Form and Flow P1 — On the Hypotheses Which Lie at the Foundations of Geometry (Part II): A Contemporary Continuation of Riemann's Methodology and the Recognition of Geometric Remainders
形与流 P1 — 论几何学所基于的假设（之二）：黎曼方法论的当代接续与几何余项的承认

A Contemporary Continuation of Riemann's Methodology and the Recognition of Geometric Remainders

(working title pending)

---

Abstract

This paper continues the methodological legacy of Riemann's 1854 inaugural lecture Über die Hypothesen, welche der Geometrie zu Grunde liegen by performing an explicit assumption review of Riemannian geometry from within the Self-as-an-End (SAE) framework. The paper does not aim to construct an axiomatic SAE-geometry; rather, as the founding paper of the "Form and Flow" (形与流) series, it provides foundational vocabulary and ontological anchors for the subsequent treatment of geometry and dynamics across multiple D-levels (4D physics, 5D-6D evolutionary dynamics, 7D-8D population dynamics, 9D+ subject dynamics). Two core propositions are advanced: first, the Gauss-Bonnet formula is reread as an exact exchange law among four geometric remainders on the manifold, with a falsifiable formal statement under standard normalization; second, the Hamilton-Perelman Ricci flow program is reread as the failure of resolution within the smooth-manifold category and the forced ascent of geometric remainder to a higher category—an instance, on the geometric side, of the SAE H'-type indissolvability theorem. A meta-level proposition is advanced: unification does not eliminate remainder; it migrates remainder—offering an ontological identification of the phenomenon that the geometric domain cannot be wholly enclosed by any single framework. On this basis, the paper operationalizes these ontological propositions into a set of reusable diagnostic tools for working geometers: the ρ-defect ledger, ρ-signature, and readout-first principle. Within this diagnostic framework, three SAE-perspective applications are proposed—as conjectural applications and open research problems, not on equal footing with the two core propositions: a quantization conjecture for Perelman's W-functional jumps in Ricci flow surgery, a dissipation-duality conjecture between geometric flows and information-geometric phase transitions, and a rereading of abnormal geodesics in sub-Riemannian geometry as topological breakthroughs of ρ. These three are concrete research invitations from the SAE perspective; their formalization and falsification are left to the geometric community. The paper explicitly reviews the five $\rho = \varnothing$ assumptions inherited from Riemannian geometry (smoothness, integer dimension, tensor metric, real-number background, point-set substrate), replaces the first three, and explicitly leaves the latter two as boundaries the present generation cannot move, to be passed to successors. The methodological posture itself—identifying assumptions, replacing some, explicitly acknowledging the unreplaced—constitutes embodied evidence for the SAE core ontological proposition that remainder is irreducible.

Keywords: Riemannian geometry; Gauss-Bonnet formula; Ricci flow; irreducibility of remainder; geometric H'; ρ-defect ledger; Perelman W-functional; information geometry; assumption-loosening partial order; Self-as-an-End framework

---

1. Introduction: Continuing Riemann's Methodology

1.1 Riemann's 1854 Inaugural Lecture and Its Methodological Legacy

On June 10, 1854, Riemann delivered his inaugural lecture at Göttingen, titled Über die Hypothesen, welche der Geometrie zu Grunde liegen [1]. The audience consisted primarily of Gauss; the text was published only posthumously, in 1868. This brief document has since been recognized in nearly every historical narrative as the starting point of modern geometry.

What it accomplished was, in essence, a single act: identifying the assumptions on which geometry rests, and proposing the replacement of one of them. The replaced assumption was the flatness of space—taken as self-evident in Euclidean geometry, demoted by Riemann to a special case, and superseded by the more general notion of variable metric. This replacement opened an entire workspace of non-Euclidean geometries, subsequently extended by Ricci, Levi-Civita, Einstein, and continued today across various directions: Finsler geometry, sub-Riemannian geometry, Lorentzian geometry, fractal geometry, noncommutative geometry, p-adic geometry, tropical geometry, information geometry.

The reason this lecture deserves to be reread, however, is not solely the specific assumption Riemann replaced, but rather the methodology it exemplified: geometry is not a completed edifice but a set of assumptions perpetually open to renewed scrutiny. The work of each generation of geometers consists not in adding bricks to that edifice, but first in identifying what the previous generation took for granted, why it was taken for granted, whether it remains necessary—and then, where replacement is possible, in replacing; where replacement is not possible, in explicitly acknowledging and passing the assumption forward. Riemann's own term was Hypothesen—hypotheses—and he was acutely aware that his work was not the unveiling of ultimate truth, but the identification of assumptions and the replacement of some among them.

1.2 The Position of the Present Paper

The present paper continues this lineage. The choice to title it as a sequel to Riemann's lecture is not a claim to comparability with Riemann himself but rather an act of explicitly embedding the present work within that methodological lineage: Riemann identified one assumption and replaced it; subsequent generations of geometers each identified other assumptions and replaced several; the present paper continues this same act, performing one more round of assumption review from the SAE perspective, replacing those assumptions that can be replaced, and explicitly acknowledging those that the present generation cannot move, to be passed to successors. The designation "Part II" signals not equivalence in workload to Riemann's original, but a refusal to subsume all post-Riemannian geometric work under "Part I"—an attempt to reactivate the act of assumption review itself as a methodological continuation, leaving room for "Part III," "Part IV," and beyond.

1.3 A Retrospective Acknowledgment Concerning the SAE Framework and the "Recognition of ρ" Reading

It must be stated explicitly at the outset: the framework of "recognizing ρ" employed in this paper is the SAE framework's retrospective reading of geometric history from its own ontological standpoint. It is not a vocabulary Riemann himself used, nor a position Riemann himself held [2,3]. One of Riemann's actual concerns in 1854 was the physical applicability of geometry—he proposed in the closing section of the lecture that the metric relations of space ought to be determined by the binding forces acting upon it, a thought that later directly inspired Einstein's general relativity. This is an insight coupling mathematical geometry with physical dynamics, akin in spirit to the SAE proposition of "recognizing remainder," but not identifiable with it.

What the present paper does is not to impose a modern framework upon Riemann, but to acknowledge that the SAE retrospective reading is a directional, positioned reinterpretation: SAE perceives a particular face of Riemann's work and names it "the recognition of geometric remainder," but this naming is the work of SAE, and Riemann is not responsible for it.

This explicit acknowledgment matters because it determines the paper's posture. The present paper does not stand outside Riemann to declare that "Riemann recognized ρ but not enough," nor does it stand above Riemann to declare that "SAE has completed Riemann's unfinished project." It stands within the methodological lineage of identifying assumptions, doing the same kind of work Riemann did: identifying the assumptions still defaulted in our generation, replacing the few that can be replaced, explicitly acknowledging the few that cannot. This is a continuation, not a transcendence.

1.4 The "Form and Flow" Series and What This Paper Carries

This paper carries the founding-paper task of the "Form and Flow" (形与流) series. "Form and Flow" is an independent SAE series treating the geometry-and-dynamics facet of the SAE program across D-levels: the 2D mathematical archetype of manifolds (the present paper); the 4D return of continuous fields in physics (subsequent P2, fluid dynamics); 5D-6D evolutionary dynamics (P3); 7D-8D population dynamics (P4); 9D+ subject dynamics (P5); causal-slot operators as smoothing functors (P6); and a concluding synthesis paper (P0). The series shares its framework with the existing SAE Physics, SAE Information Theory, SAE Biology Notes, SAE Moral Law, and Human Total Construct series, as a parallel facet of the same ontological program rather than a hierarchically related sub-program.

As the founding paper, the present work carries five distinct tasks:

(i) Locating Riemann within the methodological genealogy of "recognizing geometric remainder," with explicit acknowledgment that this is a SAE retrospective reading;

(ii) Establishing two core propositions—the Gauss-Bonnet formula as a four-fold remainder exchange law (§2), and the Ricci flow program as a smooth-manifold-category resolution failure with categorial ascent (§3);

(iii) Establishing geometric vocabulary repeatedly invoked in subsequent P2-P5—metric, connection, curvature, geodesic, singularity, conservation as reread under the SAE perspective (§4);

(iv) Explicitly enumerating those assumptions the present generation cannot replace, articulating why, and proposing a SAE meta-proposition (§5);

(v) Operationalizing the foregoing ontological propositions into reusable diagnostic tools for working geometers, with three concrete SAE-perspective applications, so that the paper's contribution is immediately usable in geometric research practice (§6).

What the paper explicitly does not undertake includes: axiomatization of SAE-geometry, category-theoretic formalization, the systematic treatment of complex / Kähler geometry, the SAE rereading of spectral geometry and index theory. Each is a substantial independent project, and the present paper does not attempt to encroach upon them.

---

2. Proposition A: Gauss-Bonnet as a Four-Fold Remainder Exchange Law

2.1 The Formula and the Standard Reading

One of the oldest and most perfect identities of two-dimensional geometry is the Gauss-Bonnet formula [4,5]. For a compact, orientable two-dimensional Riemannian surface $M$ with smooth boundary $\partial M$:

$$\int_M K \, dA + \int_{\partial M} k_g \, ds = 2\pi \chi(M).$$

Here $K$ is the Gaussian curvature, $k_g$ the geodesic curvature on the boundary, $\chi(M)$ the Euler characteristic of the manifold, and $dA$, $ds$ the area form and arclength form respectively. For piecewise-smooth boundary, exterior-angle terms are added on the right; for clarity of exposition, the present paper treats only the smooth-boundary case, with the SAE rereading extending directly to the more general case.

The standard reading of this formula is that it relates local curvature (a differential-geometric object) to global topology (an algebraic-topological object) through the transcendental constant $2\pi$. The present paper does not aim to replace this standard reading but to perform an SAE rereading of the formula, identifying the ontological structure it expresses.

2.2 The Ontological Identification of Four Remainders

Proposition 2.1 (Gauss-Bonnet as four-fold remainder exchange law). In the SAE perspective, the Gauss-Bonnet formula expresses an exact exchange law among four distinct remainders:

$$\frac{\text{interior curvature remainder} + \text{boundary remainder}}{\text{Archimedean circular unit } 2\pi} = \text{topological integer remainder}.$$

Each of the four quantities corresponds to a distinct facet of geometric remainder; the formula provides the precise exchange relation among them.

The first is the interior curvature remainder, corresponding to $\int_M K \, dA$. This term integrates local curvature over the entire manifold, representing the ontological resistance against flatness manifested at every point: for Euclidean space the term vanishes identically; for any nonflat space it surfaces with nonzero integral. This is precisely the remainder Riemann recognized: Euclidean geometry assumed flatness; Riemann identified the failure of this assumption and made variable metric the new working platform. As the archetype of geometric remainder, curvature is recorded with integral precision in this formula.

The second is the boundary remainder, corresponding to $\int_{\partial M} k_g \, ds$. Its presence indicates that, when the manifold is not closed, a portion of what would otherwise belong to the interior curvature remainder "spills over" onto the boundary, manifesting as the forced bending of geodesics. The boundary remainder and the interior remainder are the same ontological resistance manifesting differently—one inside, one on the edge—not two independent quantities. The Gauss-Bonnet formula expresses the legitimacy of continuous transfer between them: for the same manifold, decreasing interior curvature increases boundary curvature, and vice versa. This legitimate interior-boundary exchange is the ontological reason why the geodesic-curvature term appears as an independent additive term in the formula.

The third is the Archimedean circular unit, the constant $2\pi$ on the right-hand side. The ontological status of this term requires careful handling. The constant $2\pi$ is not an accident of the formula, but neither is it the ontological core of the Gauss-Bonnet theorem—this distinction must be made explicit. The Gauss-Bonnet formula admits a renormalized form $\int_M e(TM) = \chi(M)$, where $e(TM)$ is the Chern-Weil representative of the Euler class, expressed under the standard Chern-Weil normalization as $e(\Omega) = (2\pi)^{-n} \mathrm{Pf}(\Omega)$ [6]. The implication is that $2\pi$ is a normalization constant: it is "the exchange factor through which the real-valued reading of curvature must pass to enter the integer-valued reading of topology, under the standard Archimedean real-geometric normalization."

The role of $2\pi$ is not mystical but irreducible: it is the inevitable exchange rate between continuous real geometry and integer topology. The remainder appears specifically as $2\pi$ because we have adopted standard radian-angle and standard area-form conventions; under different normalization conventions, the rate appears as $\pi$, $4\pi$, or $(2\pi)^k$, but its ontological role as "real/integer exchange factor" is invariant. From the SAE perspective, this exchange factor is the geometric manifestation of arithmetic remainder—it cannot be eliminated, only renormalized.

The fourth is the topological integer remainder, corresponding to $\chi(M)$ on the right-hand side. The Euler characteristic compresses the manifold's full topological information—genus, connectivity, hole structure—into a single integer. This integer is discrete, in sharp contrast to the continuous curvature integral. Yet the Gauss-Bonnet formula asserts that, after multiplication by an Archimedean exchange factor, the two sides are equal. In other words: the total continuous geometric remainder, after passing through the standard Archimedean normalization, equals exactly the discrete integer reading of topological remainder.

This exchange law is not a new mathematical result—the formula was completed by Bonnet in 1848 [4] and extended to higher dimensions by Chern in 1944 [5]. Rather, it is an ontological rereading of an established mathematical result. The substantive contribution lies in identifying the remainder-status of each quantity in the formula, treating the entire identity as evidence for the legitimacy of precise exchanges among remainders of different facets.

2.3 A Falsifiable Formal Statement

To prevent Proposition 2.1 from remaining at the level of ontological exegesis alone, a falsifiable formal statement is given.

Proposition 2.2 (Standard-normalization version, equivalent to Proposition A1). Suppose the standard Archimedean normalization convention is adopted: standard radian angles, standard area form, and standard Gaussian curvature induced from the Euclidean plane (henceforth "standard convention"). Under this convention, if there exists a constant $a \in \overline{\mathbb{Q}}$ (i.e., an algebraic number) such that

$$\int_M K \, dA = a \cdot \chi(M)$$

holds for all compact orientable closed Riemannian surfaces $M$, a contradiction follows.

Proof. Take the unit sphere $S^2$, on which Gaussian curvature satisfies $K \equiv 1$, hence $\int_{S^2} K \, dA = 4\pi$, while $\chi(S^2) = 2$. Substituting into the assumed identity yields $4\pi = 2a$, i.e., $a = 2\pi$. If $a$ were algebraic, then $\pi = a/2$ would also be algebraic, contradicting the Lindemann theorem [7] ($\pi$ is transcendental). $\square$

On the boundary of this proposition: the qualifier "standard convention" cannot be omitted. Under non-standard normalization (e.g., redefining angle units so that $2\pi$ corresponds to an algebraic multiple, or redefining the area form to absorb the $2\pi$ factor), both sides of the identity may be rational or even integer—this is a legitimate renormalization, not a counterexample to Proposition 2.2. What the proposition asserts is: under the standard Archimedean real-geometric normalization, arithmetic remainder must manifest precisely in the form of the transcendental constant $2\pi$. Surrendering the standard convention can relocate this manifestation, but the relocation must in the new convention explicitly absorb the arithmetic remainder—it cannot be eliminated, only repositioned. This is precisely the manifestation of meta-proposition 5.3 ("unification is migration") on the arithmetic-remainder facet.

2.4 Higher-Dimensional Generalizations, the Exchange Law Family, and Period Theory

The exchange law is not a coincidence of two-dimensional surfaces. Chern's 1944 generalization [5] yields the $2n$-dimensional Chern-Gauss-Bonnet formula:

$$\int_M \mathrm{Pf}(\Omega) = (2\pi)^n \chi(M),$$

or equivalently, under standard Chern-Weil normalization:

$$\int_M \frac{1}{(2\pi)^n} \mathrm{Pf}(\Omega) = \chi(M).$$

Thus, as the manifold dimension increases, the exchange factor between real numbers and integers is an integer power of $2\pi$. Concomitantly, the normalizations of higher-dimensional characteristic classes—Pontryagin, Chern, Euler—exhibit factors of $(2\pi)^k$ and $(2\pi i)^k$ throughout [6]. These normalization factors form a complete exchange law family.

Observation 2.3 (Exchange law family). Real- or complex-valued continuous readings in the world of differential forms must, in order to enter integer-valued cohomology, pay a normalization cost of the form $(2\pi i)^k$.

The ontological identity of this cost is the geometric manifestation of arithmetic remainder, appearing in different dimensions as different powers of $\pi$.

A formal clarification is required here, lest the SAE perspective on the exchange law be read with excessive strength. To predict that all "continuous-to-discrete" geometric identities must contain transcendental factors is an oversimplification: normalization conventions can be chosen, and under particular conventions both sides of an identity may be rational or even integer. The appearance of $2\pi$ in the Gauss-Bonnet formula depends on the standard Archimedean real-geometric normalization—it is not an intrinsic invariant of the geometric structure itself. A more reliable ontological identification places the locus at the comparison coefficient between de Rham cohomology and Betti cohomology—that is, at periods. The period theory of Kontsevich and Zagier [8] systematizes this class of constants—including $2\pi i$, special values of the $\zeta$-function, algebraic periods—into a number-theoretic family deeply coupled to algebraic geometry. From the SAE perspective, periods are the standard manifestation locus of arithmetic remainder in geometry: they do not depend on convention choice but are invariants of the geometric structure itself.

This observation is formalized as a directional statement.

Pointing Proposition 2.4 (Periods as the standard manifestation of arithmetic remainder). Let $X$ be an algebraic variety defined over $\mathbb{Q}$, $\omega$ an algebraic differential form on $X$, and $\gamma$ a rational homology cycle in $X(\mathbb{C})$. The period $\int_\gamma \omega$ provides the exchange between the algebraic side (geometric structure) and the arithmetic side (transcendental or algebraic numbers). From the SAE perspective, the field to which a period belongs (transcendental, algebraic, rational) precisely identifies the arithmetic-remainder level of the corresponding geometric structure. Period identities thus form a remainder-conservation family across algebraic geometry and number theory. The constant $2\pi$ in Gauss-Bonnet is the manifestation of this conservation family in its simplest case (real manifold / Euler-class integration).

This proposition is offered as a directional statement; its formalization is not undertaken here. The SAE rereading of period theory itself—the appropriate categorical structure, the ρ-defect-ledger form on the space of periods, the precise duality between arithmetic and geometric remainders in the period sense—is an independent project, left for future SAE Spectral Geometry or SAE Arithmetic Geometry treatments. The $2\pi$ phenomena discussed in Propositions 2.1, 2.2, and Observation 2.3 of the present paper should be understood as manifestations of period theory in its simplest setting.

Deeper remainder phenomena in spectral geometry—the $\pi^k$ terms in the Selberg trace formula, the normalization constants in heat-kernel expansion coefficients, the appearance of $\zeta$-function special values in index theorems and analytic torsion [9]—can be regarded as spectral-form extensions of the exchange law family. These phenomena point toward a broader "spectral-geometry remainder-conservation family," not developed here. One open observation worth recording: $\zeta(2k)$ is a rational multiple of $\pi^{2k}$, while the arithmetic nature of $\zeta(2k+1)$ remains undetermined (Apéry [10] proved only the irrationality of $\zeta(3)$ in 1978; its transcendence remains open). This perhaps points to an as-yet-unrecognized deeper level of arithmetic remainder in geometry. The observation is left for future work.

---

3. Proposition B: Ricci Flow and the Geometric H'-Type Categorial Ascent

3.1 From Static Bookkeeping to Dynamical Conservation

If Proposition A in §2 treats the static bookkeeping of geometric remainder—exact exchanges among curvature, boundary, Archimedean factor, and integer topology—then Proposition B in this section treats the dynamical conservation of geometric remainder. The claim: geometric remainder is irreducible not only statically but also dynamically. Any attempt to "smooth away" curvature by means of a geometric flow will, within finite time, encounter the rebound of remainder; this rebound is to be read, in the SAE perspective, not as the failure of an elimination attempt, but as the recognition of remainder by a higher-categorial structure into which it has ascended. The Ricci flow program is the most cogent contemporary case for this proposition.

3.2 The Ricci Flow Program: Brief Overview

Ricci flow was introduced by Hamilton in 1982 [11], defined by

$$\frac{\partial g(t)}{\partial t} = -2 \, \mathrm{Ric}(g(t)).$$

The geometric intuition is to evolve the metric along the negative Ricci-curvature direction—shrinking high-curvature regions, expanding low-curvature ones—so that the manifold trends toward "more uniformly curved" states. Under certain ideal initial conditions, this evolution drives the manifold to an Einstein metric (i.e., $\mathrm{Ric}(g) = \lambda g$). Hamilton's foundational 1982 theorem proved that any compact 3-manifold with positive Ricci curvature in the initial metric flows under normalized Ricci flow to a constant-curvature metric. This result presented the geometric community with an attractive program: perhaps any manifold's geometry could be "smoothed" via Ricci flow, with its underlying topology read off from the limit.

But this program quickly encountered the rebound of geometric remainder. Under more general initial conditions, Ricci flow does not deliver the manifold into a smooth limit state but rather, within finite time, forms singularities—regions where curvature diverges and the manifold locally "necks" into a measure-zero point. These singularities cannot be processed within the original smooth-manifold category; the smooth evolution breaks down here. Perelman's work of 2002–2003 [12,13,14] showed that these singularities are not accidental failures but structured, classifiable, controlled: their local models (such as neckpinches and degenerate neckpinches) form a countable list of standard forms, and each singularity form precisely encodes local topological information about the manifold. Perelman thereby designed the "singular surgery" procedure—at the moment of singularity formation, excising the singular neighborhood, repairing the opening, and continuing the flow—and ultimately, through this procedure, proved the Poincaré conjecture and the Thurston geometrization conjecture.

3.3 The Geometric H'-Type Reading

Before stating the proposition, a brief clarification of the term "H'" is in order. "H'" is the proposition-type symbol established within the ZFCρ trunk of the present series [17,18], referring generically to a class of indissolvability theorems—their abstract content being that, for an original category $\mathcal{C}$ and a family of resolution operations $R$, the residue $\rho \neq 0$ cannot be entirely eliminated. Specific H'-type propositions admit different instantiations across domains (set theory, geometry, formal systems, dynamical systems), but the abstract pattern is identical (formalized in §3.6). The present section identifies the phenomenon arising in the Ricci flow program, on the geometric side, as a geometric H'-type proposition: namely, that under the "smoothing-resolution" operation within the smooth-manifold category, geometric remainder cannot be eliminated and can only be lifted into a more general category for renewed recognition. Readers unfamiliar with the ZFCρ series may simply read "geometric H'" as "an indissolvability theorem within the geometric category"; this section is self-contained at that level.

This is one of the great achievements in the history of geometric analysis. From the SAE perspective, the ontological significance of this achievement can be precisely identified as a H'-type indissolvability theorem.

Proposition 3.1 (Ricci flow as geometric H'-type categorial ascent). Smooth Ricci flow cannot provide a singularity-free global smooth resolution process for arbitrary initial geometry. In general, geometric remainder must, within finite time, manifest anew as singularities; for the resolution process to continue, it must ascend into a more general object category beyond the original smooth-manifold category—a category encompassing singular spacetimes, surgeries, weak flows, and topological jumps. In this ascended category, geometric remainder is recognized again, in new form, and continues to be processed.

The ontological substance of this proposition is: geometric remainder is not eliminated; it is forced to ascend a category. The failure of resolution within the original smooth-manifold category does not mean the remainder has been removed—on the contrary, it manifests in sharper form (as singularities) and forces the observer into a more general category to handle it. Perelman's singular surgery is not "the elimination of remainder" but "the lifting of remainder from the smooth-manifold level to the surgery-bearing singular-spacetime level, where it is recognized at the new level."

