. The continuous limit of calculus is precisely this operation made rigorous.
The entire Tsallis q-family is thus an exact interpolation family parameterized by K, connecting two poles: the DD resolvent (discrete, single-step, one layer of feedback) and Boltzmann (continuous, infinitely subdivided, infinite layers). K's physical interpretation is the effective feedback order.
V. The Channel-Averaged Shielding Conjecture
Beyond the interpolation family identity, Paper IV proposes a structural conjecture: in an f/r dual-channel system,
q = Ω_eff / n_ch
where Ω_eff is the effective shell depth (the physical generalization of ZFCρ's prime-factor count Ω) and n_ch is the number of channels. For dual channels, n_ch = 2, giving q = Ω_eff / 2.
The conjecture has exact endpoint anchoring: Ω_eff = 2 gives q = 1 (the Boltzmann end), Ω_eff = 4 gives q = 2 (the DD resolvent end). It provides a structural explanation for q ≥ 1: at least n_ch prime factors are needed to distribute shielding across channels. And it gives physical readings of observed q values: solar wind q ≈ 1.7 corresponds to Ω_eff ≈ 3.4, near but not at the Boltzmann limit; gravitational systems at q ≈ 2.0 sit at the resolvent end.
The channel-averaged shielding law is a structurally motivated conjecture, not a theorem. The interpolation family identity itself is an exact algebraic result requiring no conjecture status.