A precision of technical formulation is required. The phrase "geometric remainder must, within finite time, manifest anew as singularities" in Proposition 3.1 refers to the general case, not literally to every non-Einstein manifold. Ricci flow exhibits considerable behavioral diversity across dimensions, normalizations, and curvature signs: two-dimensional Ricci flow's long-time behavior classifies according to Euler characteristic, while negatively-curved Einstein metrics under unnormalized flow expand, and so on. The accurate technical formulation should be: Ricci flow cannot provide for general initial geometry a global resolution process that remains within the original smooth-manifold category and introduces no additional structure. Einstein metrics and Ricci solitons (self-similar solutions modulo diffeomorphism and scaling) are the "fixed points" of Ricci flow in this sense, but the submanifold of these fixed points does not cover general initial conditions—in general, the flow must leave the original smooth category.

3.4 Contemporary Research Directions Following the Ascent, and an Open Prediction

This ascent has, in the contemporary research landscape, given rise to an independent subfield. Following Perelman's singular-surgery program, geometric analysts have further developed "Ricci flow through singularities" theory—constructing more general objects defined even at singular times, allowing the manifold to traverse singularities continuously without manual excision [15,16]. The specific work of these directions is not the focus of this section (with concrete applications addressed in §6.5); their existence supports the SAE ascent reading: generalized flows can exist, but only with the introduction of additional objects and structures (singular spacetimes, measure-weak structures, topological jumps). In other words, the irreducibility of geometric remainder does not mean "any treatment fails," but rather "any treatment must, in some manner, ascend into a more general category." This is a stronger, more precise, and more contemporary-research-consistent proposition than "necessary failure."

A prediction compatible with the present research landscape may then be offered:

Conjecture 3.2 (Original-category irreducibility prediction). There does not exist a global Ricci-flow resolution theory that remains entirely within the original smooth-manifold category and introduces no additional structure (singular spacetimes, surgeries, weak flows, topological jumps).

This conjecture admits future falsification: if one constructs a global resolution theory operating only within the original smooth-manifold category, the SAE ascent proposition will require revision.

3.5 Structural Analogy with ZFCρ H'

Proposition 3.1 stands in structural analogy with the H'-type proposition in the ZFCρ trunk of the present series [17,18]. ZFCρ H' addresses the set-theoretic facet: any attempt to "close" set theory by adding large-cardinal axioms re-exposes itself, after the new axiom is added, in the form of a fresh structural gap—remainder is not eliminated, only pushed to a higher level. The geometric phenomenon described by Proposition 3.1 shares this abstract pattern: any attempt to "smooth out" curvature via geometric flow re-exposes itself, within finite time, as singularities—remainder is not eliminated, only pushed into a more general category. The shared abstract pattern can be expressed as:

Abstract Pattern 3.3 (H'-type indissolvability pattern). Given an object category $\mathcal{C}$, a resolution operation $R$, a structural gap $\Phi$ that resolution attempts to address, and the residue $\rho(R^t X)$ following $R$'s action: the H'-type proposition asserts that for every admissible resolution operation $R$, the residue $\rho \neq 0$ cannot be entirely eliminated.

In specific instantiations across domains, $\mathcal{C}$ may be a model of set theory, a smooth manifold, a formal system, or a dynamical system; $R$ may be axiom extension, geometric flow, formalization, or phase-space evolution; but the abstract pattern is the same.

The present paper identifies this as a "structural analogy," not a "category-theoretically rigorous isomorphism"—the latter would require the explicit specification of categories, functors, and natural transformations under which the two H' instances become two presentations of the same object. Such formalization is an independent project, not undertaken here. What is claimed here is that the geometric H' and the ZFCρ H' share the same abstract pattern; the formal redemption of this shared pattern is left as a commitment, to be discharged in the series's concluding paper P0.

3.6 The Family of Remainder-Persistence Phenomena: Other Antecedents

It is worth noting in passing that the abstract pattern "resolution attempt → ascent of remainder" has antecedents across multiple domains of mathematics. Gödel's incompleteness theorems [19], Tarski's undefinability of truth, and Turing's undecidability all describe the remainder a formal system encounters when attempting to fully internalize its own truth, provability, or decidability. The KAM theorem [20] addresses the fate of invariant tori in non-integrable Hamiltonian systems under small perturbation, where Diophantine conditions on the arithmetic side compel most tori to persist while some are torn apart. Bifurcation theory and catastrophe theory identify degenerate points not as pathologies but as classification objects. Morse theory reads critical points as structural information. Causal-set theory reads continuous spacetime as a macroscopic approximation of an underlying discrete causal structure, where the discrete-sprinkling dynamics corresponding to a geometric manifold has smooth transitions and topological jumps that map naturally onto, on the continuous side, singularity formation and categorial ascent. These are not the same theorem, but they may be jointly identified as a "family of remainder-persistence phenomena"—the same ontological pattern manifesting repeatedly across different mathematical subdomains. The unified articulation of this family is not undertaken here, but is listed as a direction for future work in SAE-Geometry and SAE-Mathematics.

---

4. Geometric Vocabulary and Information Geometry

4.1 Geometric Concepts Reread Under SAE

The first two sections established the two core propositions. This section provides the geometric vocabulary repeatedly invoked in the subsequent papers of the "Form and Flow" series. The objects treated by P2 through P5—fluids, evolutionary dynamics, population dynamics, social subjects—are not internal to the discipline of geometry, yet each can be geometrized over an appropriate state space. To make geometric vocabulary applicable to such non-traditional objects, several basic concepts must first be liberated from their purely formal definitions and granted ontological readings under the SAE perspective.

Metric, traditionally read as "the distance between two points," should be reread under SAE as "the cost structure through which difference is read out at a particular level." A metric on physical space measures length; a metric on phase space may measure dynamical distance; a metric on a statistical manifold measures distinguishability of distributions; a metric on belief space measures the cost of belief change. Each subsequent D-level has its own metric object, but all metrics share the same ontological identity: metric is the mode through which remainder, at that level, is manifested as cost.

Connection, traditionally read as "the rule of parallel transport," should be reread under SAE as "the rule for comparing the same object across positions." On physical space, connection tells how to parallel-transport a vector; on the evolutionary state space, connection tells how to compare a phenotypic attribute across genotypes; on the contextual space of subject dynamics, connection tells how to identify "the same stance" across contexts. Each subsequent level must explicitly construct a connection object, and the chosen form of connection at each level reveals the level's distinctive character.

Curvature, under the SAE perspective, should be read as "the cumulative reading of local closure failure." This is the reading already at work in §2 and §3: curvature is not "the degree of bending" in geometric intuition, but "the degree to which the attempt to return to the starting point along a closed path fails." At each subsequent level, wherever there is some "local cycle failure" phenomenon, there is a corresponding curvature object. Vorticity in fluid mechanics is the curvature of the velocity field; "fitness-cycle non-closure" in evolutionary dynamics is the curvature of the fitness landscape; "belief-loop inconsistency" in subject dynamics is some kind of curvature.

Geodesic, under SAE, should be read as "the natural path under a given cost structure." The traditional reading equating geodesic with "shortest path" is excessively narrow: in Lorentzian geometry, geodesics may be "longest" (timelike geodesics maximize proper time); in information geometry, geodesics correspond to a kind of "most natural transformation" within a distribution family. The present paper reads geodesic generally as "the natural evolutionary path under local closure conditions," a reading applicable across subsequent levels.

Singularity: the SAE reading was given in Proposition 3.1: a singularity is a place where the current smooth description can no longer absorb remainder, the locus of the forced ascent of remainder. This reading recurs across subsequent levels: turbulent bursts in fluid mechanics are singularities; species-extinction events in evolution are singularities; phase-transition jumps in population dynamics are singularities; paradigm shifts in subject dynamics are singularities. None of these is "system failure"; each is "the recognition point for remainder ascending into a higher level."

Conservation and flux, in Proposition 2.1, were already presented through Gauss-Bonnet as a template: precise exchange among remainders of distinct facets. P2 (fluid mechanics) can directly invoke this template to handle the relationships among multiple conservation laws (energy, momentum, vorticity, entropy); P3-P5 each have their own conservation-flux structures. The Gauss-Bonnet exchange law remains the archetype of this class of structures, available for parallel reference in subsequent papers.

4.2 Information Geometry as the Higher-D Return of Geometry

The state spaces treated in P3 through P5 are mostly not physical spaces or geometric manifolds but statistical manifolds composed of probability distributions. The fitness landscape in evolutionary dynamics may be viewed as a scalar field over species space; the phase space of population dynamics is the probability simplex spanned by species ratios; the public-opinion field in subject dynamics is the distribution density of viewpoints over belief space. To extend the vocabulary of Riemannian geometry to such levels, a bridge is needed: information geometry provides this bridge [21].

Information geometry places the Fisher information metric

$$g_{ij}(\theta) = E_\theta\left[\frac{\partial \log p(x; \theta)}{\partial \theta^i} \cdot \frac{\partial \log p(x; \theta)}{\partial \theta^j}\right]$$

on the parameter space of distributions, making it a Riemannian manifold. On this manifold, the "distance" between two points measures distinguishability between distributions; geodesics correspond to natural transformations within a distribution family; curvature corresponds to the curved structure of the distribution family. The framework developed by Amari, Chentsov, and others [21,22] makes the evolution of probability distributions rigorously geometrizable.

The present paper regards information geometry as a paradigmatic return of Riemannian geometry at higher D-levels in the present series—not claiming it is historically the "first" or "only" return (such first-priority claims invite needless dispute), nor demanding that all subsequent levels adopt Riemannian form. In fact, the state spaces of P3 through P5 may employ a variety of geometric objects: manifolds, stratified spaces, probability simplices, Wasserstein spaces, metric measure spaces, networks, categories, sheaves. What the present paper provides is geometric vocabulary, without imposing a Riemannian template: subsequent papers may freely choose the geometric form most suited to their level, provided the SAE-perspective ontological readings of "metric–connection–curvature–geodesic–singularity–conservation" are preserved.

---

5. Assumption Review, Loosening Partial Order, and Meta-Level Proposition

5.1 Five $\rho = \varnothing$ Assumptions Inherited from Riemannian Geometry

Following the Riemannian methodology continued by this paper, the next step is to review the assumptions Riemannian geometry itself still defaults to. Five are identified.

First, the smoothness assumption. Riemannian geometry assumes the manifold is smooth everywhere; singularities are treated as exceptions, as objects to be excised, rather than as legitimate components of manifold structure. This assumption has been partially loosened in the latter half of the twentieth century by multiple directions: stratified spaces, orbifolds, Alexandrov spaces [23], Ricci flow with surgery, geometric measure theory, singular Ricci flow. From the SAE perspective, this assumption can be replaced by: a singularity is not a failure of geometric object, but the locus of forced manifestation of remainder there.

Second, the integer-dimension assumption. Riemannian geometry assumes manifold dimension is a non-negative integer; fractal dimensions are regarded as "pathological." This assumption admits partial loosening: Hausdorff dimension, Minkowski dimension, spectral dimension already formalize legitimate non-integer-dimensional objects; Connes's spectral-triple framework [24] redefines dimension from the operator-algebraic viewpoint. From the SAE perspective, this assumption can be replaced by: dimension extends from topological integer to a measure / spectrum / operator-growth reading—with the caveat that this extension does not deny integer dimension but treats integer dimension as a special case of the broader concept.

Third, the tensor-metric assumption. Riemannian geometry assumes the metric is a smooth tensor field. This assumption has been partially loosened by multiple directions: Finsler geometry permits direction-dependent norms; Lorentzian geometry permits indefinite signature; information geometry introduces the Fisher metric on statistical manifolds; sub-Riemannian geometry permits some directions to be inaccessible. From the SAE perspective, this assumption can be partially replaced by: metric is not a given background but the mode through which remainder is read out at the level. The present paper only establishes this vocabulary, without constructing a systematic theory.

Fourth, the real-number background assumption. Riemannian geometry assumes the underlying number domain is the real continuum. This is one of the deepest scaffoldings of Riemannian geometry—it bundles together several mutually independent commitments (Cauchy completeness, the Archimedean property, order completeness, the field structure under continuous topology, the uncountable cardinality of the Cantor continuum). Multiple directions have attempted to loosen it: p-adic geometry, Berkovich spaces, adic spaces, motive theory [25], tropical geometry, surreal numbers, smooth infinitesimal analysis. Each direction pays a cost for surrendering the reals (loss of standard analysis, of constructivity, of some topological intuition). The present generation does not replace this assumption: it implicates not only geometric foundations but also the wholesale reconstruction of mathematical and logical foundations, exceeding the burden of this paper. This assumption is one of the boundaries explicitly left for successors.

Fifth, the point-set substrate assumption. Riemannian geometry assumes that space is, first and foremost, a point set, with geometric structure (topology, smooth structure, metric) attached to it. This assumption differs from the real-number background assumption: the latter concerns the value range of coordinates, the former concerns the ontology of space. Connes's noncommutative geometry [24], locale and topos theory [26], derived geometry, and other directions have begun loosening it—they admit that spatial structure may be attached not to a point set but to algebras, categories, or spectra. The present generation likewise does not replace this assumption: its loosening would also implicate too deep a reconstruction of mathematical foundations, beyond the scope of this paper. This assumption is the other boundary explicitly left for successors.

The five assumptions, summarized:

Assumption	Treatment in the Present Generation
Smoothness	Replaced: singularity as locus of forced remainder manifestation
Integer dimension	Replaced: dimension extended to measure / spectrum / operator reading
Tensor metric	Partially replaced: metric as mode of remainder readout
Real-number background	Explicitly left
Point-set substrate	Explicitly left

This treatment is itself the embodied form of the paper's methodological posture. The paper does not attempt to replace all five—doing so would be a posture that treats "the geometric domain" as a one-time-completable object, structurally contradicting the ontological proposition of "the irreducibility of remainder." What the paper does is to review explicitly, replace what can be replaced, explicitly leave what cannot, and treat "non-replacement" itself as content of methodology rather than as oversight.

5.2 The Assumption-Loosening Partial Order and Nested Structure

In conventional mathematical writing, one sometimes encounters expressions of the form "Riemannian geometry $\subset$ non-Euclidean geometries $\subset$ the geometric domain," or analogously on the set-theoretic side "ZFC $\subset$ ZFCρ $\subset$ a larger class of set theories." Such use of the set-inclusion symbol $\subset$ to express disciplinary extension is, strictly speaking, inaccurate: non-Euclidean geometries are not a single set; Finsler, Lorentzian, tropical, p-adic, noncommutative geometries are not "supersets" of Riemannian geometry but rather parallel objects in different directions of assumption-loosening; ZFCρ is not a simple "containment" of ZFC but an axiom-system extension.

For clarity of expression, the present paper introduces the symbol $\preceq$ for the assumption-loosening partial order.

Definition 5.1 (Assumption-loosening partial order). Let $\mathcal{T}_0, \mathcal{T}_1$ be two mathematical theories (each characterized by the union of its axiom set and its default assumptions). Write $\mathcal{T}_0 \preceq \mathcal{T}_1$ if $\mathcal{T}_1$ retains a portion of $\mathcal{T}_0$'s operations while loosening some assumption that was defaulted as $\rho = \varnothing$ in $\mathcal{T}_0$.

Under this notation:

• Geometric side: Riemannian geometry $\preceq$ non-Euclidean family $\preceq$ the geometric domain
• Set-theoretic side: ZFC $\preceq$ ZFCρ $\preceq$ a larger class of set theories

Observation 5.2 (Loosening chains share a pattern). The two loosening chains above share the same abstract pattern: recognition of partial remainder, then recognition of more remainder, while the total class cannot be wholly enclosed by any single framework.

This property of "non-enclosability by any single framework" is itself the precise expression, at the meta-level, of the irreducibility of remainder. The present paper only identifies this observation; the formalization is not developed here, and whether the two loosening chains form a category-theoretic isomorphism is left for P0.

5.3 The Meta-Level Proposition: Unification as Migration

From the five-assumption treatment of §5.1 and the two-loosening-chain observation of §5.2, an SAE meta-level proposition is advanced.

Meta-Proposition 5.3 (Unification as migration). Unification does not eliminate remainder; it migrates remainder.

This proposition requires careful understanding. It is not the assertion that "every unification attempt necessarily fails"—homotopy type theory [27], topos theory [26], $\infty$-category theory are all strong unification attempts at the level of mathematical foundations, each having achieved substantial progress, and the paper acknowledges these advances. What the proposition genuinely asserts is: every unification framework must choose where to place the remainder—it can eliminate some differences but pushes others to the meta-level. Homotopy type theory unifies types, spaces, and logic, but introduces new internal boundaries (universe hierarchy, coherence, computational interpretation); topos theory unifies the multiplicity of set-like universes but brings the multiplicity of internal logics; $\infty$-category theory unifies homotopy invariants but pushes higher-structure complexity onto coherence data. Every unification is a costly migration of remainder, not a cost-free elimination.

From the SAE perspective, this is the ontological grounding for the non-enclosability of the geometric domain by any single framework. It is not a pessimistic conclusion about geometry but the strongest defense of the legitimacy of future geometric work: as long as remainder remains unrecognized, new geometry can be opened.

---

6. Operationalizing Geometric ρ: From Exchange Law to Defect Ledger

6.1 From Ontology to Methodology: Why Operationalization

§§2 through 5 establish the ontological propositions of the paper: Gauss-Bonnet as the four-fold remainder exchange law (Proposition 2.1), Ricci flow singularities as the forced ascent of geometric ρ (Proposition 3.1), and "unification as migration" as the meta-level proposition (Meta-Proposition 5.3). Together these constitute the SAE rereading of geometry at the ontological level.

But ontological rereading by itself is not yet a directly usable tool for working geometers in their concrete research. A geometer working on Ricci-flow singularity classification, moduli-space compactification, bubble-tree convergence, derived-geometry category selection, or weak-solution defect measures—even if endorsing the ontological identifications of §§2 through 5—needs a set of reusable diagnostic tools for the SAE perspective to genuinely enter research practice.

This section operationalizes the foregoing ontological propositions into concrete tools facing working geometers. The tools are organized in two layers. §6.2 through §6.4 are diagnostic frameworks—the ρ-defect ledger, ρ-signature, and readout-first principle—providing geometers with reusable conceptual structures for examining any geometric ascent, singularity treatment, or category extension. §6.5 through §6.7 are three SAE-perspective concrete applications—a quantization conjecture for the Perelman W-functional under Ricci flow surgery, a dissipation-duality conjecture between geometric flows and information-geometric phase transitions, and the rereading of abnormal geodesics in sub-Riemannian geometry as topological breakthroughs of ρ—demonstrating how these diagnostic tools yield concrete, falsifiable, and geometer-relevant outputs in three distinct subfields. §6.8 presents the minimal-ρ-migration principle in checklist form as a methodological supplement for category selection. §6.9 summarizes what this section has committed to working geometers.

A clarification is required: the diagnostic frameworks given here are not new mathematical discoveries—defect measures, bubble-tree identities, Chern-Weil boundary terms, Perelman entropy, and other concrete tools have long existed in the geometric-analysis literature. What this section does is to identify these locally-developed tools, scattered across subfields, as instances of the same "remainder-ledger" structure, unified under the SAE exchange-law perspective. The framework is cross-subfield diagnostic, not new technique within any single subfield.

6.2 The ρ-Defect Ledger: Remainder Bookkeeping for Geometric Ascent

Consider a geometric ascent process

$$X_t \longrightarrow X_\infty$$

where $X_t$ lies in the original category $\mathcal{C}$ (e.g., smooth manifolds, smooth metrics, smooth embedded surfaces, algebraic varieties of fixed topology), and $X_\infty$ lies in the ascended category $\mathcal{C}^\uparrow$ (e.g., singular spaces, surgered spaces, weak-flow limits, stratified spaces, orbifolds, metric-measure limits, derived stacks). The ascent may be implemented by geometric flows (Ricci flow, mean curvature flow, Yang-Mills flow), compactifications (Gromov-Hausdorff limits, bubble-tree convergence, moduli compactifications), or category extensions (schemes to derived schemes, manifolds to orbifolds), among other concrete processes.

Select an invariant family $\mathcal{I} = \{I_\alpha\}_{\alpha \in A}$. Concrete examples include the Euler characteristic $\chi$, signature $\sigma$, curvature integrals $\int \lvert\mathrm{Rm}\rvert^{n/2}$, analytic indices, Perelman entropy $\lambda$ or $\nu$, Chern numbers, Pontryagin numbers, volume growth.

Definition 6.1 (ρ-defect ledger). For an ascent process $X_t \to X_\infty$ and an invariant family $\mathcal{I}$, the ρ-defect ledger is the system of decompositions:

$$I_\alpha(X_t) = I_\alpha(X_\infty^{\mathrm{reg}}) + \sum_{s \in \mathcal{S}(X_\infty)} \rho_{\alpha,s} + \varepsilon_\alpha,$$

where:

• $X_\infty^{\mathrm{reg}}$ is the regular part of the limit object, i.e., the subobject still treatable as a smooth/classical object within the original category $\mathcal{C}$;
• $\mathcal{S}(X_\infty)$ is the set of ascent loci, including singularities, bubble centers, neck regions, surgery operation sites, collapsed strata, and all other locations in $X_\infty$ that exceed $\mathcal{C}$;
• $\rho_{\alpha,s}$ is the remainder released or absorbed at ascent locus $s$ as read by invariant $I_\alpha$, i.e., the precise SAE-quantity for the manifestation of geometric ρ at that locus;
• $\varepsilon_\alpha$ is the non-localizable error term, which should be zero, controlled, or explicitly marked as the part of the ascent theory not yet closed.

ρ-Ledger Adequacy Principle. A geometric ascent theory $\mathcal{C} \to \mathcal{C}^\uparrow$ that claims to "resolve" or "classify" the singularity/degeneration problems of the original category must be able to provide a defect ledger for the invariant family $\mathcal{I}$ of that theory's interest. If it cannot, the theory has not completed remainder bookkeeping; under the SAE perspective, the theory is incomplete.

This principle is not a universal theorem. It does not assert "every geometric ascent automatically possesses a complete ledger," nor does it assert "ascent theories with complete ledgers are the only legitimate ones." Its precise formulation: for any ascent theory claiming to preserve or classify a chosen invariant family $\mathcal{I}$, a necessary adequacy diagnostic is whether the defect ledger relative to $\mathcal{I}$ exists.

On the Relationship to Existing Defect Measure / Bubble Tree Tools. The ρ-defect-ledger framework is not a new mathematical technique, but it carries a key diagnostic difference from the local tools already present in geometric analysis—defect measures, bubble-tree identities, Chern-Weil boundary terms, Uhlenbeck compactness, stratified intersection cohomology. Existing tools typically permit nonzero "residue" or "error" terms within their respective subfields—for instance, the defect-measure framework permits the measure itself to act as "residue not localizable to specific singularities"; bubble-tree identities, in some cases, permit a structural surplus between total energy and the sum over individual bubbles. Within the SAE perspective, the ρ-defect ledger imposes a stricter diagnostic requirement: the existence of $\varepsilon_\alpha$ must be explicitly marked as the part of the ascent theory not yet completed—either $\varepsilon_\alpha = 0$ (full closure) or $\varepsilon_\alpha$ is explicitly acknowledged as "the theory's remainder on this invariant has not yet been localized" (explicit blank). "Vague, unmarked, defaulted residue" is, under SAE diagnosis, inadequate because it conflates "not yet completed" with "completed," obscuring the genuine commitment-status of the ascent theory. This stricter diagnostic requirement is the SAE perspective's concrete contribution to the methodology of geometric analysis—not a replacement for existing tools, but an explicit commitment-status review standard for them.

The four examples below illustrate how the ledger framework manifests in already-known geometric theories.

Example 6.2.1 (Gauss-Bonnet as the simplest ledger). Proposition 2.1 itself is a ledger example. Set $X_t = M$ a bounded smooth surface, $X_\infty^{\mathrm{reg}}$ the interior of $M$, $\mathcal{S}(X_\infty)$ the boundary $\partial M$, and the invariant $I = (1/2\pi)\int_M K\, dA$. The Gauss-Bonnet formula

$$\frac{1}{2\pi}\int_M K\, dA = \chi(M) - \frac{1}{2\pi}\int_{\partial M} k_g\, ds$$

reads as: interior curvature reading = topological reading − remainder released by the boundary. Here $\rho_{\partial M} = -(1/2\pi)\int_{\partial M} k_g\, ds$ is the remainder released by the boundary locus, with $\varepsilon = 0$ (the smooth-boundary case is fully closed).

Example 6.2.2 (Ricci flow with surgery ledger). Set $X_t$ as the smooth manifold before Ricci-flow surgery, $X_\infty$ as the manifold after surgery. Choose $\mathcal{I} = \{\chi, \text{Perelman } \mathcal{W}\}$. The ledger reads:

$$\chi(X_t) = \chi(X_\infty^{\mathrm{reg}}) + \sum_{s \in \mathcal{S}_{\mathrm{surgery}}} \Delta\chi_s + \varepsilon_\chi$$

$$\mathcal{W}(g_t) = \mathcal{W}(g_{X_\infty^{\mathrm{reg}}}) + \sum_{s \in \mathcal{S}_{\mathrm{surgery}}} \Delta\mathcal{W}_s + \varepsilon_\mathcal{W}$$

Each surgery site $s$ releases concrete values $\Delta\chi_s$ and $\Delta\mathcal{W}_s$. In §6.5 this ledger is further refined at the quantization level.

Example 6.2.3 (Yang-Mills bubbling ledger). Set $X_t$ as the smooth connection before Yang-Mills-flow bubbling, $X_\infty$ as the bubbling limit. Choose $\mathcal{I} = \{\text{energy}, c_2\}$ (energy and second Chern class). The ledger reads:

$$E(X_t) = E(X_\infty^{\mathrm{reg}}) + \sum_{\mathrm{bubbles}} E_{\mathrm{bubble}} + \varepsilon_E$$

Each bubble releases an integer multiple of the "instanton energy" $8\pi^2/g^2$—a perfect ledger (the content of Uhlenbeck's compactness theorem [32], identified by the SAE perspective as a ledger paradigm).

Example 6.2.4 (Gromov-Hausdorff collapse ledger). Set $X_t$ as a family of Riemannian manifolds, $X_\infty$ as their GH limit (potentially of lower dimension). $\mathcal{I}$ includes volume, diameter, lower Ricci-curvature bounds. The collapse loci release dimensional information (from the $n$-dimensional manifold to a lower-dimensional metric-measure space). The precise form of the ledger here is still an active direction in geometric analysis [33]; the SAE perspective identifies this direction as an attempt to provide GH collapse with a complete defect ledger.

6.3 The ρ-Signature: Singularity Fingerprint

A direct consequence of the ρ-defect ledger is that every ascent locus $s$ is endowed with a remainder vector:

Definition 6.3 (ρ-signature). Let $X_t \to X_\infty$ be an ascent process possessing a defect ledger, with chosen invariant family $\mathcal{I} = \{I_\alpha\}_{\alpha \in A}$. For each ascent locus $s \in \mathcal{S}(X_\infty)$, the ρ-signature is the vector

$$\rho_s(\mathcal{I}) := (\rho_{\alpha,s})_{\alpha \in A}.$$

The ρ-signature is the "fingerprint" of the ascent locus under the chosen invariant family—it precisely characterizes how much of which kinds of remainder that locus releases or absorbs.

The ρ-signature elevates singularity classification from object-classification to fingerprint-classification. In conventional geometry, singularities are classified by local object type—conical singularities, cusp singularities, neckpinches, bubbles, orbifold points, and so on—which is essentially classification by the local geometric object at the singularity. The ρ-signature, by contrast, gives classification by the release pattern of the singularity over the chosen invariants—the same local-object type may yield different signatures under different invariant families, and different local-object types may yield the same signature under a particular invariant family. This classification is research-question-oriented: whichever invariants concern you, the ρ-signature tells you the release pattern of the singularity over those invariants.

As a concrete application of the ρ-signature, consider the Gauss-Bonnet remainder-defect classification of isolated singularities on two-dimensional manifolds.

Gauss-Bonnet remainder defect classification. For an isolated singularity $p$ on a two-dimensional manifold, define the Gauss-Bonnet remainder defect:

$$\Delta_\rho(p) := \lim_{\epsilon \to 0^+}\left[2\pi\chi(M_\epsilon) - \int_{M_\epsilon} K\, dA - \int_{\partial M_\epsilon} k_g\, ds\right]$$

where $M_\epsilon = M \setminus B_\epsilon(p)$. Under the choice $\mathcal{I} = \{\chi\}$, $\Delta_\rho(p)$ is the ρ-signature of the singularity $p$. The signature falls into three classes:

• Class A ($\Delta_\rho(p) \in 2\pi\mathbb{Z}$): smoothable singularities, with signature on the standard integer exchange grid;
• Class B ($\Delta_\rho(p) \in 2\pi\mathbb{Q} \setminus 2\pi\mathbb{Z}$): conical / orbifold singularities, with signature on the rational extension of the exchange grid;
• Class C ($\Delta_\rho(p) \notin 2\pi\overline{\mathbb{Q}}$): ontological singularities, with signature outside the standard algebraic exchange.

Problem 6.4 (Existence of Class C). Does Class C exist? Can it be found via standard geometric constructions?

This is an open problem. The Gauss-Bonnet defects of all known isolated singularities (conical, cusp, orbifold) are rational numbers, falling into Class A or Class B. If Class C exists, it would be the SAE-perspective "ontological singularity"—not resolvable to Class A or Class B by any homotopic smooth deformation, marking the irreducible manifestation of underlying ρ outside standard algebraic exchange.

The paper does not assert that Class C must exist. What it does assert: the ρ-signature framework gives rise to the concrete question "does Class C exist?"—a question that does not naturally arise from the pure-geometric perspective, surfacing only under the SAE Archimedean exchange-law perspective. This is one concrete contribution of the SAE perspective to geometry: it identifies a new research direction.

Why working geometers should care about Class C. The Class-C existence question is not merely abstract; it interfaces directly with several active research directions.

First, moduli compactification. In the Deligne-Mumford compactification of the surface moduli space $\mathcal{M}_g$, the boundary corresponds to surface degenerations (cycle contraction, node formation). All such degenerations are of Class A or Class B type (integer or rational multiples of $2\pi$ defect), consistent with known Hodge theory. If Class C singularities exist, they would be degeneration modes uncapturable by any Deligne-Mumford-type compactification—forcing geometers to seek entirely new compactification objects. "Does Class C exist?" is equivalent to asking "Is the Deligne-Mumford compactification complete?"

Second, bubble-tree convergence. In harmonic-map / minimal-surface / Yang-Mills bubbling, bubbles release integer multiples of "instanton energy" or "harmonic energy"—the typical Class A behavior. Class B occurs in orbifold-quotient and similar settings. If Class C exists, it would mean some bubble releases a transcendental multiple of energy—fundamentally altering the form of the bubble-tree compactness theorem and forcing the introduction of new invariant families.

Third, stratified geometry. Whitney stratification, Goresky-MacPherson intersection cohomology, and related theories all rest on the implicit assumption that "the link of a singularity is a rational-cohomology object" (Class A/B types). If Class C singularities exist, the stratified theory would be required to extend its underlying number domain from rationals to a period number field—a directional reorientation crossing algebraic geometry and topology.

These interfaces show that the existence of Class C is not a question "of philosophical concern but of no consequence to geometers." It is a concrete diagnostic question directly affecting three active research directions of working geometers. The paper raises the question; the concrete construction of existence, the proof of non-existence, or the precise interface with existing theory is left for the geometric community.

6.4 The Readout-First Principle

The ρ-defect ledger and ρ-signature address the question of "how to keep the books." But in actual research, geometers face a more upstream question: what should be controlled? Local data (curvature, energy, defect measures) have many components; each invariant reads only a portion. Which components should be prioritized?

The default of conventional geometric analysis is to first control total energy or total norm, then handle the details. Yet Gauss-Bonnet, the signature theorem, and the index theorem all teach the same lesson: global invariants do not look at the total energy of the local field; they look at a particular readout functional.

Principle 6.5 (Readout-first principle). In the study of a global geometric invariant $I$, prioritize control over the projection of local remainder onto the readout direction of $I$, rather than over the total norm of local remainder.

The formalization of this principle: given a global invariant $I = \mathcal{R}[F]$, where $\mathcal{R}$ is the readout functional from local data $F$ to the global invariant. Decompose the local field as

$$F = F_{\mathrm{readout-visible}} + F_{\mathrm{readout-invisible}},$$

where $F_{\mathrm{readout-visible}}$ is the part seen by $\mathcal{R}$ and $F_{\mathrm{readout-invisible}}$ the part invisible to $\mathcal{R}$ (contributing nothing to $I$). The principle asserts: the energy-dominant mode is not necessarily the invariant-dominant mode; the invariant-dominant mode is the readout-visible mode.

Three examples illustrate the principle.

Example 6.4.1 (Gauss-Bonnet). The global invariant $\chi$ reads the Pfaffian form $\mathrm{Pf}(\Omega)$, not the curvature tensor in its entirety. The symmetric portions of the curvature tensor (Ricci scalar, etc.) make no contribution to $\chi$; they are readout-invisible. The implication: controlling $\int \lvert\mathrm{Rm}\rvert^2$ as total energy does not directly control $\chi$. What truly matters is control over the antisymmetric component of the Pfaffian.

Example 6.4.2 (Hirzebruch signature theorem). The global invariant $\sigma(M)$ reads the Chern-Weil representative of the $L$-class polynomial $L(p_1, p_2, \ldots)$, not the Pontryagin classes in their entirety. This assigns a precise weight to each curvature component for its "contribution to signature."

Example 6.4.3 (Atiyah-Singer index theorem). The global invariant $\mathrm{ind}(D)$ reads the pairing between the characteristic form and the Chern character, not the elliptic operator's full analytic energy. This is precisely why the index is a topological invariant: the bulk of analytic variation is readout-invisible.

The principle's specific guidance for geometric-analysis practice manifests in three directions. In Ricci-flow blow-up analysis, asking "is the curvature norm I control readout-visible or readout-invisible?" can help identify whether the blow-up sequence captures the genuine mechanism of topological change. In weak-convergence and defect-measure analyses, identifying which mode is readout-visible avoids spending effort on the readout-invisible portion. In normalization choices, asking "does my normalization mask the readout-visible mode?" guards against fake closure—the rhetorical "resolution" by which a defect is defined to zero through normalization.

The readout-first principle is not a new tool, but the distillation of the already-known phenomenon of "global invariant vs local energy" into a reusable methodological precept. Its SAE irreplaceability lies in this: the pure-geometric perspective typically thinks in terms of "object–attribute" (objects have curvature, energy, index), whereas the SAE perspective thinks in terms of "object–readout–remainder" (objects manifest as remainder through readout functionals, with the readout-invisible portion neither seen nor booked).

6.5 Application I: Quantization Conjecture for the W-Functional in Ricci-Flow Surgery

This section demonstrates the ρ-defect-ledger framework's concrete application in geometric analysis.

In 2002–2003, Perelman introduced the W-functional (entropy functional) [13]:

$$\mathcal{W}(g, f, \tau) = \int_M \left\tau(R + \lvert\nabla f\rvert^2) + f - n\right^{-n/2} e^{-f}\, dV$$

It is monotone non-decreasing under Ricci-flow evolution. Perelman further defined invariants $\lambda(g) = \inf_f \mathcal{W}(g, f, \tau=1)$ and $\nu(g) = \inf_{\tau,f} \mathcal{W}(g, f, \tau)$, also monotone under Ricci-flow evolution.

Under Ricci flow with surgery, $\mathcal{W}$ exhibits a jump at surgery moments:

$$\Delta\mathcal{W}_s := \lim_{t \to t_s^-}\mathcal{W}(g(t)) - \lim_{t \to t_s^+}\mathcal{W}(g(t))$$

where $t_s$ is the surgery time. Bamler's 2018 work [16] extends the monotonicity of $\mathcal{W}$ to the weak-flow framework and shows $\Delta\mathcal{W}_s$ is bounded.

Conjecture 6.6 (Quantization conjecture for the W-functional jump). At each singularity surgery moment $t_s$ in Ricci flow with surgery, the jump $\Delta\mathcal{W}_s$ is quantized: there exists a non-trivial function $Q(\Delta\chi_s, \Delta b_0, \tau_s, \ldots)$, determined by the topological transition data of the surgery (change in Euler characteristic, change in connected components, surgery type, etc.), such that

$$\Delta\mathcal{W}_s = Q(\text{topological transition data at } s).$$

Furthermore, under appropriate normalization, $Q$ is a discrete-valued function—its value set is a countable discrete set determined by the topological transition data.

The conjecture deliberately leaves the specific functional form of $Q$ open (not specifying $Q = -2\pi \cdot \Delta\chi$ or $Q = c \cdot \Delta\chi^2$ or any other concrete formula). The reason: existing entropy bounds from Bamler and others are of inequality form (monotonicity + boundedness); a specific "equality-type" quantization formula remains an open problem, with concrete coefficients dependent on precise normalization and dimension. The conjecture provides a structural prediction—the jump is a discrete value controlled by topological transition data—while the specific functional form is left to subsequent work.

The SAE-perspective irreplaceability is clear here. Pure-geometric inspection of Perelman entropy gives monotonicity + boundedness; there is no intrinsic reason for strict-equality quantization. The SAE ρ-defect-ledger perspective suggests a stronger structure: $\mathcal{W}$ is a ledger invariant, surgery is an ascent locus, hence $\Delta\mathcal{W}_s$ should be the precise component of that locus's ρ-signature on $\mathcal{W}$. The ledger metaphor naturally inclines toward closure as strict equality rather than inequality—this inclination is not a derived theorem but a structural expectation of the SAE framework over geometric analysis. Whether geometric-analysis practice will discover that the W-functional jump is in fact strict equality (supporting the SAE expectation), or substantive inequality with irreducible error (challenging the SAE expectation), is precisely what this conjecture leaves as the falsifiable content for future work.

The conjecture admits future falsification or refinement. If a concrete instance shows $\Delta\mathcal{W}_s$ to be continuous-valued under reasonable normalization, the conjecture is falsified. If a concrete functional form $Q$ is proved, the conjecture is sharpened to a theorem. If $Q$ is shown to quantize for some surgery types but not others, the conjecture requires classificational refinement.

Whatever the outcome, the conjecture provides geometric-analysis researchers with a concrete, falsifiable, naturally-SAE-perspective research question—which is precisely this section's goal: not to give geometers a new theorem, but to give them a new question.

6.6 Application II: Dissipation Duality Between Geometric Flows and Information-Geometric Phase Transitions

This section demonstrates the ρ-defect-ledger framework's application as a cross-subfield bridge, simultaneously providing a shared geometric tool for the subsequent papers P2 through P5 of the "Form and Flow" series.

Perelman's W-functional

$$\mathcal{W}(g, f, \tau) = \int_M \left\tau(R + \lvert\nabla f\rvert^2) + f - n\right^{-n/2} e^{-f}\, dV$$

is, on its surface, a geometric-analysis object, yet its structure bears a striking resemblance to the Boltzmann-Gibbs entropy functional in statistical physics (under a particular Lagrange-multiplier form). Indeed, Perelman himself in [13] noted that $\mathcal{W}$ admits interpretation as a Lagrange-multiplier-form statistical-mechanics functional. The work of Lott-Villani [34], Sturm [35], Otto-Villani [36], and others has further connected, in the framework of Wasserstein geometry and optimal transport, Ricci curvature with the entropy geometry on statistical manifolds: specifically, the lower Ricci-curvature bound $\mathrm{Ric} \geq K$ is equivalent to $K$-convexity of the entropy functional on Wasserstein space, the foundation of the RCD($K$, $N$) space theory.

This phenomenon, from the SAE perspective, should not be read as coincidence but as the necessary consequence of the irreducibility-of-ρ law: the geometric and informational levels must possess isomorphic dissipation equations.

Conjecture 6.7 (Dissipation duality between geometric flows and information-geometric phase transitions). The processes of metric collapse (volume collapse), singularity formation, and surgery ascent in Ricci flow are, in mathematical structure, dual to the processes of degeneration, singularization, and reconfiguration of the Fisher metric on the underlying probability-distribution family during phase transitions of statistical-physics systems in the thermodynamic limit.

More precisely, the conjecture contains a level distinction. The established RCD-level connection: Lott-Villani-Sturm have established at the static level the precise equivalence between lower Ricci-curvature bounds and Fisher / entropy geometry (the RCD($K$, $N$) space theory)—a mature bridge between geometry and information geometry at the static-curvature level. The extension proposed by this conjecture: this bridge can be extended to the dynamical level—Ricci-flow surgery topological transitions on one side should correspond to statistical-manifold phase-transition topological reorganizations on the other; the jump $\Delta\mathcal{W}_s$ on the Ricci-flow side should correspond to phase-transition entropy jumps on the statistical-manifold side.

It must be made explicit: the dynamical-level extension is itself an open conjecture, not yet fully established in the existing geometric / probabilistic literature even outside the SAE perspective. The conjecture is not "an existing-literature phenomenon SAE merely renames" but "a direction explicitly proposed for the first time in this series." The contribution of the SAE perspective is to provide a unified vocabulary for this direction (the ρ-defect ledger applies to both sides); the formal completion of the direction (concrete categorical structure of the transformation, ledger correspondence, precise duality of Wasserstein geometry and Ricci flow at the surgery / phase-transition level) is left for future work. Within the present paper, "duality" is used as informal vocabulary, awaiting future formalization into a suitable categorical structure (which may be a functor, may be a natural transformation, may be a derived correspondence; the specific form remains to be determined).

The SAE-perspective ontological reading: geometric ρ (curvature as local irreducible resistance) and informational ρ (distribution distinguishability as irreducible cost) are the same ontological resistance manifesting at two distinct D-levels. They must satisfy the same dissipation equation at the dynamical level because what drives both dynamics is the same ρ. Dissipation duality is no mathematical coincidence; it is the mirrored necessity of the SAE irreducibility-of-ρ proposition at two levels.

The cross-subfield significance of this conjecture lies in its simultaneous connection to three originally-separated research communities. Differential geometry and geometric analysis (Hamilton-Perelman lineage) concern themselves with Ricci flow with surgery, Perelman entropy, and geometric H'-type ascent. Information geometry (Amari-Chentsov lineage) concerns itself with the Fisher metric, $\alpha$-connections, and statistical-manifold curvature. Optimal transport (Otto-Villani-Sturm-Lott lineage) concerns itself with Wasserstein geometry, entropic transport, and RCD spaces. Each of these fields has, over the past one to two decades, developed deep geometric tools, but the concrete bridging among them has remained predominantly at the "lower-curvature-bound ↔ entropy-convexity" RCD phenomenon. The conjecture predicts: this bridge can be extended to the more dynamical level of singularities / phase transitions / surgery / ascent, and the SAE ρ-defect-ledger framework is the unifying vocabulary for this extension.

The conjecture also directly provides a shared geometric tool for the subsequent papers in the series. The boundary layers, turbulence, and singularity bursts in P2 fluid mechanics may be read as geometric duals of Fisher-metric phase transitions. The fitness-landscape collapses and species-extinction events in P3 evolutionary dynamics may be read as phase transitions and ascent on Wasserstein metrics. The replicator flows and phase-space bifurcations in P4 population dynamics may directly invoke the unified framework given by the conjecture. The opinion-field flows and belief-distribution evolutions in P5 subject dynamics may be handled in the coupled framework of statistical manifold + geometric flow.

This conjecture is the most important geometric-tool bridge within the "Form and Flow" series. The present paper only establishes its posture and statement; concrete formalization and applications across D-levels are left for P2 through P5.

6.7 Application III: Abnormal Geodesics in Sub-Riemannian Geometry as Topological Breakthroughs of ρ

This section demonstrates the ρ-defect-ledger perspective's rereading of a classical geometric "pathological" phenomenon, exemplifying the SAE framework's "turning enemies into allies" posture.

Sub-Riemannian geometry is an important extension of Riemannian geometry: on a manifold $M$, specify a rank-$k$ distribution $\mathcal{D} \subset TM$ with $k < \dim M$, requiring "motion only within $\mathcal{D}$." This kind of geometry has concrete applications in control theory (robotic motion, wheeled rolling, quantum control systems), heat diffusion (degenerate Laplacians), and Carnot-group analysis.

A core difficulty of sub-Riemannian geometry is the existence and properties of abnormal geodesics—curves satisfying the local-shortness condition but not arising from the standard variational principle of the sub-Riemannian Hamiltonian. Their existence was systematically proposed as the century-long puzzle by Sussmann in 1996 [37] and remains, even today, one of the deepest open problems in sub-Riemannian geometry and geometric control. Liu-Sussmann in 1995 constructed concrete abnormal-geodesic examples, demonstrating their existence and that they may be locally strictly shortest; but whether abnormal geodesics are always smooth, whether they constitute legitimate components of the geometric object, and how they should be handled in classification theory all remain without unified answer.

The treatment of abnormal geodesics in conventional sub-Riemannian-geometry literature is heterogeneous: one strand of the literature (especially in classification-theoretical orientations) regards them as "exceptional cases," requiring "discussion outside the abnormal geodesics" in order for standard theorems to hold; another strand (the lineage of Liu-Sussmann, Hsu, Agrachev-Sachkov, Montgomery's monograph) treats them as legitimate objects of active research. Even within the active-research lineage, the concrete properties of abnormal geodesics—smoothness, classification, relations to other features of the geometric object—remain unsettled open problems. The SAE perspective does not claim that geometers "have not seen" abnormal geodesics; rather, it offers a rereading lens for already-extant research.

SAE Rereading 6.8 (Abnormal geodesics as topological breakthrough of ρ). Abnormal geodesics are not pathological; they are the topological breakthrough produced by ρ as it seeks the path of least resistance within the highly-compressed constraint space (rank $k < n$ distribution).

To explain concretely: in standard Riemannian geometry, all directions are accessible, and ρ manifests as curvature, the cumulative reading of local bending. In sub-Riemannian geometry, the directions of motion are strictly restricted, and ρ cannot manifest through ordinary curvature because most directions "do not exist." This restriction does not eliminate ρ—by Meta-Proposition 5.3 ("unification as migration"), ρ must migrate to a new locus. Abnormal geodesics are the locus to which ρ has migrated: they are the products of geometric ρ "breaking through" the standard Hamiltonian framework under extreme constraint, seeking alternative manifestation paths.

In the ρ-defect-ledger framework, this phenomenon admits formalization. Set $X_t$ as the unconstrained (full-Riemannian) object ($\mathcal{D} = TM$), $X_\infty$ as the sub-Riemannian object ($\mathcal{D} \subsetneq TM$). This process is a "constraint ascent"—the categorial extension from rank $= n$ to rank $= k < n$. Choose $\mathcal{I} = \{\text{geodesic structure}, \text{cut locus}, \text{minimizer set}\}$. The ledger:

$$\text{geodesics}(X_t) = \text{Hamiltonian geodesics}(X_\infty) + \text{abnormal geodesics}(X_\infty) + \varepsilon$$

Abnormal geodesics are the nonzero component of the ascent's ρ-signature on geodesic structure.

The significance of this rereading must be precisely articulated. It does not resolve the Sussmann puzzle—the concrete properties of abnormal geodesics (smoothness, classification, etc.) remain open. It also does not claim that the entire sub-Riemannian community treats abnormal geodesics as pathological while only SAE perceives their legitimacy—a substantial body of work already takes them seriously. What it does is to provide, for the heterogeneous existing treatments, a unified ontological coordinate: in the conventional perspective, "abnormal geodesics as pathological exceptions" and "abnormal geodesics as legitimate objects" can be simultaneously accommodated, the former corresponding to the "elimination-attempt" posture and the latter to the "recognition-of-ρ-manifestation" posture; the SAE perspective holds that the latter is closer to the ontological identity of abnormal geodesics, without denying the engineering value of the former.

This is a concrete demonstration of the SAE perspective's "turning enemies into allies" posture—providing for objects regarded by some literature as "pathological exceptions" a unified framework for their rereading as "legitimate manifestations of ρ." The paper does not claim this rereading automatically resolves the Sussmann puzzle, but holds that it offers sub-Riemannian-geometry research a possibility extension: between "avoiding abnormal geodesics" and "analyzing abnormal geodesics themselves," the SAE perspective proposes a third path—analyzing the ρ-signature structure of abnormal geodesics, treating them as fingerprint objects within the SAE-ascended category.

6.8 The Minimal-ρ-Migration Principle: Category-Selection Checklist

This section provides a methodological supplement for category selection. Meta-Proposition 5.3 ("unification as migration") induces a concrete principle in the practice of category selection.

When the original category $\mathcal{C}$ cannot accommodate the limits, singularities, solutions, or classifications of an object, geometric researchers must select an ascended category $\mathcal{C}^\uparrow$. The choices are multiple. For Ricci-flow singularity treatment: surgery, singular Ricci flow, weak flow, through-singularities flow. For algebraic geometry: schemes, stacks, derived schemes, log schemes, $\infty$-stacks. For metric geometry: GH limits, metric-measure spaces, RCD spaces, smooth metric measure spaces. For differential geometry: orbifolds, stratified spaces, currents, varifolds. For moduli theory: classical moduli, virtual fundamental class, derived enhancements.

Each kind of ascent "resolves" a particular problem of the original category but also "introduces" new internal boundaries—new strata, new coherence data, new categorial levels, new singularity types. SAE Meta-Proposition 5.3 asserts that this is necessary: any ascent does not eliminate ρ but only migrates it.

Principle 6.9 (Minimal-ρ-migration principle). In the selection of an ascended category, the legitimate choice is the minimal-ρ-migration path: under the constraint that the original problem remain solvable, the path introducing the fewest new ρ data (new internal boundaries + new coherence data) is preferred.

Formalized as an operational checklist: an ascent $\mathcal{C} \to \mathcal{C}^\uparrow$ is ρ-minimal relative to invariant family $\mathcal{I}$ if it satisfies:

1. Full embedding: $\mathcal{C}$ is fully faithfully embedded into $\mathcal{C}^\uparrow$, i.e., the objects and morphisms of the original category are preserved within the new category;
2. Solvability: $\mathcal{C}^\uparrow$ accommodates the limits, singularities, solutions, or classifications required by the original problem;
3. Ledger existence: $\mathcal{C}^\uparrow$ provides a defect ledger relative to $\mathcal{I}$;
4. Coherence explicitness: all new coherence data introduced by the ascent are explicit, identifiable, and countable;
5. Irreducibility: there exists no strictly smaller ascent $\mathcal{C}^\uparrow_0 \subsetneq \mathcal{C}^\uparrow$ simultaneously satisfying (1)–(4).

An ascent satisfying these five is ρ-minimal. An ascent satisfying only (1)–(2) but not (3)–(5) "resolves" the problem without completing remainder bookkeeping; under SAE diagnosis, it is over-enlargement—introducing unnecessary remainder migration.

This principle yields methodological judgment for concrete category-selection problems. In derived geometry, when choosing among derived schemes / derived stacks / $\infty$-stacks, ask: "does the invariant family of my interest genuinely require higher-order derived structure? Are the higher coherence data introduced strictly necessary?" In moduli theory, when choosing between classical moduli and virtual fundamental class, ask: "is the obstruction theory introduced by the virtual class genuinely utilized by the invariant family I care about?" In singular geometry, when choosing among orbifolds / stratified spaces / currents / varifolds, ask: "which class of objects gives my invariant family the most complete and concise ledger?"

The minimal-ρ-migration principle is not a revolution against existing mathematical practice—many mathematicians intuitively apply similar judgment in category selection ("use the minimal sufficient category"). What this principle does is to explicitly formalize this intuitive judgment as a concrete operational checklist and provide the SAE ontological grounding for "why these five conditions." It does not replace researchers' specific judgment but offers a unified diagnostic framework.

A deeper formalization—rigorous metrization of "ρ-migration cost," construction of "migration functors" at the level of category hierarchies, proof of the minimality of specific ascents—is an independent project, left for P0 and future SAE Category-Theory treatments. What this section provides is the principle and the checklist, not a theorem.

6.9 Summary: This Section's Commitments to Working Geometers

The §6 commitments to working geometers are summarized as follows.

Diagnostic tools (reusable): The ρ-defect ledger (Definition 6.1) is a unified framework for examining whether any ascent theory has completed its remainder bookkeeping; the ρ-signature (Definition 6.3) elevates singularity classification from object-classification to fingerprint-classification; the readout-first principle (Principle 6.5) is the methodological precept of prioritizing readout-visible modes over total energy.

Concrete falsifiable conjectures: The W-functional quantization conjecture (Conjecture 6.6) predicts that the Perelman entropy jump in Ricci-flow surgery is a discrete value controlled by topological transition data; the geometric-flow / information-geometric dissipation duality conjecture (Conjecture 6.7) predicts that Ricci flow and Fisher-metric phase transitions are structurally dual at the dynamical level, extending the existing static RCD-level connection, as the foundation for shared geometric tools in P2 through P5.

Research-direction rereading: The reading of abnormal geodesics in sub-Riemannian geometry as topological breakthroughs of ρ (SAE Rereading 6.8) is a concrete demonstration of the "turning enemies into allies" posture.

Methodological supplement: The minimal-ρ-migration principle and its 5-point checklist (Principle 6.9) is the SAE-perspective criterion for category selection.

These tools and conjectures do not require geometers to accept the entire SAE framework. Each may be independently examined, used, falsified, or developed. The contribution of the SAE perspective is to identify the unity among these tools and conjectures—they all stem from the ontological proposition that "ρ is irreducible; unification is migration"—but each tool can survive independently of this unity.

This section's goal is thereby fulfilled: that working geometers, after reading the present paper, be willing to attempt the SAE perspective on their own concrete problems—because they have seen concrete usable diagnostic frameworks, falsifiable conjectures, rereadable phenomena, and operational checklists.

---

7. Remaining Work: Complex Geometry, Relations to Existing SAE Series, Explicit Blanks

7.1 The Complex-Geometry Placeholder

The geometry treated in this paper is real Riemannian geometry and several extensions over the standard real-number background. Complex geometry—Hermitian geometry, Kähler geometry, Calabi-Yau manifolds [28], complex algebraic geometry—is another face of geometric remainder, not addressed here. Complex structure is itself a non-trivial remainder; the vocabulary of real Riemannian geometry cannot directly capture it. The paper does not undertake any treatment of this face but explicitly marks it as future work: "On the Hypotheses Which Lie at the Foundations of Geometry, Part III" may be the inception of SAE Complex Geometry, to be carried out by successors. This is an independent direction parallel to, not subordinate to, the present paper.

Similarly, the precise exchange between analytic and topological invariants revealed by spectral geometry and the index theorem (Atiyah-Singer et al. [9]) is a deeper version of the exchange-law framework of Proposition 2.1, also not developed here, left as another independent direction.

7.2 Relations to Existing SAE Series

The "Form and Flow" series shares its framework with the existing SAE Physics series [29], SAE Information Theory series [30], SAE Biology Notes, SAE Moral Law series [31], SAE Law series, SAE Epistemology series, SAE Economics series, and Human Total Construct series [2,3]. These series unfold the SAE ontological proposition across distinct facets—SAE Physics on the law-of-nature facet, SAE Information Theory on the physics–information interface, SAE Biology Notes on the life-science facet, SAE Moral Law on the ethical-structure facet, Human Total Construct on the intellectual-history facet—while "Form and Flow" treats the geometry-and-dynamics facet across D-levels. The series do not stand in hierarchical relation to one another; they are parallel facets of the same ontological framework. The present paper opens the "Form and Flow" series; subsequent P2 through P5 and P0 unfold in turn. The paper does not attempt to cover topics already addressed by other series.

The dissipation duality conjecture in §6.6 is the most important bridge within the "Form and Flow" series: it simultaneously underwrites the geometric tools of P2 (fluids), P3 (evolution), P4 (population), and P5 (subject dynamics). Subsequent unfoldings at each D-level will repeatedly refer back to this duality conjecture and to the diagnostic frameworks in §6 (ρ-defect ledger, ρ-signature, readout-first principle).

7.3 Explicit Blanks

Items the paper explicitly does not undertake, left for future SAE-Geometry treatments or for successors, are listed as follows:

(i) The categorical formalization of the structural analogy between geometric H' and ZFCρ H' (specifying the categories, functors, and natural transformations);

(ii) The unified statement, at the abstract level, of the family of fixed-point theorems (Brouwer, Banach, Lefschetz, Tarski, etc.) and "the irreducibility of remainder";

(iii) The spectral-geometry remainder-conservation family: SAE rereadings of $\zeta$-function special values, the Selberg trace formula, heat-kernel coefficients, analytic torsion, indices of elliptic operators; the SAE-perspective treatment of period theory (Kontsevich-Zagier);

(iv) The SAE deduction for the complex-geometry / Kähler / Calabi-Yau face—"On the Hypotheses Which Lie at the Foundations of Geometry, Part III";

(v) The complete SAE rereading of higher-dimensional characteristic classes (Pontryagin, Chern, Euler);

(vi) The substantive replacement work for the real-number background and point-set substrate assumptions;

(vii) The SAE treatment of the constructive / computability dimension of mathematics (a deeper-level blank);

(viii) The determination of the specific functional form $Q$ in the W-functional quantization conjecture (Conjecture 6.6);

(ix) The formalization of the geometric-flow / information-geometric dissipation duality conjecture (Conjecture 6.7): concrete categorical transformation, ledger correspondence, precise duality of Wasserstein geometry and Ricci flow at the surgery / phase-transition level;

(x) The existence problem and concrete construction of Class-C singularities (§6.3);

(xi) The rigorous metrization of "ρ-migration cost" in the minimal-ρ-migration principle (Principle 6.9), the construction of "migration functors" at the level of category hierarchies, and proofs of minimality for specific ascents;

(xii) The concrete analysis of the ρ-signature structure of abnormal geodesics (§6.7).

7.4 Claim-Status Map: Explicit Hierarchy of the Paper's Commitments

To make the level-status of every formal statement in the paper visible at a glance, the formal statements are organized into four levels of epistemic strength, plus the explicit-blanks layer.

Level 1: Hard claims advanced as propositions

Label	Content	Nature
Proposition 2.1	Gauss-Bonnet as four-fold remainder exchange law	Ontological rereading of established mathematics
Proposition 2.2	Falsifiable form of Gauss-Bonnet exchange law under standard normalization	Rigorous proof, contingent on Lindemann's theorem
Proposition 3.1	Geometric H'-type reading of Ricci flow	Ontological rereading of the Hamilton-Perelman program

These three are the paper's core commitments, submitted to the geometric community for substantive scrutiny.

Level 2: Methodological frameworks and meta-level propositions

Label	Content	Nature
Abstract Pattern 3.3	H'-type indissolvability pattern	Cross-subfield abstract-pattern identification
Definition 5.1	Assumption-loosening partial order $\preceq$	Notational clarification
Observation 5.2	Loosening chains share a pattern	Cross-subfield structural observation
Meta-Proposition 5.3	Unification as migration	SAE meta-level ontological proposition
Pointing Proposition 2.4	Periods as standard manifestation of arithmetic remainder	Directional statement, awaiting formalization

This level provides the methodological and ontological framework, supporting the argumentation of the core propositions and laying groundwork for subsequent papers.

Level 3: Diagnostic tools (the §6 operational framework)

Label	Content	Nature
Definition 6.1	ρ-defect ledger	Cross-subfield diagnostic framework
Definition 6.3	ρ-signature	Tool for fingerprint-classification of singularities
Principle 6.5	Readout-first principle	Methodological precept
Principle 6.9	Minimal-ρ-migration principle and 5-point checklist	Methodological criterion for category selection

This level operationalizes the Level-1 and Level-2 ontological propositions into reusable diagnostic tools for working geometers. Not new mathematical discoveries, but a unifying framework for already-existing geometric-analysis tools.

Level 4: Conjectural applications and open research problems

Label	Content	Nature
Conjecture 3.2	Original-category irreducibility prediction	Falsifiable conjecture
Problem 6.4	Existence of Class C singularities	Open problem
Conjecture 6.6	W-functional quantization conjecture	Falsifiable conjecture (structural form)
Conjecture 6.7	Geometric-flow / information-geometric dissipation duality	Falsifiable conjecture (structural-duality form)
SAE Rereading 6.8	Abnormal geodesics as topological breakthrough of ρ	Research-direction rereading, not resolution of a specific puzzle

This level is not on equal footing with Levels 1 and 2. It consists of the SAE perspective's concrete research invitations to working geometers—falsifiable conjectures or new research directions. Their formalization, calculation, and verification are left to the geometric community. The epistemic status of this level is "falsifiable conjecture and research prompt," not "argued conclusion."

Explicit blanks (§7.3) form a fifth level: directions the paper explicitly does not undertake, left for future work. Twelve items are listed.

These four levels plus the explicit-blanks layer constitute the complete hierarchical map of the paper's commitments.

7.5 Conclusion

The two propositions undertaken at the core of this paper—the Gauss-Bonnet four-fold remainder exchange law (Propositions 2.1, 2.2) and the geometric H'-type ascent reading of Ricci flow (Proposition 3.1)—are the ontological rereadings the paper offers. On this basis, the paper provides five assumption treatments and methodological tools: the geometric vocabulary, the information-geometry bridge, the assumption-loosening partial order, the meta-level proposition "unification as migration," and the §6 diagnostic frameworks (ρ-defect ledger, ρ-signature, readout-first principle, minimal-ρ-migration checklist). Finally, the paper offers three concrete research invitations to the geometric community—as conjectural applications and open research problems—the W-functional quantization conjecture, the dissipation-duality conjecture, and the abnormal-geodesics-as-topological-breakthrough rereading.

These three groups of contributions are not on equal epistemic footing: the core propositions are the paper's hard claims; the methodological and diagnostic frameworks are supporting tools for the argumentation; the conjectural applications are concrete research invitations to working geometers. The claim-status map of §7.4 lays out this hierarchy explicitly.

Continuing the methodology of Riemann's 1854 lecture, the paper has reviewed several assumptions, replaced those that can be replaced, explicitly acknowledged those that cannot, and operationalized the ontological propositions into a set of reusable tools facing working geometers. The form of the paper itself—identifying assumptions, replacing some, explicitly leaving blanks—constitutes embodied evidence for the ontological proposition that remainder is irreducible.

The contribution of the present paper is not to establish an SAE-geometry system, but to reactivate Riemann's methodology under the SAE perspective: to identify assumptions, to replace some, to explicitly acknowledge those unreplaced. The remaining work is left to successors.

---

Appendix A: Low-Dimensional Concrete Ledger Calculations

This appendix provides the unfolded calculations of the three diagnostic tools and three applications of §6, on low-dimensional concrete examples. The purpose is to enable working geometers to see directly what the ρ-defect ledger / ρ-signature / readout-first principle look like on familiar objects, so they may attempt to apply these tools in their own research.

The contents of the appendix are demonstrative rather than new mathematical results—the Gauss-Bonnet calculations in A.1 are standard textbook material since the nineteenth century; the neckpinch framework in A.2 is based on the existing Hamilton-Perelman theory; the Heisenberg-group abnormal geodesics in A.3 are based on the Liu-Sussmann 1995 construction. The work of this appendix is to re-express these known examples in the vocabulary of the ρ-defect ledger, demonstrating how the SAE framework manifests in concrete computation.

A.1 Low-Dimensional Calculation of the Gauss-Bonnet ρ-Defect Ledger

A.1.1 The Unit Sphere $S^2$

Set $M = S^2$ with the standard metric, $K \equiv 1$. No boundary, $\partial M = \varnothing$.

Choose the invariant family $\mathcal{I} = \{\chi\}$. The Gauss-Bonnet ledger reads:

$$\frac{1}{2\pi}\int_M K\, dA = \chi(M) - \frac{1}{2\pi}\int_{\partial M} k_g\, ds.$$

Substituting: $\int_{S^2} K\, dA = 4\pi$, and $\partial M = \varnothing$ gives $\int_{\partial M} k_g\, ds = 0$. Hence

$$\frac{4\pi}{2\pi} = 2 - 0,$$

i.e., $2 = \chi(S^2) = 2$. The ledger closes completely, $\varepsilon_\chi = 0$.

ρ-signature reading: $S^2$ has no ascent loci; all curvature manifests as readout-visible mode in the interior geometric remainder, and the Archimedean exchange factor $2\pi$ wholly converts the continuous reading into the discrete integer $\chi = 2$. This is the simplest manifestation of the ledger.

A.1.2 The Standard Disk $D^2$ (Boundary Example)

Set $M = D^2$ as the unit Euclidean disk, $K \equiv 0$ (flat metric), $\partial M = S^1$, with boundary geodesic curvature $k_g \equiv 1$ (unit circle). $\chi(D^2) = 1$.

Ledger:

$$\frac{1}{2\pi}\int_{D^2} K\, dA + \frac{1}{2\pi}\int_{S^1} k_g\, ds = \chi(D^2)$$

$$0 + \frac{1}{2\pi} \cdot 2\pi = 1.$$

Perfect closure, $\varepsilon = 0$.

ρ-signature reading: $D^2$ has one ascent locus $s = \partial M$ (boundary), with ρ-signature $\rho_{\chi, s} = 1$—the boundary releases the entire value of $\chi$. The interior geometric remainder is zero (flat); all $\chi$ information flows from the boundary remainder. This aligns with the readout-first principle of §6.4: in the flat $D^2$ case, the readout of $\chi$ is wholly determined by the boundary $k_g$ term, not by interior curvature.

A.1.3 The Torus $T^2$ and the Connected Sum $T^2 \# T^2$

Set $T^2$ with the flat metric ($K \equiv 0$), no boundary. The ledger gives $0 = \chi(T^2)$, i.e., $\chi(T^2) = 0$. Perfect closure. The ρ-signature on $T^2$ is trivial—neither interior nor boundary remainder, no ascent loci.

The genus-2 surface $\Sigma_2 = T^2 \# T^2$ with the hyperbolic metric ($K \equiv -1$, area $4\pi$ by Gauss-Bonnet). Ledger:

$$\frac{1}{2\pi}\int_{\Sigma_2} K\, dA = \frac{-4\pi}{2\pi} = -2 = \chi(\Sigma_2).$$

Closes. Note $\chi = -2 = 2 - 2g$ confirms the operation of Gauss-Bonnet as ledger on higher-genus surfaces.

A.1.4 The Conical-Singularity Case (Class-B Example)

Consider a surface with a single conical singularity: $M = $ flat $\mathbb{R}^2$ with a conical point at the origin of total angle $\theta_0 \in (0, 2\pi)$. Smooth and flat away from the singularity, $K \equiv 0$ on $M \setminus \{0\}$.

By the definition of the Gauss-Bonnet remainder defect in §6.3:

$$\Delta_\rho(0) = 2\pi - \theta_0 \in (0, 2\pi).$$

If $\theta_0 = \pi$ (right-angle cone), $\Delta_\rho = \pi \in 2\pi \cdot (1/2) \subset 2\pi\mathbb{Q}$—Class B, signature on the rational extension of the exchange grid.

If $\theta_0 = 2\pi/n$ ($n \in \mathbb{Z}_+$, orbifold cone), $\Delta_\rho = 2\pi(1 - 1/n) \in 2\pi\mathbb{Q}$—again Class B.

The ρ-signatures of all standard conical singularities (orbifold types, rationally-tilted cones) fall into Class B. Class C singularities require $\theta_0 / 2\pi$ to be transcendental; standard geometric constructions cannot naturally produce such singularities—this is the concrete grounding of Problem 6.4 in §6.3.

A.2 The ρ-Signature Framework Example for Ricci-Flow Surgery

A.2.1 The Standard Neckpinch: Brief Introduction

Consider Ricci flow on a three-dimensional manifold undergoing a standard neckpinch surgery. The typical scenario in the Hamilton-Perelman program: the manifold $M_t$ forms a cylindrical neck region collapsing to zero volume; at the singular time $t_s$, the singular neighborhood is excised, with hemispherical caps glued to the two sides, yielding the topologically-changed manifold $M_{t_s^+}$.

Suppose $M_{t_s^-}$ is a connected compact 3-manifold before surgery, while $M_{t_s^+}$ splits into two connected components after surgery (the typical neckpinch divides one manifold into two "dumbbell heads"). Then under the invariant family $\mathcal{I} = \{\chi, b_0\}$ (where $b_0$ is the number of connected components):

$$\Delta\chi_s = \chi(M_{t_s^+}) - \chi(M_{t_s^-})$$

$$\Delta b_0 = b_0(M_{t_s^+}) - b_0(M_{t_s^-}) = 1$$

The concrete values depend on the specific surgery type (e.g., 3-sphere-type surgery gives different $\Delta\chi$ than lens-space-type surgery), determined precisely by the local model of the surgery (neckpinch, degenerate neckpinch, etc.).

A.2.2 The Hierarchical Form of the ρ-Signature on the W-Functional

By Conjecture 6.6 of §6.5, the W-functional jump $\Delta\mathcal{W}_s$ at the surgery moment should be a discrete value controlled by $(\Delta\chi_s, \Delta b_0, \tau_s, \ldots)$. This appendix does not provide the concrete numerical value of $\Delta\mathcal{W}_s$—doing so would require complete calculation within the existing Bamler entropy framework, involving specific dimensional normalization and surgery-type classification, beyond the burden of this paper. The appendix provides the structural ledger form:

$$\mathcal{W}(g_{t_s^-}) = \mathcal{W}(g_{t_s^+, \mathrm{reg}}) + \Delta\mathcal{W}_s + \varepsilon_\mathcal{W}.$$

In the SAE framework, Conjecture 6.6 holds that $\varepsilon_\mathcal{W} = 0$ (strict equality, quantization form), whereas Bamler's existing results give $\varepsilon_\mathcal{W} \neq 0$ but bounded (inequality form). The genuine difference between these two claims is precisely what the conjecture leaves for future falsifiable testing: under reasonable normalization, is $\Delta\mathcal{W}_s$ a continuous value or quantized by topological data?

A.2.3 Comparison with the Static RCD Level

By Conjecture 6.7 of §6.6, the ρ-signature on the geometric-flow side should be dual to the ρ-signature on statistical-manifold phase transitions. At the static RCD level, Lott-Villani-Sturm have established the precise equivalence $\mathrm{Ric}(g) \geq K \iff K\text{-convexity}$. At the dynamical level, the conjecture predicts:

$$\Delta\mathcal{W}_s^{\mathrm{Ricci}} \longleftrightarrow \Delta S_s^{\mathrm{phase}}$$

The W-functional jump on the Ricci-flow side should be dual to the entropy jump on the statistical-phase-transition side. The concrete form of this duality (functor, natural transformation, derived correspondence) awaits formalization, but its structural prediction is already available for borrowing in P2 fluid mechanics as a unified framework for turbulence bursts and statistical phase transitions.

A.3 The ρ-Signature Framework for Abnormal Geodesics on the Heisenberg Group

A.3.1 The Heisenberg Group as the Simplest Non-Trivial Sub-Riemannian Geometry

The three-dimensional Heisenberg group $\mathbb{H}^3$ is the most classical non-trivial example of sub-Riemannian geometry. It is the Lie group $\mathbb{R}^3$ equipped with a left-invariant rank-2 distribution: in standard coordinates $(x, y, z)$, the distribution $\mathcal{D}$ is spanned by two vector fields,

$$X = \partial_x - \frac{y}{2}\partial_z, \quad Y = \partial_y + \frac{x}{2}\partial_z.$$

Note $[X, Y] = \partial_z$, i.e., the $z$-direction "lies outside the distribution" but is reachable via Lie bracket—the most-studied feature of the Heisenberg group (Hörmander condition / bracket-generating). The sub-Riemannian metric is determined by orthonormalizing $X$ and $Y$; only curves along $\mathcal{D}$ have defined length, while instantaneous displacement along $\partial_z$ has no defined length (since it lies outside the distribution).

The shortest-path problem on the Heisenberg group has been systematically studied by Brockett, Hsu, Beals-Gaveau-Greiner, and others; the standard Hamiltonian geodesics are given by the transversal extremality principle of the Hamilton equations. The point is: all shortest geodesics on the Heisenberg group arise from the normal Hamiltonian extremality principle. On the standard Heisenberg group there are no abnormal geodesics.

A.3.2 Abnormal Geodesics in Higher-Step Sub-Riemannian Structures

Abnormal geodesics arise in sub-Riemannian structures of higher step (i.e., sub-Riemannian structures where bracket-generating requires more layers of Lie bracket). Liu-Sussmann in 1995 constructed concrete strict-shortest abnormal geodesics on Engel / Goursat distributions (rank-2 distributions on 4-manifolds, bracket-generating step 3). The concrete properties of these examples (smoothness, uniqueness, stability) remain partially open.

A.3.3 The ρ-Signature Framework Reading of Abnormal Geodesics

By SAE Rereading 6.8 of §6.7, abnormal geodesics in the ρ-defect-ledger framework are read as the nonzero component of the ρ-signature on the geodesic-structure invariant under constraint ascent (from rank-$n$ full-Riemannian to rank-$k$ sub-Riemannian). Concretely:

Set $X_t = $ the unconstrained full-Riemannian object ($\mathcal{D} = TM$), $X_\infty = $ the Engel/Goursat sub-Riemannian object ($\mathcal{D} \subsetneq TM$). Choose $\mathcal{I} = \{\text{normal geodesics}, \text{abnormal geodesics}, \text{cut locus}, \text{minimizer set}\}$. Ledger:

$$\text{geodesic structure}(X_t) = \text{normal geodesics}(X_\infty) + \text{abnormal geodesics}(X_\infty) + \varepsilon$$

Abnormal geodesics are the nonzero component of the ρ-signature of this ascent on geodesic structure. The difference between the Heisenberg group ($\varepsilon = 0$ because no abnormal geodesics exist) and the Engel/Goursat distribution ($\varepsilon \neq 0$ but explicitly carried as the abnormal-geodesics term) is precisely the SAE-framework manifestation of how "the step-number of the constraint ascent" determines whether the ρ-signature has an abnormal component.

This framework does not resolve the Sussmann puzzle (the concrete properties of abnormal geodesics remain open), but provides for existing research a unified coordinate: from the SAE perspective, abnormal geodesics are not pathological exceptions but legitimate ρ-signature components necessarily produced by constraint ascent at step higher than 2. The reason the Heisenberg group "has no abnormal geodesics" is not its peculiarity but the fact that step-2 constraint ascent does not suffice to produce abnormal ρ-manifestation; constraint ascent at step 3 and above produces it as legitimate bookkeeping for deeper-level constraint.

The concrete computation of the abnormal ρ-signature (explicitly writing out the abnormal-geodesics' invariant contribution on Engel distributions) is an independent project, left for sub-Riemannian-geometry treatments. The appendix establishes the framework; specific calibration is left for subsequent work.

---

References

[1] B. Riemann, Über die Hypothesen, welche der Geometrie zu Grunde liegen, Habilitationsvortrag, Göttingen, 1854; published posthumously in Abhandlungen der Königlichen Gesellschaft der Wissenschaften zu Göttingen, 13 (1868), 133–150.

[2] H. Qin, Self-as-an-End, Paper 1: Ontological Foundation of Moral Law, Zenodo, 2025. DOI: 10.5281/zenodo.18528813.

[3] H. Qin, Self-as-an-End, Paper 2: Chisel-Construct Cycle and Methodological Foundation, Zenodo, 2025. DOI: 10.5281/zenodo.18666645.

[4] O. Bonnet, Mémoire sur la théorie générale des surfaces, Journal de l'École Polytechnique, 19 (1848), 1–146.

[5] S. S. Chern, A simple intrinsic proof of the Gauss-Bonnet formula for closed Riemannian manifolds, Annals of Mathematics, 45 (1944), 747–752.

[6] R. Bott and L. W. Tu, Differential Forms in Algebraic Topology, Graduate Texts in Mathematics 82, Springer, 1982.

[7] F. Lindemann, Über die Zahl π, Mathematische Annalen, 20 (1882), 213–225.

[8] M. Kontsevich and D. Zagier, Periods, in Mathematics Unlimited—2001 and Beyond, Springer, 2001, 771–808.

[9] M. F. Atiyah and I. M. Singer, The index of elliptic operators on compact manifolds, Bulletin of the American Mathematical Society, 69 (1963), 422–433.

[10] R. Apéry, Irrationalité de ζ(2) et ζ(3), Astérisque, 61 (1979), 11–13.

[11] R. S. Hamilton, Three-manifolds with positive Ricci curvature, Journal of Differential Geometry, 17 (1982), 255–306.

[12] G. Perelman, The entropy formula for the Ricci flow and its geometric applications, arXiv:math/0211159, 2002.

[13] G. Perelman, Ricci flow with surgery on three-manifolds, arXiv:math/0303109, 2003.

[14] G. Perelman, Finite extinction time for the solutions to the Ricci flow on certain three-manifolds, arXiv:math/0307245, 2003.

[15] B. Kleiner and J. Lott, Singular Ricci flows I, Acta Mathematica, 219 (2017), 65–134.

[16] R. H. Bamler, Convergence of Ricci flows with bounded scalar curvature, Annals of Mathematics, 188 (2018), 753–831.

[17] H. Qin, ZFCρ Series, Paper 71: Structural Statement of the H' Indissolvability Theorem, Zenodo, 2026. DOI: 10.5281/zenodo.20011553.

[18] H. Qin, ZFCρ Series, Paper 72: Spectral Bridge Front II, Zenodo, 2026. DOI: 10.5281/zenodo.20029941.

[19] K. Gödel, Über formal unentscheidbare Sätze der Principia Mathematica und verwandter Systeme I, Monatshefte für Mathematik und Physik, 38 (1931), 173–198.

[20] V. I. Arnold, Mathematical Methods of Classical Mechanics, Graduate Texts in Mathematics 60, Springer, 1989.

[21] S. Amari, Differential-Geometrical Methods in Statistics, Lecture Notes in Statistics 28, Springer, 1985.

[22] S. Amari and H. Nagaoka, Methods of Information Geometry, Translations of Mathematical Monographs 191, American Mathematical Society, 2000.

[23] D. Burago, Y. Burago, and S. Ivanov, A Course in Metric Geometry, Graduate Studies in Mathematics 33, American Mathematical Society, 2001.

[24] A. Connes, Noncommutative Geometry, Academic Press, 1994.

[25] V. Voevodsky, A¹-homotopy theory, Documenta Mathematica, Extra Volume ICM 1998, I, 579–604.

[26] S. Mac Lane and I. Moerdijk, Sheaves in Geometry and Logic: A First Introduction to Topos Theory, Springer, 1992.

[27] The Univalent Foundations Program, Homotopy Type Theory: Univalent Foundations of Mathematics, Institute for Advanced Study, 2013.

[28] S.-T. Yau, On the Ricci curvature of a compact Kähler manifold and the complex Monge-Ampère equation, I, Communications on Pure and Applied Mathematics, 31 (1978), 339–411.

[29] H. Qin, SAE Physics Series: Four Forces Paper 1, Zenodo, 2026. DOI: 10.5281/zenodo.19342106.

[30] H. Qin, SAE Information Theory Series, Paper 4: Black Hole Information via Causal Spectrum, Zenodo, 2026. DOI: 10.5281/zenodo.19880111.

[31] H. Qin, SAE Moral Law Series, Paper 1: Foundational Theorems, Zenodo, 2026. DOI: 10.5281/zenodo.20011018.

[32] K. Uhlenbeck, Connections with $L^p$ bounds on curvature, Communications in Mathematical Physics, 83 (1982), 31–42.

[33] J. Cheeger and T. Colding, On the structure of spaces with Ricci curvature bounded below. I, Journal of Differential Geometry, 46 (1997), 406–480.

[34] J. Lott and C. Villani, Ricci curvature for metric-measure spaces via optimal transport, Annals of Mathematics, 169 (2009), 903–991.

[35] K.-T. Sturm, On the geometry of metric measure spaces. I, Acta Mathematica, 196 (2006), 65–131.

[36] F. Otto and C. Villani, Generalization of an inequality by Talagrand and links with the logarithmic Sobolev inequality, Journal of Functional Analysis, 173 (2000), 361–400.

[37] H. J. Sussmann, A cornucopia of four-dimensional abnormal sub-Riemannian minimizers, in Sub-Riemannian Geometry, Progress in Mathematics 144, Birkhäuser, 1996, 341–364.

黎曼方法论的当代接续与几何余项的承认

(工作标题待定)

---

摘要

本文接续黎曼一八五四年就职演讲《论几何学所基于的假设》的方法论传承,在 SAE(Self-as-an-End)框架下对黎曼几何做一次显式的假设审查。本文不试图建构 SAE 几何学的公理化体系,而是作为「形与流」系列的开篇,为后续若干 D 层级(物理 4D、生命演化 5D-6D、生物种群 7D-8D、主体动力学 9D+)上的几何与动力学讨论提供基础语汇与本体论锚点。本文立两条核心命题:其一,把 Gauss-Bonnet 公式重读为流形上四重几何余项之间的精确兑换律,并给出可证伪的标准归一化形式陈述;其二,把 Hamilton-Perelman 的 Ricci flow 程序重读为光滑流形范畴内消解尝试的失败与几何余项的强制升层,作为几何侧的 H' 型不可消解定理实例。本文同时立下一条 SAE 元层命题——统一不是消灭余项,而是迁移余项——作为对几何全域不可被任何单一框架完整收拢这一现象的本体论指认。在此基础上,本文把上述本体论命题操作化为一组面向几何学家的可重复使用诊断工具:ρ-defect ledger(余项账本)、ρ-signature(奇点指纹)、readout-first principle(读数优先原则)。在此诊断框架下,本文进一步提出三条 SAE 视角下的具体应用——作为 conjectural applications 与开放研究问题:Ricci flow surgery 中 Perelman W-泛函突变的量子化猜测、几何流与信息几何相变之间的耗散对偶猜想、次黎曼几何反常测地线作为 ρ 拓扑打穿的重读。这三条不与两条核心命题等权,而是 SAE 视角对几何学家发出的具体研究邀约,具体形式化与可证伪检验留待几何学共同体。本文显式审查黎曼几何遗留的五条 ρ=∅ 假设(光滑性、整数维、张量度规、实数背景、点集基底),替换前三条,显式留下后两条作为本代不能动的边界,留给后人。本文的方法论姿态本身——指认假设、替换部分、显式承认未替换部分——构成对"余项不可消"这一 SAE 核心本体论命题的具身证据。

关键词: 黎曼几何;Gauss-Bonnet 公式;Ricci flow;余项不可消;几何 H';ρ-defect ledger;Perelman W-泛函;信息几何;假设松动偏序;SAE 框架

---

1. 引言:黎曼传承的方法论接续

1.1 黎曼一八五四年就职演讲及其方法论遗产

一八五四年六月十日,黎曼在哥廷根做了他的就职演讲,题目是《论几何学所基于的假设》(Über die Hypothesen, welche der Geometrie zu Grunde liegen)[1]。这场演讲的听众主要是高斯,演讲的内容只在黎曼去世后才正式发表(一八六八年)。这一篇短短的文本后来被几乎所有的几何史叙事认定为现代几何学的起点。

它做的事情其实只有一件:指认几何学所基于的若干假设,并提议替换其中的一条。被替换的那一条是空间的平直性——欧氏几何把它当作不证自明的前提,黎曼把它降格为一种特殊情形,代之以更一般的"度规可变"。这一替换打开了一整个非欧几何的工作场域,后来由 Ricci、Levi-Civita、Einstein 接续,直到今天仍在被各种新方向(Finsler、sub-Riemannian、Lorentzian、分形、非交换、p-adic、tropical、信息几何)继续延展。

但是这一篇就职演讲值得被反复重读的理由,不只是它替换了哪一条具体假设,而是它示范了一种方法论:几何学不是一座完工的建筑,而是一组始终需要被重新审视的假设。每一代几何学家的工作都不是给这座建筑添砖加瓦,而是首先指认前一代默认了什么、为什么默认、是否仍然必要;然后在能够替换的地方替换,在不能替换的地方显式承认,留给下一代继续推进。黎曼自己用的词是 "Hypothesen"——假设——他非常清楚他的工作不是揭示终极真理,而是指认假设并替换其中一部分。

1.2 本文的定位

本文正是接续这条传承的当代尝试。本文之所以选择把标题接续黎曼那篇就职演讲,并非要冒认与黎曼并列的位置,而是要把所做的事情显式地嵌入这一脉传承之中:黎曼指认了一条假设并替换了它,后来的几代几何学家各自指认了别的假设并替换了若干,本文接续这一动作,在 SAE 视角下再做一次假设审查,替换其中能够替换的几条,显式承认其中本代尚不能动的几条,留给后人。本文之所以是"之二",不是因为本文的工作量与原作相当,而是因为本文拒绝把黎曼之后的全部工作都归并入"之一"——本文试图把"几何学所基于的假设"这一审查动作本身重新激活,作为一种方法论的延续,留下"之三""之四"等更多的位置给未来。

1.3 关于 SAE 框架与"承认 ρ"读法的回溯性说明

需要在一开始就显式说明的一点是:本文所采用的"承认 ρ"这一框架,是 SAE(Self-as-an-End)从其本体论立场出发对几何史的一次回溯解读,不是黎曼自己曾经使用过的语汇,也不是黎曼自己曾经持有的立场[2,3]。黎曼一八五四年实际关心的核心问题之一是几何的物理可应用性——他在演讲的最后一节明确提出,空间的度规关系应当由"作用于其上的束缚力"来决定,这一思想后来直接启发了爱因斯坦的广义相对论。这是一种把数学几何与物理动力学耦合在一起的洞察,与"承认 ρ"的本体论命题在精神上有相通之处,但并不能被直接等同。

本文所做的不是把现代框架强加于黎曼,而是承认 SAE 的回溯解读是一次有方向、有立场的重读——SAE 看见黎曼工作中的某一面,把它命名为"承认几何余项",但这种命名本身是 SAE 的工作,黎曼并不为这种命名负责。

这一显式标注非常重要,因为它决定了本文的姿态。本文不是站在黎曼之外宣告"黎曼承认了 ρ 但还不够",也不是站在黎曼之上宣告"SAE 完成了黎曼未竟之业",而是站在黎曼指认假设的方法论传承之内,做与黎曼相同的事情:指认本代仍在默认的假设,替换能够替换的若干,显式承认不能替换的若干。这是一次接续,不是一次超越。

1.4 「形与流」系列与本文的承担

本文承担的是「形与流」系列的开篇任务。「形与流」是一个独立的 SAE 系列,处理的是几何与动力学这一侧面在 SAE 各 D 层级上的展开:2D 数学的流形祖型(本文)、4D 物理的连续场回归(后续 P2,流体力学)、5D-6D 生命演化动力学(P3)、7D-8D 生物种群动力学(P4)、9D 以上主体动力学(P5)、因果槽作为光滑化算子(P6)、以及最终回收的总收篇(P0)。本系列与已有的 SAE 物理系列、SAE 信息论系列、SAE 生物学注、SAE 道德律系列、人类总构系列共享同一框架,在不同侧面上平行展开,各系列之间不构成层级关系。

作为系列开篇,本文的承担有五件事:

(一)把黎曼定位在"承认几何余项"的方法论谱系之中,并显式标注这一定位是 SAE 回溯解读;

(二)立下两条核心命题——把 Gauss-Bonnet 公式重读为四重余项之间的兑换律(§2),把 Ricci flow 程序重读为光滑流形范畴内消解尝试的失败与升层承认(§3);

(三)为后续 P2 至 P5 立下可反复回指的几何语汇——度规、连接、曲率、测地线、奇点、守恒在 SAE 视角下的重读(§4);

(四)显式列出本代尚不能替换的几条假设,说明为什么不能替换,并立下 SAE 元层命题(§5);

(五)把以上本体论命题操作化为一组面向几何学家的可重复使用工具,并给出三条 SAE 视角下的具体应用,使本文的贡献在几何学家的研究实践中可立即被使用(§6)。

本文明确不承担的事情包括:SAE 几何学的公理化、范畴论意义上的形式化、复几何/Kähler 这一格的系统处理、谱几何与指标定理的 SAE 重读。这些工作各自有其独立的体量,本文不试图越位完成。

---

2. 命题 A:Gauss-Bonnet 作为四重余项的兑换律

2.1 公式与标准解读

二维几何学最古老也最完美的恒等式之一是 Gauss-Bonnet 公式 [4,5]。对于一个紧致有向的二维黎曼曲面 $M$,带光滑边界 $\partial M$,有

$$\int_M K \, dA + \int_{\partial M} k_g \, ds = 2\pi \chi(M).$$

其中 $K$ 是 Gauss 曲率,$k_g$ 是边界上的测地曲率,$\chi(M)$ 是流形的 Euler 示性数,$dA$ 与 $ds$ 分别是面积形式与弧长形式。若边界为分段光滑而非全光滑,则需要在右边加上各角点处的转角项,本文为表述清晰起见只处理光滑边界情形,角项的 SAE 重读由相同框架直接推广。

这条公式的标准解读是:它把局部曲率(微分几何对象)与整体拓扑(代数拓扑对象)用一个超越数 $2\pi$ 联系起来。本文要做的不是替换这一标准解读,而是在 SAE 视角下对它做一次重读,指认这条公式所表达的本体论结构。

2.2 四重余项的本体论指认

命题 2.1(Gauss-Bonnet 作为四重余项兑换律). Gauss-Bonnet 公式可以在 SAE 视角下表达为四重余项之间的精确兑换律:

$$\frac{\text{几何曲率余项} + \text{边界余项}}{\text{Archimedean 圆周单位}\, 2\pi} = \text{拓扑整数余项}.$$

公式两边的四个量分别对应四种不同侧面的几何余项,公式则给出这四种余项之间的精确兑换关系。

第一种是几何曲率余项,对应公式左边第一项 $\int_M K \, dA$。这一项是局部曲率的全局积分,代表着流形在每一点上"无法保持平直"的本体论阻力——如果空间是欧氏的,这一项恒为零;空间一旦弯曲,这一项就以非零积分的形式显形出来。这正是黎曼所承认的那一项余项:欧氏几何假设空间是平直的,黎曼指认这一假设的失败,并把度规的可变性作为新的工作平台。曲率作为几何余项的祖型,在这一公式里被精确地以积分形式记账。

第二种是边界余项,对应公式左边第二项 $\int_{\partial M} k_g \, ds$。这一项的存在意味着:当流形并不闭合时,本应当属于内部几何余项的曲率,有一部分"溢出"到了边界上,表现为测地线的强迫弯曲。边界余项与几何余项是同一种本体论阻力的不同显形——一个在内部,一个在边界——但二者并非各自独立的量。Gauss-Bonnet 公式所表达的是,大自然允许这两种余项之间发生连续的流动交换:同一个流形,如果减小其内部弯曲度,边界曲率就会相应增加;反之亦然。这种内外交换的合法性,正是测地曲率项在公式中独立出现的本体论理由。

第三种是 Archimedean 圆周单位,对应公式右边的常数 $2\pi$。这一项的本体论地位需要谨慎处理。$2\pi$ 不是公式中"碰巧出现"的某个数,但它也不是 Gauss-Bonnet 定理"必须如此"的本体核心——这一点必须明确。事实上,Gauss-Bonnet 公式可以重新归一化写成 $\int_M e(TM) = \chi(M)$,其中 $e(TM)$ 是 Euler 类的 Chern-Weil 代表,在标准 Chern-Weil 归一化下表示为 $e(\Omega) = \frac{1}{(2\pi)^n}\,\text{Pf}(\Omega)$ [6]。这意味着 $2\pi$ 是归一化常数——它是"标准 Archimedean 实几何归一化下,曲率的实数读数进入整数拓扑读数所必须经过的兑换因子"。

$2\pi$ 的角色不是神秘奇迹,而是实连续几何与整数拓扑之间不可消的兑换率。这一余项之所以以 $2\pi$ 这一具体超越数显形,是因为我们采用了标准弧度角与标准面积形式作为度量约定;若采用不同的归一化约定,这一兑换率会以 $\pi$、$4\pi$ 或 $(2\pi)^k$ 等不同形式出现,但其作为"实/整数兑换因子"的本体论位置不变。在 SAE 视角下,这一兑换因子就是数论余项的几何显形——它不能被消去,只能被重新归一化。

第四种是拓扑整数余项,对应公式右边的 $\chi(M)$。Euler 示性数把流形的全部拓扑信息——亏格、连通性、孔洞结构——压缩为一个整数。这一整数是离散的,与连续的曲率积分形成鲜明对比。但 Gauss-Bonnet 公式恰恰说,这两侧的量在乘以一个 Archimedean 兑换因子之后是相等的。换言之:连续几何余项的总量,经过标准 Archimedean 归一化的兑换,恰好等于离散拓扑余项的整数读数。

这一兑换律不是新的数学结果——公式本身在一八四八年就由 Bonnet 完成 [4],一九四四年由 Chern 推广至高维 [5]——而是对这一已有数学结果的本体论重读。这一重读的实质贡献在于,它指认了公式中每一个量的余项归属,并把整个公式作为不同侧面余项之间精确兑换的合法性证据。

2.3 可证伪形式陈述

为了让命题 2.1 不停留于本体论解释,本文给出一条可证伪的形式陈述。

命题 2.2(标准归一化版,即命题 A1). 设采用欧氏平面诱导的标准弧度角、标准面积形式、标准 Gauss 曲率定义这一 Archimedean 归一化约定(下称"标准约定")。在这一约定下,若存在常数 $a \in \overline{\mathbb{Q}}$(即代数数)使得对所有紧致有向无边二维黎曼曲面均成立

$$\int_M K \, dA = a \cdot \chi(M),$$

则导出矛盾。

证明. 取单位球面 $S^2$,其上 Gauss 曲率 $K \equiv 1$,故 $\int_{S^2} K \, dA = 4\pi$;而 $\chi(S^2) = 2$。代入假设式得 $4\pi = 2a$,即 $a = 2\pi$。若 $a$ 为代数数,则 $\pi = a/2$ 亦为代数数,与 Lindemann 定理7矛盾。$\square$

关于命题边界的明确说明. 命题 2.2 的"标准约定"限定不可省略。若采用非标准归一化(例如重新定义角度单位使 $2\pi$ 对应到代数数倍数,或重新定义面积形式吸收 $2\pi$ 因子),则等式两边可以同为有理数甚至整数——这是合法的归一化重选,不构成对命题 2.2 的反例。本命题所断言的是:在标准 Archimedean 实几何归一化下,数论余项必然以 $2\pi$ 这一超越数形式精确显形。若放弃标准约定,这一显形位置可被搬移,但搬移过程本身必须在新约定中显式承担数论余项——它不能被消去,只能被重新定位。这正是元命题 5.3"统一即迁移"在数论余项侧面的具体体现。

这一形式陈述说明,在标准 Archimedean 归一化约定下,实数兑换因子必然以 $2\pi$ 这一超越数的形式出现——不是因为 $2\pi$ 本身不可替代,而是因为标准约定一旦给定,这一兑换因子就被锁死为 $2\pi$。任何试图在保持标准约定的同时把这一因子替换为代数数的尝试都不可能成功。Archimedean 兑换余项是不可消的,在这个意义上,数论余项确实在几何中精确显形——但显形的形式是兑换因子,而非孤立常数。

2.4 高维推广、兑换律族与周期理论

兑换律不是二维曲面的偶然现象。Chern 一九四四年的推广 [5] 给出 $2n$ 维 Chern-Gauss-Bonnet 公式:

$$\int_M \text{Pf}(\Omega) = (2\pi)^n \chi(M),$$

或在标准 Chern-Weil 归一化下写成

$$\int_M \frac{1}{(2\pi)^n} \text{Pf}(\Omega) = \chi(M).$$

这意味着随着流形维度的升高,实数与整数之间的兑换因子是 $2\pi$ 的整数次幂。与之相伴,Pontryagin 类、Chern 类、Euler 类等高维特征类的归一化中处处出现 $(2\pi)^k$ 与 $(2\pi i)^k$ 因子 [6]。这些归一化因子构成了一个完整的兑换律族:

观察 2.3(兑换律族). 微分形式世界中的实/复连续读数,要进入整数上同调世界,必须付出 $2\pi i$ 型的归一化代价。

这一代价的本体论身份就是数论余项的几何显形,在不同维度上以 $\pi$ 的不同次幂呈现。

需要在这里做一个重要的形式说明,以避免 SAE 视角对兑换律的一种过强读法。把"必有超越数因子"作为对所有"连续到离散"几何等式的预测是过度简化的——归一化约定本身可以选择,在特定约定下等式两边可以都是有理数甚至整数。$2\pi$ 在 Gauss-Bonnet 中的出现依赖标准 Archimedean 实几何归一化,不是几何结构本身的内禀不变量。更稳妥的本体论指认,应当是 de Rham 上同调与 Betti 上同调之间的比较系数——即周期(period)。Kontsevich 与 Zagier 的周期理论 [8] 把这一类常数(包括 $2\pi i$、$\zeta$ 函数特殊值、algebraic periods 等)系统化为一个与代数几何深度耦合的数论对象族。在 SAE 视角下,周期是数论余项在几何中的标准显形位置——它不依赖于约定选择,而是几何结构本身的不变量。

把这一观察形式化为一条指向性陈述:

指向性命题 2.4(周期作为数论余项的标准显形). 设 $X$ 是定义在 $\mathbb{Q}$ 上的代数簇,$\omega$ 是 $X$ 上的代数微分形式,$\gamma$ 是 $X(\mathbb{C})$ 中的有理同调链。周期 $\int_\gamma \omega$ 给出从代数侧(几何结构)到数论侧(超越或代数数)的兑换。在 SAE 视角下,周期所属的数域(超越/代数/有理)正是该几何结构所对应的数论余项层级:周期恒等式构成一个跨代数几何与数论的余项守恒族,Gauss-Bonnet 的 $2\pi$ 是这一守恒族在最简单情形(实流形 / Euler 类积分)下的具体显形。

本命题作为指向性陈述给出,不在本文中展开形式化论证。周期理论本身的 SAE 重读——具体的范畴论结构、周期空间的 ρ-defect ledger 形式、数论余项与几何余项在周期意义下的精确对偶——是一项独立工作,留待未来 SAE 谱几何或 SAE 数论几何专项。本文的命题 2.1、2.2、观察 2.3 中所讨论的 $2\pi$ 现象,应当理解为周期理论在最简单情形下的几何显形。

谱几何中的更深一层余项现象——Selberg 迹公式中的 $\pi^k$ 项、热核展开系数中的归一化常数、$\zeta$ 函数特殊值在指标定理与 analytic torsion 中的出现 [9]——可以视为兑换律族在频谱化版本下的延伸。这些现象指向一个更广的"谱几何余项守恒族",但本文不展开这一方向。一个值得记下的开放观察是,$\zeta(2k)$ 是有理数与 $\pi^{2k}$ 的乘积,而 $\zeta(2k+1)$ 的算术性质至今未明(Apéry 一九七八年仅证明 $\zeta(3)$ 的无理性 [10],其超越性仍未决);这或许指向数论余项在几何中尚有未被承认的更深一层。这一观察留给未来工作。

---

3. 命题 B:Ricci flow 与几何 H' 型升层承认

3.1 从静态记账到动态保守

如果 §2 的命题 A 处理的是几何余项的静态记账——曲率、边界、Archimedean 因子、整数拓扑之间的精确兑换——那么本节的命题 B 要处理的就是几何余项的动态保守性。本文主张:几何余项不仅在静态意义上不可消去,在动态意义上同样不可消去——任何试图通过几何流把曲率"光滑掉"的尝试,都会在有限时间内遭遇余项的反扑,这种反扑在 SAE 视角下不应被读为"消除尝试的失败",而应被读为"余项升入更高层结构后被新范畴承认"。这一命题以 Ricci flow 程序作为最具说明力的当代案例。

3.2 Ricci flow 程序简述

Ricci flow 由 Hamilton 一九八二年引入 [11],定义为

$$\frac{\partial g(t)}{\partial t} = -2 \, \text{Ric}(g(t)).$$

其几何直觉是让度规沿着 Ricci 曲率的负方向演化——曲率高的地方收缩,曲率低的地方扩张——使流形向"曲率更均匀"的状态趋近。在某些理想初始条件下,这种演化会把流形带入 Einstein 度规(即 $\text{Ric}(g) = \lambda g$)的状态。Hamilton 一九八二年的奠基定理证明,任何紧致三维流形若初始度规具有正 Ricci 曲率,则其归一化 Ricci flow 趋向常曲率度规。这一结果给了几何学界一个极具诱惑力的纲领:或许通过 Ricci flow,可以把任意流形的几何"光滑化",并由此读出其底层拓扑。

但这一纲领很快遇到了几何余项的反扑。在更一般的初始条件下,Ricci flow 不会把流形带入光滑的极限状态,而是在有限时间内形成奇点——某些区域曲率发散,流形局部"颈缩"成一个零测度的点。这些奇点在原始的光滑流形范畴内不能被继续处理,光滑演化在此中断。Perelman 二零零二至二零零三年的工作 [12,13,14] 指出,这些奇点并非偶然失败,而是有结构、可分类、可控的:奇点的局部模型(如 neckpinch、degenerate neckpinch)是几种可枚举的标准形式,而每一种奇点形式都精确编码了流形局部的拓扑信息。Perelman 由此设计了"奇点手术"程序——在奇点形成时切除奇异邻域、修补开口、让流继续演化——并最终通过这一程序证明了 Poincaré 猜想与 Thurston 几何化猜想。

3.3 几何 H' 读法

在进入命题之前,需要对术语"H'"做一个简短说明。"H'"是本系列在 ZFCρ 主干 [17,18] 中确立的命题类型符号,泛指一类不可消解定理——其抽象内容是:对一个原范畴 $\mathcal{C}$ 与一族消解操作 $R$,残余项 $\rho \neq 0$ 不可被完全消去。具体的 H' 型命题在不同领域有不同实例化(集合论、几何、形式系统、动力系统等),抽象模式相同(在 §3.6 给出统一形式)。本节将 Ricci flow 程序在几何范畴中的对应现象指认为一条几何 H' 型命题,即对光滑流形范畴内的"光滑化消解"操作,几何余项不可被消去,只能升入更广义的范畴中被重新承认。读者在没有阅读 ZFCρ 系列的情况下,可以把"几何 H'"直接理解为"几何范畴内的不可消解定理",不影响本节的完整性。

这是几何分析史上的一次伟大成就。在 SAE 视角下,这一成就的本体论意义可以被精确地指认为一条 H' 型不可消解定理。

命题 3.1(Ricci flow 几何 H' 读法). 光滑 Ricci flow 不能为任意初始几何提供一个无奇点的全局光滑消解过程。一般情形下,几何余项必然在有限时间内以奇点的形式重新显形;而消解过程要继续,就必须升入比原始光滑流形范畴更广义的对象范畴——包含奇异时空、手术、弱流、拓扑跳变在内的扩展范畴。在这一升层后的范畴中,几何余项重新被承认,以新的形式继续被处理。

这一命题的本体论实质可以表达为:几何余项不被消灭,而是被强制升层。原始光滑流形范畴内的消解失败,不意味着余项被消去——恰恰相反,余项以更尖锐的形式(奇点)重新显形,并迫使观察者进入一个更广义的范畴去处理它。Perelman 的奇点手术不是"消除了余项",而是"把余项从光滑流形层级提升到了带手术的奇异时空层级,并在新层级承认它"。

需要在技术表述上保持精确。命题 3.1 中所表述的"几何余项必然在有限时间内以奇点形式重新显形"指的是一般情形,不是字面上对每一个非 Einstein 流形的断言——Ricci flow 在不同维度、不同归一化、不同曲率符号下的具体行为差异颇大,二维曲面的 Ricci flow 长期行为依 Euler 示性数不同而分类,负曲率 Einstein 度规的 unnormalized 流呈扩张,等等。准确的技术表述应该是:Ricci flow 不能为一般初始几何提供一个仍停留在原始光滑流形范畴内、不引入额外结构的全局消解过程。Einstein 度规和 Ricci soliton(在微分同胚与尺度变换模掉之后的自相似解)是 Ricci flow 在这种意义上的"固定点",但这些固定点构成的子流形并不覆盖一般初始条件——一般情形下,流必然要离开原始光滑范畴。

3.4 升层后的当代研究方向与开放预测

这一升层的当代研究方向已经形成一个独立的子领域。Perelman 的奇点手术程序之后,几何分析界进一步发展了"穿过奇点的 Ricci flow"理论——构造在奇点处仍然定义的更广义对象,使流形可以在不依赖人工切除的情况下连续穿越奇点 [15,16]。这些方向的具体工作不是本文要展开的,但它们的存在恰好支持 SAE 的升层读法:广义流可以存在,但必须引入额外的对象与结构(奇异时空、measure-weak 结构、拓扑跳变等)。换言之,几何余项的不可消并不意味着"任何处理都失败",而是意味着"任何处理都必须以某种方式升入更广义的范畴"。这是一个比"必然失败"更强、更精确、也更符合现代几何分析现状的命题。

由此可以提出一条与现有研究方向相容的开放预测:

猜想 3.2(原始范畴不可闭合预测). 不存在一种仍停留在原始光滑流形范畴内、同时不引入额外结构(奇异时空、手术、弱流、拓扑跳变等)的全局 Ricci flow 消解理论。

这一猜想可以被未来工作证伪——若有人构造出仅在原始光滑流形范畴内运作的全局消解理论,则 SAE 升层命题需要重新审视。

3.5 与 ZFCρ H' 的结构性类比

命题 3.1 与本系列已有的 ZFCρ 主干中的 H' 命题之间存在结构性类比 [17,18]。ZFCρ H' 处理的是集合论侧面的现象:任何试图通过添加大基数公理来"封闭"集合论的尝试,都会在新公理被加入之后,以新的结构性缺口的形式重新暴露——余项不被消灭,只被推到更高层级。命题 3.1 所描述的几何侧面现象在抽象结构上与之相似:任何试图通过几何流来"光滑化"曲率的尝试,都会在有限时间内以奇点的形式重新暴露——余项不被消灭,只被推到更广义的范畴。两者共享的抽象模式可以表述为:

抽象模式 3.3(H' 型不可消解模式). 给定对象范畴 $\mathcal{C}$,消解操作 $R$,被试图消解的结构缺口 $\Phi$,以及在 $R$ 作用之后的余项 $\rho(R^t X)$。H' 型命题断言,对所有 admissible 的消解操作 $R$,余项 $\rho \neq 0$ 不可被完全消去。

这一抽象框架在不同领域的具体实例化中,$\mathcal{C}$ 可以是集合论模型、光滑流形、形式系统、动力系统;$R$ 可以是公理扩张、几何流、形式化过程、相空间演化;但抽象模式相同。

本文愿意把这一类比指认为"结构性类比"(structural analog),而非"范畴论意义上的严格同构"——后者需要给出明确的范畴、函子与自然变换,使两侧的 H' 成为同一对象的两个表示。这种严格形式化是一项独立的工作,不在本文承担之内。本文所主张的是:几何 H' 与 ZFCρ H' 共享同一抽象模式,这一共享的形式化兑现作为一项承诺,留给系列总收的 P0。

3.6 余项持续性家族:其他先例

需要附带提到的是,这种"消解尝试 → 余项升层"的抽象模式在数学的多个领域都有先例。Gödel 不完备定理 [19]、Tarski 真值不可定义性、Turing 不可判定性,描述的都是形式系统试图把"自身的真值/可证性/可判定性"完全内化时遭遇的余项;KAM 定理 [20] 处理的是不可积哈密顿系统在小扰动下不变环面的命运,数论侧的丢番图条件强制大部分环面持续存在而部分被撕裂;分歧理论与 catastrophe 理论指认退化点不是病态而是分类对象;Morse 理论把临界点读为结构信息;因果集(causal set)理论把时空连续体读为底层离散因果结构的宏观近似,几何流形对应的离散撒点动力学中,光滑过渡与拓扑跃迁恰好对应连续侧的奇点形成与升层。这些不是同一个定理,但它们可以被指认为"余项持续性家族"——同一种本体论模式在不同数学子领域的多次显形。本文不展开这一家族的统一论证,但把它列为 SAE 几何与 SAE 数学未来工作的方向。

---

4. 几何语汇与信息几何

4.1 几何概念的 SAE 重读

本文的前两节立下两条核心命题。本节要做的是为「形与流」系列后续提供可反复回指的几何语汇。后续 P2 至 P5 处理的对象——流体、生命演化、生物种群、社会主体——都不是几何系内部的对象,但它们都可以在合适的状态空间上被几何化处理。要让几何语汇能在这些非传统对象上发挥作用,需要先把若干基本概念从纯形式定义中解放出来,赋予 SAE 视角下的本体论读法。

度规(metric)在传统读法中是"两点之间的距离",在 SAE 视角下应当被读为"在某一层级中差异被读出的代价结构"。物理空间的度规度量长度,但相空间的度规可以度量动力学距离,统计流形的度规度量分布的可区分性,信念空间的度规度量观点变换的成本。后续各 D 层级各自有其度规对象,但所有度规都共享同一本体论身份:它是该层级中余项被显形为代价的方式。

连接(connection)在传统读法中是"平行移动的规则",在 SAE 视角下应当被读为"跨位置比较同一对象的规则"。在物理空间上,连接告诉我们如何把一个向量从一点平移到另一点;在生命演化的状态空间上,连接告诉我们如何在不同基因型之间比较同一个表型属性;在主体动力学的语境空间中,连接告诉我们如何在不同语境之间识别"同一立场"。连接是后续每一层级都必须显式构造的对象,而不同层级所选择的连接形式恰好揭示了该层级的特征。

曲率(curvature)在 SAE 视角下应当被读为"局部闭合失败的可积累读数"。这是 §2 与 §3 中已经在使用的读法:曲率不是"弯曲程度"这种几何直观,而是"试图沿闭合路径回到出发点而失败的程度"。后续每一层级中,只要存在某种"局部循环失败"的现象,就有相应的曲率对象。流体力学中的涡度(vorticity)是速度场的曲率;演化动力学中的"适应度循环不闭合"是适应度景观的曲率;主体动力学中的"信念回路不一致"也是某种曲率。

测地线(geodesic)在 SAE 视角下应当被读为"在给定代价结构下的自然路径"。传统读法把测地线等同于"最短路径"是过度狭窄的——在 Lorentzian 几何中测地线可以是"最长"的(类时测地线极大化固有时);在信息几何中测地线对应分布族中的某种"最自然变换"。本文主张把测地线一般地读作"局部闭合条件下的自然演化路径",这个读法在后续各层级都能适用。

奇点(singularity)的 SAE 读法已经在命题 3.1 中给出:奇点是当前光滑描述不能继续吸收余项的地方,是余项强制升层显形的位置。这一读法在后续各层级都会反复使用——流体中的湍流爆发是奇点,演化中的物种灭绝事件是奇点,种群动力学中的相变跃迁是奇点,主体动力学中的范式转移是奇点。这些都不是"系统失败",而是"余项升入更高层级的承认点"。

守恒与通量(conservation and flux)在命题 2.1 中已经以 Gauss-Bonnet 公式作为模板呈现:不同侧面余项之间的精确兑换。后续 P2 流体可以直接调用这一模板处理能量、动量、涡度、熵等多重守恒之间的关系;P3 至 P5 各自也有相应的守恒-通量结构。本文把 Gauss-Bonnet 兑换律作为这一类结构的祖型留下,后续各篇可以把各自层级的守恒律与之并置参照。

4.2 信息几何作为高 D 层级的几何回归

后续 P3 至 P5 所处理的状态空间,大多不是物理空间或几何流形,而是概率分布构成的统计流形。生命演化的适应度景观可以被看作物种空间上的标量场;种群动力学的相空间是物种比例所张成的概率单纯形;主体动力学的舆论场是观点分布在信念空间上的概率密度。要让黎曼几何的语汇在这些层级延续使用,需要一个桥梁——信息几何提供了这一桥梁 [21]。

信息几何把 Fisher 信息度规

$$g_{ij}(\theta) = E_\theta\left[\frac{\partial \log p(x; \theta)}{\partial \theta^i} \cdot \frac{\partial \log p(x; \theta)}{\partial \theta^j}\right]$$

放在分布参数空间上,使该空间成为黎曼流形。在这一流形上,两点之间的"距离"度量分布的可区分性,测地线对应分布之间的最自然变换,曲率对应分布族的弯曲结构。Amari、Chentsov 等人发展的这一框架 [21,22],使得概率分布的演化可以被严格地几何化。

本文主张把信息几何视为本系列中黎曼几何在高 D 层级的一个典型回归形式——并不主张它是历史上"第一次"或"唯一"的回归(此类历史第一性的判断容易引发不必要的争议),也不主张后续所有层级都必须采用 Riemannian 形式。事实上,P3 至 P5 各层级的状态空间可能采用流形、分层空间、概率单纯形、Wasserstein 空间、度量测度空间、网络、范畴或 sheaf 等多种几何对象。本文所提供的是几何语汇,而不强制 Riemannian 模板——后续各篇可以自由选择最适合该层级的几何形式,只要保留 SAE 视角下"度规-连接-曲率-测地-奇点-守恒"这一套概念结构的本体论读法即可。

---

5. 假设审查、松动偏序与元层命题

5.1 黎曼几何的五条 ρ=∅ 假设

按照本文所接续的黎曼方法论,下一步是审查黎曼几何本身仍在默认的若干假设。本文指认五条。

第一,光滑性假设。黎曼几何把流形假定为处处光滑的,奇点被视为例外、被处理为要被切除的对象,而不是流形结构的合法组成部分。这一假设在二十世纪后半已被多个方向部分松动:stratified spaces、orbifolds、Alexandrov 空间 [23]、Ricci flow with surgery、几何测度论、奇异 Ricci 流等。SAE 视角下这一假设可以被替换为:奇点不是几何对象的失败,而是余项在该处的强制显形位置。

第二,整数维假设。黎曼几何把流形维度假定为非负整数,分形维数被视为"病态"。这一假设可以被部分松动:Hausdorff 维度、Minkowski 维度、谱维度等概念已经形式化了非整数维的合法对象;Connes 的 spectral triple 框架 [24] 在算子代数视角下重新定义了维度。SAE 视角下这一假设可以被替换为:维度从拓扑整数扩展为测度/谱/算子增长读数——但需要注意,这种扩展并非否定整数维,而是把整数维视为更广维度概念的特殊情形。

第三,张量度规假设。黎曼几何把度规假定为光滑张量场。这一假设已被多个方向部分松动:Finsler 几何允许度规为方向相关的范数,Lorentzian 几何允许度规非正定,信息几何在统计流形上引入 Fisher 度规,sub-Riemannian 几何允许某些方向不可达。SAE 视角下这一假设可以被部分替换为:度规不是给定背景,而是余项在该层级被读出的方式。本文仅立下这一词汇,不在本文中建构系统理论。

第四,实数背景假设。黎曼几何把底层数域假定为实数连续统。这一假设是黎曼几何最深的脚手架之一——它绑定了若干彼此相互独立的承诺(Cauchy 完备性、Archimedean 性质、序完备性、连续拓扑下的域结构、Cantor 连续统的不可数基数性等)。已有多个方向尝试松动它:p-adic 几何、Berkovich 空间、adic spaces、motive 理论 [25]、tropical 几何、surreal numbers、smooth infinitesimal analysis 等,但每一个方向都要为放弃实数付出某种代价(失去标准分析、失去构造性、失去某种拓扑直觉等)。本代不替换这一假设——它牵涉的不只是几何基础,还涉及数学基础与逻辑基础的整体重构,体量超出本文承担。这一假设作为本文显式留下的边界之一,留给后人。

第五,点集基底假设。黎曼几何把空间假定为首先是一个点集,几何结构(拓扑、光滑结构、度规)附着于这一点集之上。这一假设与实数背景假设不同——后者关于坐标值域,前者关于空间本体。Connes 非交换几何 [24]、locale/topos 理论 [26]、derived geometry 等方向已在松动这一假设——它们承认空间结构可以不附着于点集,而附着于代数、范畴或谱等更抽象的对象。本代亦不替换这一假设——其松动同样会牵涉过深的数学基础重构,不在本文承担之内。这一假设作为本文显式留下的另一条边界,亦留给后人。

把以上五条放在一张表里:

假设	本代处理
光滑性	替换:奇点是余项强制显形位置
整数维	替换:维度扩展为测度/谱/算子读数
张量度规	部分替换:度规即余项读出方式
实数背景	显式留下
点集基底	显式留下

这一处理本身就是本文方法论姿态的具身。本文不试图替换全部五条——那是一种把"几何全域"视为可一次性完成的对象的姿态,与"承认余项不可消"的本体论命题在结构上自相矛盾。本文所做的是显式审查、替换能替换的几条、显式留下不能替换的几条,把"未替换"本身作为方法论的内容,而不是疏忽。

5.2 假设松动偏序与嵌套结构

在数学界传统的表述中,人们有时会写"黎曼几何 $\subset$ 非欧几何 $\subset$ 几何全域",或类似的"集合论侧:ZFC $\subset$ ZFCρ $\subset$ 集合论更大类"。这种用集合包含符号 $\subset$ 来表达学科扩展关系的写法在严格意义上并不准确——非欧几何不是一个单一的集合,Finsler、Lorentzian、tropical、p-adic、非交换几何等不是黎曼几何的"超集",而是不同假设松动方向上的并列对象;ZFCρ 也不是 ZFC 的简单"包含",而是公理系统的扩展关系。

本文为表述清晰,引入"假设松动偏序"符号 $\preceq$。

定义 5.1(假设松动偏序). 设 $\mathcal{T}_0, \mathcal{T}_1$ 为两个数学理论(以公理集与默认假设的并集刻画)。称 $\mathcal{T}_0 \preceq \mathcal{T}_1$,若 $\mathcal{T}_1$ 保留了 $\mathcal{T}_0$ 的一部分操作,同时松动了某些 $\mathcal{T}_0$ 中曾被默认的 $\rho = \varnothing$ 假设。

在这一符号下,可以表达:

• 几何侧:黎曼几何 $\preceq$ 非欧几何族 $\preceq$ 几何全域
• 集合论侧:ZFC $\preceq$ ZFCρ $\preceq$ 集合论更大类

观察 5.2(松动链共享模式). 上述两条松动偏序链共享同一抽象模式——承认部分余项,然后承认更多余项,而总类无法被任何单一框架完整收拢。

这种"无法被任何单一框架完整收拢"的特性本身,正是余项不可消在元层面的精确表述。本文仅指认这一观察,不在此处展开形式化论证;两条松动链之间是否构成范畴论意义上的同构,留待 P0 处理。

5.3 元层命题:统一即迁移

由 §5.1 的五条假设处理与 §5.2 的两条松动链观察,可以提出本文所主张的一条 SAE 元层命题:

元命题 5.3(统一即迁移). 统一不是消灭余项,而是迁移余项。

这条命题需要被精确理解。它不是说"任何统一尝试必然失败"——同伦类型论 [27]、topos 理论 [26]、$\infty$-范畴等都是数学基础层面的强统一尝试,它们各自都取得了实质性的进展,本文承认这些进展。这条命题真正主张的是:任何统一框架都必须选择把余项放在哪里——它能消除一些差异,但会把另一些差异推到元层级。同伦类型论统一了类型、空间与逻辑,但引入了 universe hierarchy、coherence、computational interpretation 等新的内部边界;topos 理论统一了 set-like 宇宙的多样性,但带来了 internal logic 的多样性;$\infty$-范畴学统一了同伦不变量,但把高阶结构的复杂性推到了 coherence 数据上。每一种统一都是有代价的余项迁移,不是免费的余项消灭。

在 SAE 视角下,这就是几何全域不可被任何单一框架收拢的本体论根据。它不是几何学的某种悲观结论,而是对几何学未来工作之合法性的最强辩护——只要还有未被承认的余项,就还有新的几何学可以被开启。

---

6. 几何 ρ 的操作化:从兑换律到余项账本

6.1 从本体论到方法论:为什么需要操作化

§2 至 §5 立下了本文的本体论命题:Gauss-Bonnet 是四重余项兑换律(命题 2.1),Ricci flow 奇点是几何 ρ 的强制升层(命题 3.1),"统一即迁移"是元层命题(元命题 5.3)。这些命题共同构成 SAE 视角对几何学的本体论重读。

但本体论重读本身,对几何学家在自己的具体研究上还不是直接可用的工具。一位正在处理 Ricci flow 奇点分类、moduli space 紧化、bubble tree 收敛、derived geometry 范畴选择、weak solution defect measures 等具体问题的几何学家,即使认同 §2 至 §5 的本体论指认,也需要一套可重复使用的诊断工具才能让 SAE 视角真正进入研究实践。

本节就是把前述本体论命题操作化为一组面向几何学家的具体工具。这些工具有两个层次。§6.2 至 §6.4 是诊断框架——ρ-defect ledger、ρ-signature、readout-first principle——给几何学家在审查任何几何升层、奇点处理、范畴扩展时可以反复使用的概念结构。§6.5 至 §6.7 是三条 SAE 视角下的具体应用——Ricci flow surgery 中 Perelman W-泛函突变的量子化猜测、几何流与信息几何相变之间的耗散对偶猜想、次黎曼几何反常测地线作为 ρ 拓扑打穿的重读——展示这些诊断工具在三个不同几何子领域中如何产生具体的、可证伪的、几何学家关心的结果。§6.8 给出最小 ρ 迁移原则的 checklist 形式,作为范畴选择的方法论补充。§6.9 总结本节对几何学家的具体承诺。

需要明确指出的是:本节给出的诊断框架不是新的数学发现——defect measures、bubble tree identities、Chern-Weil boundary terms、Perelman entropy 等具体工具在几何分析文献中早已存在。本节做的是把这些散布在不同子领域的局部工具识别为同一种"余项账本"结构,并由 SAE 兑换律视角统一。它是跨子领域的诊断框架,而非任一子领域内部的新技术。

6.2 ρ-defect ledger:几何升层的余项账本

考虑一个几何升层过程

$$X_t \longrightarrow X_\infty$$

其中 $X_t$ 在原范畴 $\mathcal{C}$ 中(例如光滑流形、光滑度规、光滑嵌入曲面、带固定拓扑的代数簇等),而 $X_\infty$ 在升层范畴 $\mathcal{C}^\uparrow$ 中(例如带奇点空间、手术后空间、弱流极限、stratified space、orbifold、metric-measure 极限、derived stack 等)。这一升层可以由几何流(Ricci flow、mean curvature flow、Yang-Mills flow)、紧化(Gromov-Hausdorff 极限、bubble tree 收敛、moduli compactification)、或范畴扩展(从 schemes 到 derived schemes、从 manifolds 到 orbifolds)等具体过程实现。

选定一族几何读数或不变量 $\mathcal{I} = \{I_\alpha\}_{\alpha \in A}$。具体例子可以包括 Euler 示性数 $\chi$、signature $\sigma$、曲率积分 $\int \lvert\text{Rm}\rvert^{n/2}$、analytic index、Perelman entropy $\lambda$ 或 $\nu$、Chern numbers、Pontryagin numbers、volume growth 等。

定义 6.1(ρ-defect ledger). 对升层过程 $X_t \to X_\infty$ 与不变量族 $\mathcal{I}$,ρ-defect ledger(余项账本)是一组分解式:

$$I_\alpha(X_t) = I_\alpha(X_\infty^{\mathrm{reg}}) + \sum_{s \in \mathcal{S}(X_\infty)} \rho_{\alpha,s} + \varepsilon_\alpha,$$

其中:

• $X_\infty^{\mathrm{reg}}$ 是极限对象的正则部分(regular part),即升层后仍然在原范畴 $\mathcal{C}$ 中可以作为光滑/经典对象处理的子对象;
• $\mathcal{S}(X_\infty)$ 是升层位置集合,包含奇点、bubble 中心、neck 区域、surgery 操作位置、坍缩 stratum 等所有 $X_\infty$ 中超出 $\mathcal{C}$ 的位置;
• $\rho_{\alpha,s}$ 是该升层位置 $s$ 在不变量 $I_\alpha$ 上释放或吸收的余项,即 SAE 视角下"几何 ρ 在该位置的精确显形量";
• $\varepsilon_\alpha$ 是不可定位误差项,应当为 0、可控、或被显式标注为该升层理论尚未闭合的部分。

ρ-ledger 充分性原则. 一个几何升层理论 $\mathcal{C} \to \mathcal{C}^\uparrow$,若声称"解决"或"分类"了原范畴中的奇点/退化问题,则必须能给出该理论所关心的不变量族 $\mathcal{I}$ 上的 defect ledger。若不能给出,则该升层理论尚未完成余项记账,SAE 视角下该理论是不完整的。

这一原则不是万能定理。它不断言"所有几何升层都自动具有完整 ledger",也不断言"具有完整 ledger 的升层理论是唯一合法的"。它的精确陈述应当是:对任何声称保留或分类某选定不变量族 $\mathcal{I}$ 的升层理论,一个必要的充分性诊断是,相对于 $\mathcal{I}$ 的 defect ledger 是否存在。

与现有 defect measure / bubble tree 工具的关系. ρ-defect ledger 框架不是新数学技术,但它与几何分析中已有的 defect measure、bubble tree identity、Chern-Weil boundary terms、Uhlenbeck 紧致性、stratified intersection cohomology 等局部工具有一个关键的诊断差异。已有工具通常在具体子领域内允许"残余项"或"误差项"非零——例如,defect measure 框架允许测度本身作为"无法定位到具体奇点的残余",bubble tree identity 在某些情形下允许整体能量与单个 bubble 之和之间存在结构性差额。ρ-defect ledger 在 SAE 视角下提出一个更严格的诊断要求:$\varepsilon_\alpha$ 项的存在必须被显式标注为该升层理论尚未完成的部分——即,要么 $\varepsilon_\alpha = 0$(完全闭合),要么 $\varepsilon_\alpha$ 被显式承认为"该理论在这个不变量上的余项尚未被定位"(显式留白)。"模糊的、未被标注的、被默认掉的残余"在 SAE 诊断下是不充分的,因为它把"尚未完成"与"已完成"混在一起,使升层理论的真实承担状态变得不透明。这一更严格的诊断要求是 SAE 视角对几何分析方法论的具体贡献——不是替换已有工具,而是为已有工具提供一个承担状态的显式审查标准。

下面用四个例子说明 ledger 框架在已有几何理论中如何具体显形。

例 6.2.1(Gauss-Bonnet 作为最简单的 ledger). 命题 2.1 本身就是一个 ledger 例子。设 $X_t = M$ 是带边界的光滑曲面,$X_\infty^{\mathrm{reg}}$ 仍是 $M$ 的内部,$\mathcal{S}(X_\infty)$ 是边界 $\partial M$,不变量 $I = (1/2\pi)\int_M K\, dA$。Gauss-Bonnet 公式

$$\frac{1}{2\pi}\int_M K\, dA = \chi(M) - \frac{1}{2\pi}\int_{\partial M} k_g\, ds$$

可以读为:内部曲率读数 = 拓扑读数 - 边界释放的余项。这里 $\rho_{\partial M} = -(1/2\pi)\int_{\partial M} k_g\, ds$ 是边界位置释放的余项,$\varepsilon = 0$(光滑边界情形完全闭合)。

例 6.2.2(Ricci flow with surgery 的 ledger). 设 $X_t$ 是 Ricci flow 在 surgery 之前的光滑流形,$X_\infty$ 是 surgery 之后的流形。选 $\mathcal{I} = \{\chi, \text{Perelman } \mathcal{W}\}$。Ledger 形式:

$$\chi(X_t) = \chi(X_\infty^{\mathrm{reg}}) + \sum_{s \in \mathcal{S}_{\mathrm{surgery}}} \Delta\chi_s + \varepsilon_\chi$$

$$\mathcal{W}(g_t) = \mathcal{W}(g_{X_\infty^{\mathrm{reg}}}) + \sum_{s \in \mathcal{S}_{\mathrm{surgery}}} \Delta\mathcal{W}_s + \varepsilon_\mathcal{W}$$

每一个 surgery 位置 $s$ 释放具体的 $\Delta\chi_s$ 与 $\Delta\mathcal{W}_s$。在 §6.5 中,这一 ledger 将进一步在量子化层面被精化。

例 6.2.3(Yang-Mills bubbling 的 ledger). 设 $X_t$ 是 Yang-Mills 流在 bubbling 之前的光滑联络,$X_\infty$ 是 bubbling 极限。选 $\mathcal{I} = \{\text{energy}, c_2\}$(能量与第二 Chern 类)。Ledger 形式:

$$E(X_t) = E(X_\infty^{\mathrm{reg}}) + \sum_{\mathrm{bubbles}} E_{\mathrm{bubble}} + \varepsilon_E$$

每个 bubble 释放整数倍的"瞬子能量" $8\pi^2/g^2$,完美的 ledger(这是 Uhlenbeck 紧致性定理 [32] 的内容,SAE 视角把它识别为 ledger 的范例)。

例 6.2.4(Gromov-Hausdorff collapse 的 ledger). 设 $X_t$ 是一族 Riemannian 流形,$X_\infty$ 是其 GH 极限(可能维度降低)。$\mathcal{I}$ 包括 volume、diameter、Ricci 下界等。坍缩位置释放维度信息(从 $n$ 维流形到低维 metric-measure space)。这里 ledger 的精确形式仍是几何分析活跃方向 [33],SAE 视角指认这一方向就是在尝试给 GH collapse 一个完整 defect ledger。

6.3 ρ-signature:奇点指纹

ρ-defect ledger 的一个直接推论是,每一个升层位置 $s$ 都被赋予一个余项向量:

定义 6.3(ρ-signature). 设 $X_t \to X_\infty$ 是一个具有 defect ledger 的升层过程,选定不变量族 $\mathcal{I} = \{I_\alpha\}_{\alpha \in A}$。对每一个升层位置 $s \in \mathcal{S}(X_\infty)$,ρ-signature 是向量

$$\rho_s(\mathcal{I}) := (\rho_{\alpha,s})_{\alpha \in A}.$$

ρ-signature 是该升层位置在选定不变量族下的"指纹"——它精确刻画了该位置释放或吸收了多少哪几种余项。

ρ-signature 把奇点从对象分类升级为指纹分类。传统几何中,奇点按局部对象类型分类——锥奇点、cusp 奇点、neckpinch、bubble、orbifold point 等。这种分类本质上是按奇点的局部几何对象分类。ρ-signature 给出的是按奇点在选定不变量上的释放模式分类——同一种局部对象在不同不变量族下可能给出不同 signature,而不同局部对象在某一选定不变量族下也可能给出相同 signature。这种分类是面向研究者的具体问题的:你关心哪些不变量,ρ-signature 就告诉你奇点在这些不变量上的释放模式。

作为一个具体的 ρ-signature 应用例子,考虑二维流形孤立奇点的 Gauss-Bonnet 余项缺口分类。

Gauss-Bonnet 余项缺口分类. 对二维流形的孤立奇点 $p$,定义 Gauss-Bonnet 余项缺口为

$$\Delta_\rho(p) := \lim_{\epsilon \to 0^+}\left[2\pi\chi(M_\epsilon) - \int_{M_\epsilon} K\, dA - \int_{\partial M_\epsilon} k_g\, ds\right]$$

其中 $M_\epsilon = M \setminus B_\epsilon(p)$。在 $\mathcal{I} = \{\chi\}$ 选择下,$\Delta_\rho(p)$ 就是奇点 $p$ 的 ρ-signature。该 signature 可以落入以下三类:

• Class A($\Delta_\rho(p) \in 2\pi\mathbb{Z}$):光滑可消解奇点,signature 落在标准整数兑换网格上;
• Class B($\Delta_\rho(p) \in 2\pi\mathbb{Q} \setminus 2\pi\mathbb{Z}$):锥奇点 / orbifold 奇点,signature 落在有理兑换扩展上;
• Class C($\Delta_\rho(p) \notin 2\pi\overline{\mathbb{Q}}$):本体论奇点,signature 落在标准代数兑换之外。

问题 6.4(Class C 存在性). Class C 奇点是否存在?在标准几何构造下是否可被找到?

这是一个开放问题。所有已知的孤立奇点(锥奇点、cusp 奇点、orbifold 奇点)的 $\Delta_\rho$ 都是有理数,落入 Class A 或 Class B。Class C 奇点若存在,将是 SAE 视角下的"本体论奇点"——它不能通过任何同伦类内的光滑变形被解决到 Class A 或 Class B,标记 underlying ρ 在标准代数兑换之外的不可消显形。

本文不断言 Class C 一定存在。本文断言的是:ρ-signature 给出了"是否存在 Class C 奇点"这个具体问题——这是纯几何视角下不会自然提出的问题,只有在 SAE Archimedean 兑换律视角下才会浮出。这就是 SAE 视角对几何学的一个具体贡献:它指认了一个新的研究方向。

为什么几何学家会关心 Class C. Class C 存在性问题不只是抽象指认,它与几何学家正在做的若干具体研究方向有直接接口。

第一,moduli compactification。在曲面 moduli 空间 $\mathcal{M}_g$ 的 Deligne-Mumford 紧化中,边界对应曲面的退化(收缩 cycle、形成节点等)。这些退化都是 Class A 或 Class B 类型(整数倍或有理倍 $2\pi$ 缺口),与已知的 Hodge 理论一致。Class C 奇点若存在,将是无法被任何 Deligne-Mumford 型紧化捕获的退化模式——它会迫使几何学家寻找全新形式的紧化对象。"Class C 是否存在"等价于问"Deligne-Mumford 紧化是否完备?"

第二,bubble tree 收敛。在 harmonic map / minimal surface / Yang-Mills bubbling 中,bubble 释放整数倍的"瞬子能量"或"调和能量",这是典型的 Class A 行为。Class B 出现在 orbifold quotient 等情形。Class C 若存在,意味着存在某种 bubble 释放超越数倍的能量——这将彻底改变 bubble tree 紧致性定理的形式,迫使引入新的不变量族。

第三,stratified geometry。Whitney stratification、Goresky-MacPherson intersection cohomology 等理论都基于"奇点的 link 是有理上同调对象"这一隐式假设(Class A/B 类型)。Class C 奇点若存在,会要求把 stratified 理论的底层数域从有理数扩张到周期数域——这是一个跨代数几何与拓扑的方向重定向。

这些接口表明,Class C 存在性不是"哲学家关心、几何学家无感"的问题,而是直接影响几何学家三个活跃研究方向的具体诊断问题。本文给出问题,具体的存在性构造、否决证明、或与已有理论的精确对接,留给几何学家。

6.4 Readout-first principle:读数优先原则

ρ-defect ledger 与 ρ-signature 处理的是"如何记账"的问题。但在实际研究中,几何学家还面临一个更前置的问题:控制什么? 局部数据(曲率、能量、defect measures)有许多分量,每一个不变量只读出其中一部分。哪些分量应当被优先控制?

传统几何分析的默认是先控制总能量、总范数,再处理其细节。但实际上,Gauss-Bonnet、signature 定理、index 定理告诉我们:全局不变量看的不是局部场的总能量,而是特定的 readout functional。

原则 6.5(读数优先 / Readout-first principle). 在研究全局几何不变量 $I$ 时,优先控制局部余项在 $I$ 的读数方向上的投影,而非优先控制局部余项的总范数。

这条原则的形式化设定如下。给定全局不变量 $I = \mathcal{R}[F]$,其中 $\mathcal{R}$ 是从局部数据 $F$ 到全局不变量的 readout functional。把局部场分解为

$$F = F_{\mathrm{readout-visible}} + F_{\mathrm{readout-invisible}},$$

其中 $F_{\mathrm{readout-visible}}$ 是 $\mathcal{R}$ 看见的部分,$F_{\mathrm{readout-invisible}}$ 是 $\mathcal{R}$ 看不见的部分(对 $I$ 不贡献)。原则断言:主导能量模不一定主导全局不变量;主导全局不变量的是 readout-visible 模。

下面用三个例子说明这一原则的具体含义。

例 6.4.1(Gauss-Bonnet). 全局不变量 $\chi$ 读的是 Pfaffian 形式 $\text{Pf}(\Omega)$,不是曲率张量的全部。曲率张量中的对称部分(Ricci scalar 等)对 $\chi$ 不贡献——它们是 readout-invisible。这意味着控制 $\int \lvert\text{Rm}\rvert^2$ 总能量并不直接控制 $\chi$;真正关键的是控制 Pfaffian 的反对称分量。

例 6.4.2(Signature theorem,Hirzebruch). 全局不变量 $\sigma(M)$ 读的是 $L$-class 多项式 $L(p_1, p_2, \ldots)$ 的 Chern-Weil 代表,不是 Pontryagin 类的全部。这给具体的曲率分量分配了"对 signature 贡献"的权重。

例 6.4.3(Atiyah-Singer index theorem). 全局不变量 $\mathrm{ind}(D)$ 读的是 characteristic form 与 Chern character 的 pairing,不是椭圆算子的全部解析能量。这就是为什么 index 是拓扑不变量——大部分解析变化是 readout-invisible。

读数优先原则对几何分析实践的具体指引体现在三个方向。在 Ricci flow blow-up analysis 中,询问"我控制的曲率范数是 readout-visible 还是 readout-invisible?"可以帮助识别 blow-up 序列是否捕获了拓扑变化的真正机制。在 weak convergence 与 defect measures 分析中,识别哪个 mode 是 readout-visible 可以避免在 readout-invisible 部分上消耗精力。在 normalization 选择中,询问"我的归一化是否屏蔽了 readout-visible 模"可以防止 fake closure(假闭合)——通过定义把 defect 归零的修辞性"解决"。

readout-first principle 不是新工具,而是把已知的"全局不变量 vs 局部能量"这一现象提炼为一条可重复使用的研究方法论指引。它的 SAE 不可替代性在于:纯几何视角通常按"对象-属性"思考(对象有曲率、有能量、有 index),SAE 视角按"对象-读数-余项"思考(对象通过 readout functional 显形为余项,readout-invisible 部分不被看见也不被记账)。

6.5 应用一:Ricci flow surgery 的 W-泛函量子化猜测

本节展示 ρ-defect ledger 框架在几何分析方向的具体应用。

Perelman 一九九五至二零零三年引入 W-泛函(熵泛函)[13]:

$$\mathcal{W}(g, f, \tau) = \int_M \left\tau(R + \lvert\nabla f\rvert^2) + f - n\right^{-n/2} e^{-f}\, dV$$

满足在 Ricci flow 演化下单调不减。Perelman 进一步定义了 $\lambda(g) = \inf_f \mathcal{W}(g, f, \tau=1)$ 与 $\nu(g) = \inf_{\tau,f} \mathcal{W}(g, f, \tau)$ 等不变量,这些不变量在 Ricci flow 演化下也具有单调性。

在带 surgery 的 Ricci flow 中,$\mathcal{W}$ 在 surgery 时刻会发生突变(jump):

$$\Delta\mathcal{W}_s := \lim_{t \to t_s^-}\mathcal{W}(g(t)) - \lim_{t \to t_s^+}\mathcal{W}(g(t))$$

其中 $t_s$ 是 surgery 发生的时刻。Bamler 二零一八年的工作 [16] 给出了 $\mathcal{W}$ 在弱流框架下的单调性扩展,并证明 $\Delta\mathcal{W}_s$ 是有界的。

猜想 6.6(W-泛函突变量子化猜想). 在 Ricci flow with surgery 的每一个奇点 surgery 时刻 $t_s$,Perelman W-泛函的突变 $\Delta\mathcal{W}_s$ 是量子化的——存在一个由 surgery 过程的拓扑跃迁数据(Euler 示性数变化、连通分量数变化、surgery 类型等)决定的非平凡函数 $Q(\Delta\chi_s, \Delta b_0, \tau_s, \ldots)$,使得

$$\Delta\mathcal{W}_s = Q(\text{topological transition data at } s).$$

进一步,在合适的归一化下,$Q$ 是离散值函数——其取值集合是由拓扑跃迁数据决定的可数离散集。

本猜想有意不固定 $Q$ 的具体函数形式(例如不指定 $Q = -2\pi \cdot \Delta\chi$ 或 $Q = c \cdot \Delta\chi^2$ 等具体公式)。理由是:Bamler 等人的现有 entropy bounds 是不等式形式(单调性 + 有界性),具体的"相等"型 quantization 公式仍是开放问题,具体系数依赖于 surgery 的精确归一化与维度。本猜想给出的是结构性预测——突变是被拓扑跃迁数据控制的离散值——而具体函数形式留给后续工作确定。

SAE 视角的不可替代性在此处体现得很清楚。纯几何视角看 Perelman entropy 的现有结果是单调性 + 有界性,没有 strict equality 形式的 quantization 内禀理由。SAE ρ-defect ledger 视角则建议:$\mathcal{W}$ 是一个 ledger 不变量,Surgery 是一个升层位置,因此 $\Delta\mathcal{W}_s$ 应当是该位置的 ρ-signature 在 $\mathcal{W}$ 上的精确分量。Ledger 的 ledger 隐喻自然倾向于 closure 的 strict equality 而非 inequality——这一倾向不是一条已被推导的定理结论,而是 SAE 框架对几何分析的结构性期待。几何分析界是否会发现 W-泛函的突变实际是 strict equality(支持 SAE 期待)、还是带不可消减误差项的 substantive inequality(挑战 SAE 期待),正是本猜想留给未来工作可证伪的内容。

这一猜想可以被未来工作证伪或精化。若有具体例子表明 $\Delta\mathcal{W}_s$ 在合理归一化下是连续值而非量子化,本猜想被证伪。若有具体函数形式 $Q$ 被证明,本猜想被精化为定理。若 $Q$ 被证明只对某类 surgery 量子化、对另一类不量子化,本猜想需要被分类细化。

无论结果如何,这一猜想给几何分析方向的研究者一个具体可证伪的、SAE 视角下自然提出的研究问题——这正是本节的目标:不是给几何学家一个新定理,而是给他们一个新问题。

6.6 应用二:几何流与信息几何相变的耗散对偶

本节展示 ρ-defect ledger 框架作为跨子领域桥梁的具体应用,同时为「形与流」系列后续 P2 至 P5 提供共享几何工具。

Perelman W-泛函的形式

$$\mathcal{W}(g, f, \tau) = \int_M \left\tau(R + \lvert\nabla f\rvert^2) + f - n\right^{-n/2} e^{-f}\, dV$$

在表观上是几何分析对象,但其结构与统计物理中的 Boltzmann-Gibbs 熵泛函(在某种 Lagrange multiplier 形式下)惊人相似。事实上,Perelman 自己在 [13] 中已经指出 $\mathcal{W}$ 可以被解读为带 Lagrange multiplier 的统计力学泛函;Lott-Villani [34]、Sturm [35]、Otto-Villani [36] 等的工作进一步在 Wasserstein 几何与最优传输的框架下,把 Ricci 曲率与统计流形上的熵几何严格联系起来——具体说,Ricci 曲率下界 $\mathrm{Ric} \geq K$ 等价于 Wasserstein 空间上熵泛函的 $K$-凸性,这就是 RCD($K$, $N$)空间理论的基础。

这一现象在 SAE 视角下不应被读为巧合,而应被读为ρ 不可消定律在几何层与信息层必须具有同构耗散方程的必然结果。

猜想 6.7(几何流与信息几何相变的耗散对偶). Ricci flow 中的度规坍缩(volume collapse)、奇点形成(singularity formation)、与升层手术(surgery)过程,在数学结构上对偶于统计物理系统在热力学极限下相变(phase transition)时,其底层概率分布族的 Fisher 度规退化(degeneration)、奇异化(singularization)、与重组(reconfiguration)过程。

更精确地,本猜想包含一个层级区分。已建立的 RCD 层级连接:Lott-Villani-Sturm 工作已经在静态层面建立了 Ricci 曲率下界与 Fisher/熵几何的等价(RCD($K$, $N$)空间理论),这是几何与信息几何之间在静态曲率层面的成熟桥梁。本猜想所提出的扩展:这一桥梁可以延伸到动力学层面——即 Ricci flow 一侧的 surgery 拓扑跃迁应当对应统计流形一侧的相变拓扑重组,Ricci flow 一侧的 $\Delta\mathcal{W}_s$ 应当对应统计流形一侧的相变熵跃变。

需要明确:这一动力学层级延伸本身就是开放猜想,即使在 SAE 视角之外也尚未被现有几何/概率论文献完全建立——本猜想并非"现有文献已知、SAE 重新命名"的现象,而是"本系列首次显式提出的方向"。SAE 视角的贡献是给这一方向一个统一语汇(ρ-defect ledger 同时适用于两侧),而方向本身的具体形式化(明确给出范畴变换的范畴论结构、ledger 之间的对应关系、Wasserstein 几何与 Ricci flow 在 surgery / 相变层面的精确对偶)需要未来工作完成。在本文范围内,"对偶"一词作为非正式语汇使用,等待未来形式化为合适的范畴论结构(可能是函子、可能是自然变换、可能是 derived 层面的对应,具体形式有待确定)。

SAE 视角下的本体论解读是:几何 ρ(曲率作为局部不可消阻力)与信息 ρ(分布可区分性作为不可消代价)是同一种本体论阻力在两个不同 D 层面的显形。它们在动力学层面必须满足相同的耗散方程,因为驱动两种动力学的是同一个 ρ。耗散对偶不是数学巧合,是 SAE ρ 不可消命题在两个层面的镜像必然。

这一猜想的跨学科辩护力体现在它同时连接三个原本分散的研究社群。微分几何与几何分析(Hamilton-Perelman 系谱)关注 Ricci flow with surgery、Perelman entropy、几何 H' 升层。信息几何(Amari-Chentsov 系谱)关注 Fisher 度规、α-连接、统计流形曲率。最优传输(Otto-Villani-Sturm-Lott 系谱)关注 Wasserstein 几何、entropic transport、RCD 空间。这三个领域过去一二十年内已经各自发展出深刻的几何工具,但它们之间的具体桥接仍主要在"曲率下界 ↔ 熵凸性"这一 RCD 现象上。本猜想预言:这一桥接可以延伸到奇点 / 相变 / surgery / 升层这一更动态的层面,而 SAE ρ-defect ledger 框架是统一这一延伸的语汇。

这一对偶猜想同时直接为「形与流」系列后续提供共享几何工具。P2 流体力学中的边界层、湍流、奇点爆发,可以读为 Fisher 度规相变的几何对偶。P3 生命演化动力学中的适应度景观坍缩、物种灭绝事件,可以读为 Wasserstein 度规上的相变与升层。P4 种群动力学中的 replicator 流、相空间分歧,可以直接调用本对偶猜想给出的统一框架。P5 主体动力学中的舆论场流、信念分布演化,可以在统计流形 + 几何流耦合的框架下处理。

本猜想是「形与流」系列内部最重要的几何工具桥梁。它在本文中只立姿态与陈述,具体形式化与各 D 层级的应用留待 P2 至 P5 各自展开。

6.7 应用三:次黎曼几何反常测地线作为 ρ 拓扑打穿

本节展示 ρ-defect ledger 视角对一个经典几何"病态"现象的重读,演示 SAE 框架"化敌为友"的姿态。

次黎曼几何(sub-Riemannian geometry)是黎曼几何的一个重要扩展:在流形 $M$ 上指定一个秩 $k < \dim M$ 的子分布 $\mathcal{D} \subset TM$,要求"运动只能在 $\mathcal{D}$ 中进行"。这类几何在控制论(机器人运动、车轮滚动、量子控制系统)、热扩散(degenerate Laplacian)、Carnot 群分析等多个方向都有具体应用。

次黎曼几何的核心难题之一是反常测地线(abnormal geodesics)的存在与性质。反常测地线是指那些满足局部最短性条件、但不来自 sub-Riemannian Hamiltonian 的标准变分原理的曲线。它们的存在是 Sussmann 一九九六年系统提出的世纪难题 [37],至今仍是次黎曼几何与几何控制论中最深的开放问题之一。Liu-Sussmann 一九九五年构造了具体反常测地线例子,证明它们确实存在且可以是局部严格最短的;但反常测地线是否总是平滑、是否构成几何对象的合法组成部分、它们在几何分类中应当如何处理,至今没有统一答案。

传统次黎曼几何文献中,反常测地线得到的处理是分歧的:部分文献(尤其分类理论方向)把它们视为"例外情况"、需要"在反常测地线之外讨论"才能让标准定理成立;另一部分文献(Liu-Sussmann、Hsu、Agrachev-Sachkov、Montgomery 的工作系谱)把它们作为合法对象积极研究。但即便在积极研究的传统中,它们的具体性质——是否平滑、如何分类、与几何对象其他特征的关系——仍是悬而未决的开放问题。SAE 视角不是声称几何学家"都没看见"反常测地线,而是为既有的研究提供一个重读视角。

SAE 重读 6.8(反常测地线作为 ρ 拓扑打穿). 反常测地线不是病态,而是ρ 在极度压缩约束空间(秩 $k < n$ 的子分布)中,寻找最小阻力路径时产生的拓扑打穿。

具体说明:在标准黎曼几何中,所有方向都可达,ρ 通过曲率显形为局部弯曲的可积累读数。在次黎曼几何中,运动方向被严格限制,ρ 不能通过常规曲率显形,因为大部分方向"不存在"。这种约束并不消灭 ρ——按元命题 5.3 "统一即迁移",ρ 必须迁移到新的位置。反常测地线就是 ρ 迁移到的新位置:它们是几何 ρ 在极度约束下"穿透"标准 Hamiltonian 框架、寻找替代显形路径的产物。

在 ρ-defect ledger 框架下,这一现象可以被形式化。设 $X_t$ 是无约束(全黎曼)对象,$X_\infty$ 是次黎曼对象($\mathcal{D} \subsetneq TM$)。这一过程是一种"约束升层"——从 rank $= n$ 到 rank $= k < n$ 的范畴扩展。选 $\mathcal{I} = \{\text{geodesic structure}, \text{cut locus}, \text{minimizer set}\}$。Ledger:

$$\text{geodesics}(X_t) = \text{Hamiltonian geodesics}(X_\infty) + \text{abnormal geodesics}(X_\infty) + \varepsilon$$

反常测地线就是该升层的 ρ-signature 在 geodesic structure 上的非零分量。

这一重读的姿态意义需要被精确说明。它不解决 Sussmann 难题——反常测地线的具体性质(是否平滑、如何分类等)仍是开放问题。它也不声称次黎曼几何整个共同体都把反常测地线当作病态而 SAE 才看见其合法性——已有相当一批工作在严肃研究这些对象。它所做的是为既有的、分歧的处理提供一个统一的本体论坐标:在传统视角下"反常测地线作为病态例外"和"反常测地线作为合法对象"这两种处理可以被同时容纳,前者对应"消除尝试"的姿态,后者对应"承认 ρ 显形"的姿态;SAE 视角主张后者更接近反常测地线的本体论身份,但不否认前者的工程价值。

这是 SAE 视角"化敌为友"姿态的一个具体演示——为传统视角下被部分文献当作"病态例外"的对象,提供一种把它"重新读为 ρ 的合法显形"的统一框架。本文不主张这一重读自动解决 Sussmann 难题,但主张它给次黎曼几何研究提供一个研究方向的可能性扩展:在"避开反常测地线"和"分析反常测地线本身"两条路径之间,SAE 视角提议增加第三条——分析反常测地线的 ρ-signature 结构,把它们作为 SAE 升层范畴中的指纹对象处理。

6.8 最小 ρ 迁移原则:范畴选择 checklist

本节给出范畴选择的方法论补充。元命题 5.3 "统一即迁移"在范畴选择实践中诱导一条具体原则。

当原范畴 $\mathcal{C}$ 无法容纳对象的极限、奇点、解或分类时,几何研究者需要选择升层范畴 $\mathcal{C}^\uparrow$。但升层有多种选择。Ricci flow 奇点处理可以选 surgery、singular Ricci flow、weak flow、through-singularities flow 等。代数几何可以选 schemes、stacks、derived schemes、log schemes、$\infty$-stacks 等。度量几何可以选 GH limits、metric-measure spaces、RCD spaces、smooth metric measure spaces 等。微分几何可以选 orbifolds、stratified spaces、currents、varifolds 等。Moduli theory 可以选 classical moduli、virtual fundamental class、derived enhancements 等。

每一类升层都"解决"了原范畴的某个问题,但也"引入"了新的内部边界——新的 strata、新的 coherence data、新的范畴层级、新的奇点类型。SAE 元命题 5.3 主张这是必然的:任何升层都不消灭 ρ,只迁移 ρ。

原则 6.9(最小 ρ 迁移原则). 在升层范畴的选择中,合法的选择是最小 ρ 迁移路径:在保持原问题可解的前提下,使新引入的 ρ(新内部边界 + 新 coherence data)最少的那一条升层路径。

形式化为可操作 checklist。一个升层 $\mathcal{C} \to \mathcal{C}^\uparrow$ 相对于不变量族 $\mathcal{I}$ 是 ρ-minimal 的,若它满足:

1. 完全嵌入:$\mathcal{C}$ 完全忠实地嵌入 $\mathcal{C}^\uparrow$,即原范畴的对象与态射在新范畴中保持不变;
2. 可解性:$\mathcal{C}^\uparrow$ 能容纳原问题所需的极限、奇点、解或分类;
3. Ledger 存在:$\mathcal{C}^\uparrow$ 提供相对于 $\mathcal{I}$ 的 defect ledger;
4. Coherence 显式:升层引入的所有新 coherence data 是显式的、可识别的、可计数的;
5. 不可缩:不存在严格更小的升层 $\mathcal{C}^\uparrow_0 \subsetneq \mathcal{C}^\uparrow$ 同时满足 (1)–(4)。

满足这五条的升层是 ρ-minimal。一个升层若仅满足 (1)–(2) 但不满足 (3)–(5),则它"解决"了问题但未完成余项记账,SAE 视角下是过度升层(over-enlargement)——引入了不必要的余项迁移。

这条原则给具体范畴选择问题提供方法论判断。在 derived geometry 中,在 derived schemes、derived stacks、$\infty$-stacks 之间选择时,询问"我所关心的不变量族是否需要更高阶的 derived 结构?所引入的更高 coherence data 是否是必要的?"在 moduli theory 中,在 classical moduli 与 virtual fundamental class 之间选择时,询问"virtual class 引入的 obstruction theory 是否被我所关心的不变量族真正利用?"在 singular geometry 中,在 orbifolds、stratified spaces、currents、varifolds 之间选择时,询问"哪一类对象给我的不变量族最完整且最简洁的 ledger?"

最小 ρ 迁移原则不是对现有数学实践的革命——许多数学家在范畴选择时已经直觉性地应用类似判断("用最小够用的范畴")。本原则做的是把这一直觉判断显式化为一组可操作 checklist,并给出"为什么是这五条"的 SAE 本体论根据。它不取代研究者的具体判断,但提供一个统一的诊断框架。

更深的形式化——把"ρ 迁移代价"严格度量化、构造范畴层级上的"迁移函子"、证明特定升层的最小性——是一项独立工作,留待 P0 与未来 SAE 范畴论专项。本节给出的是原则与 checklist,不是定理。

6.9 小结:本节对几何学家的承诺

§6 给几何学家的具体承诺总结如下。

诊断工具(可重复使用): ρ-defect ledger(定义 6.1)是审查任何升层理论是否完成余项记账的统一框架;ρ-signature(定义 6.3)把奇点从对象分类升级为指纹分类;readout-first principle(原则 6.5)是优先控制 readout-visible 模而非 total energy 的方法论指引。

具体可证伪猜想: W-泛函量子化猜想(猜想 6.6)预言 Ricci flow surgery 中 Perelman 熵突变是被拓扑跃迁数据控制的离散值;几何流与信息几何耗散对偶猜想(猜想 6.7)预言 Ricci flow 与 Fisher 度规相变在动力学层面具有结构对偶,延伸现有 RCD 层级的静态连接,作为 P2 至 P5 共享几何工具的基础。

研究方向重读: 次黎曼几何反常测地线的 ρ 拓扑打穿读法(SAE 重读 6.8)是"化敌为友"姿态的具体演示。

方法论补充: 最小 ρ 迁移原则与 5 条 checklist(原则 6.9)是范畴选择的 SAE 视角判据。

这些工具与猜想不要求几何学家接受全部 SAE 框架。它们各自可以被独立检验、独立使用、独立证伪或独立发展。SAE 视角的贡献是指认这些工具与猜想之间的统一性——它们都来自"ρ 不可消、统一即迁移"这一本体论命题——但每一个工具都可以脱离这一统一性单独存活。

本节的目标因此达成:让几何学家在读完本文后,愿意尝试用 SAE 视角推演自己的具体问题——因为他们看到了具体可用的诊断框架、可证伪的猜想、可重读的现象、可操作的 checklist。

---

7. 余下的工作:复几何占位、与已有 SAE 系列的关系、留白

7.1 复几何这一格的占位

本文所处理的几何是实 Riemannian 几何及其在标准实数背景上的若干扩展。复几何——包括 Hermitian 几何、Kähler 几何、Calabi-Yau 流形 [28]、复代数几何等——是几何余项的另一面,本文没有处理。复结构本身是非平凡的余项,实 Riemannian 几何的语汇并不能直接捕获它。本文不对这一格做任何处理,把它显式标注为留待后续的工作:「论几何学所基于的假设」之三,可以是 SAE 复几何的开端,由后人完成。这是一个独立的工作方向,与本文平行而非附属。

类似地,谱几何与指标定理(Atiyah-Singer 等 [9])所揭示的"解析不变量与拓扑不变量之间的精确兑换"是命题 2.1 兑换律框架的更深一层版本,本文亦不展开,作为另一个独立的工作方向留下。

7.2 与已有 SAE 系列的关系

「形与流」系列与已有的 SAE 物理系列 [29]、SAE 信息论系列 [30]、SAE 生物学注、SAE 道德律系列 [31]、SAE 法律系列、SAE 认识论系列、SAE 经济学系列、人类总构系列共享同一框架 [2,3]。这些系列在不同侧面上平行展开 SAE 的本体论命题——SAE 物理在自然律侧面、SAE 信息论在物理-信息接口、SAE 生物学注在生命科学侧面、SAE 道德律在伦理结构侧面、人类总构在思想史侧面——「形与流」处理的是几何与动力学侧面在各 D 层级上的展开。各系列之间不构成层级关系,而是同一本体论框架的多侧面平行展开。本文为「形与流」系列开篇,后续 P2 至 P5 与 P0 将依次展开;本文不试图覆盖其他系列已经处理的主题。

§6.6 提出的几何流与信息几何耗散对偶猜想是「形与流」系列内部最重要的桥梁:它同时打底 P2(流体)、P3(演化)、P4(种群)、P5(主体动力学)的几何工具。后续各 D 层级的展开将反复回指这一对偶猜想及 §6 给出的诊断框架(ρ-defect ledger、ρ-signature、readout-first principle)。

7.3 显式留白清单

本文明确不承担、留待未来 SAE 几何专项或后人工作的事项,集中列出如下:

(一)几何 H' 与 ZFCρ H' 之间结构性类比的范畴论形式化(明确给出范畴、函子、自然变换);

(二)不动点定理族(Brouwer、Banach、Lefschetz、Tarski 等)与"余项不可消"在抽象层面的统一表述;

(三)谱几何余项守恒族:$\zeta$ 函数特殊值、Selberg 迹公式、heat kernel 系数、analytic torsion、indices of elliptic operators 等的 SAE 重读;周期理论(Kontsevich-Zagier)的 SAE 视角处理;

(四)复几何/Kähler/Calabi-Yau 这一格的 SAE 推演——「论几何学所基于的假设」之三;

(五)高维特征类(Pontryagin、Chern、Euler 等)的完整 SAE 重读;

(六)实数背景假设与点集基底假设的实质替换工作;

(七)数学构造性/可计算性维度的 SAE 处理(更深一层的留白);

(八)W-泛函量子化猜想(猜想 6.6)中具体函数形式 $Q$ 的确定;

(九)几何流与信息几何相变的耗散对偶猜想(猜想 6.7)的形式化:具体的范畴变换、ledger 之间的对应、Wasserstein 几何与 Ricci flow 之间在 surgery / 相变层面的精确对偶;

(十)Class C 奇点(§6.3)的存在性问题及具体构造;

(十一)最小 ρ 迁移原则(原则 6.9)中"ρ 迁移代价"的严格度量化、范畴层级上"迁移函子"的构造、特定升层最小性的证明;

(十二)反常测地线的 ρ-signature 结构(§6.7)的具体分析。

7.4 Claim-status map:本文承担的层级显式

为使读者一眼可见本文各处主张的层级状态,把全文的形式陈述按 epistemic 强度归入四层:

第一层:本文承担为命题的硬主张

标号	内容	性质
命题 2.1	Gauss-Bonnet 作为四重余项兑换律	对已有数学结果的本体论重读
命题 2.2	标准归一化下 Gauss-Bonnet 兑换律的可证伪形式陈述	严格证明,带 Lindemann 定理依赖
命题 3.1	Ricci flow 几何 H' 读法	对 Hamilton-Perelman 程序的本体论重读

这三条是 P1 的核心承担,作为本文实质贡献提交给几何学共同体审视。

第二层:方法论框架与元层命题

标号	内容	性质
抽象模式 3.3	H' 型不可消解模式	跨子领域抽象模式指认
定义 5.1	假设松动偏序 $\preceq$	表述清晰用工具
观察 5.2	松动链共享模式	跨子领域结构观察
元命题 5.3	统一即迁移	SAE 元层本体论命题
指向性命题 2.4	周期作为数论余项的标准显形	指向性陈述,留待形式化

这一层是 P1 的方法论与本体论框架,服务于核心命题的论证与系列后续打底。

第三层:诊断工具(§6 操作化框架)

标号	内容	性质
定义 6.1	ρ-defect ledger	跨子领域诊断框架
定义 6.3	ρ-signature	奇点指纹分类工具
原则 6.5	Readout-first principle	研究方法论指引
原则 6.9	最小 ρ 迁移原则与 5 条 checklist	范畴选择方法论

这一层把第一、二层的本体论命题操作化为几何学家可重复使用的诊断工具。本身不是新数学发现,而是对已有几何分析工具的统一框架重组。

第四层:Conjectural applications 与开放研究问题

标号	内容	性质
猜想 3.2	原始范畴不可闭合预测	可证伪猜想
问题 6.4	Class C 奇点存在性	开放问题
猜想 6.6	W-泛函突变量子化猜想	可证伪猜想(结构性形式)
猜想 6.7	几何流与信息几何相变耗散对偶	可证伪猜想(结构对偶形式)
SAE 重读 6.8	反常测地线作为 ρ 拓扑打穿	研究方向重读,不解决具体难题

这一层与第一、二层不等权。它是 SAE 视角对几何学家的具体研究邀约——给出可证伪猜想或新研究方向,具体形式化、计算、检验留待几何学共同体。这一层的 epistemic 状态是"可证伪 conjecture 与 research prompt",不是已论证的结论。

显式留白(§7.3) 是第五个层次:本文明确不承担、留待未来工作的方向。已列出 12 项。

把这四层加上留白共五层放在一起,是本文承担状态的完整层级图。

7.5 结语

本文核心承担的两条命题——Gauss-Bonnet 四重余项兑换律(命题 2.1, 2.2)与 Ricci flow 几何 H' 升层读法(命题 3.1)——是 P1 给出的本体论重读。在此基础上,本文提供了五条假设处理与方法论工具——几何字典、信息几何桥梁、假设松动偏序、元层命题"统一即迁移"、以及 §6 的诊断框架(ρ-defect ledger、ρ-signature、readout-first principle、最小 ρ 迁移 checklist)。最后,本文向几何学共同体提出三条具体研究邀约——作为 conjectural applications 与开放研究问题——W-泛函量子化猜想、几何流与信息几何耗散对偶猜想、反常测地线 ρ 拓扑打穿读法。

这三组贡献在 epistemic 强度上不等权:核心命题是本文承担的硬主张,方法论与诊断框架是论证的支撑工具,conjectural applications 是对几何学家的具体研究邀约。§7.4 的 claim-status map 把这一层级显式列出。

接续黎曼一八五四年的方法论,本文审查了若干假设、替换了能替换的几条、显式承认了不能替换的几条,并把本体论命题操作化为一组面向几何学家的可重复使用工具。本文的形式本身——指认假设、替换部分、显式留白——构成对"余项不可消"这一本体论命题的具身证据。

P1 的贡献不是建立 SAE 几何学体系,而是把黎曼方法论在 SAE 视角下重新激活:指认假设、替换部分、显式承认未替换部分。剩下的工作交给后人。

---

附录 A:低维具体 ledger 演算

本附录给出 §6 中三条诊断工具与三条应用在低维具体例子上的展开演算。这些例子的目的是让几何学家能直接看到 ρ-defect ledger / ρ-signature / readout-first principle 在熟悉对象上长什么样,从而在自己的研究中尝试应用这些工具。

附录的内容是示范性的,不是新的数学结果——A.1 中的 Gauss-Bonnet 计算是 19 世纪以来标准教科书内容,A.2 中的 neckpinch 框架基于 Hamilton-Perelman 已有理论,A.3 中的 Heisenberg 群反常测地线基于 Liu-Sussmann 1995 构造。本附录的工作是用 ρ-defect ledger 语汇重新表达这些已知例子,展示 SAE 框架在具体计算中如何具体显形。

A.1 Gauss-Bonnet ρ-defect ledger 的低维演算

A.1.1 单位球面 $S^2$

设 $M = S^2$,标准度规,$K \equiv 1$。无边界,$\partial M = \varnothing$。

不变量族选择 $\mathcal{I} = \{\chi\}$,Gauss-Bonnet ledger 形式为

$$\frac{1}{2\pi}\int_M K\, dA = \chi(M) - \frac{1}{2\pi}\int_{\partial M} k_g\, ds.$$

直接代入:$\int_{S^2} K\, dA = 4\pi$,$\partial M = \varnothing$ 给出 $\int_{\partial M} k_g\, ds = 0$。

$$\frac{4\pi}{2\pi} = 2 - 0,$$

即 $2 = \chi(S^2) = 2$。Ledger 完全闭合,$\varepsilon_\chi = 0$。

ρ-signature:$S^2$ 无升层位置,所有曲率作为 readout-visible 模显形为内部几何余项,Archimedean 兑换因子 $2\pi$ 完整地把连续读数兑换为离散整数 $\chi = 2$。这是 ledger 在最简单情形下的样子。

A.1.2 标准圆盘 $D^2$(带边界例子)

设 $M = D^2$ 是单位欧氏圆盘,$K \equiv 0$(平直度规),$\partial M = S^1$,边界测地曲率 $k_g \equiv 1$(单位圆周)。$\chi(D^2) = 1$。

Ledger:

$$\frac{1}{2\pi}\int_{D^2} K\, dA + \frac{1}{2\pi}\int_{S^1} k_g\, ds = \chi(D^2)$$

$$0 + \frac{1}{2\pi} \cdot 2\pi = 1.$$

完美闭合,$\varepsilon = 0$。

ρ-signature 解读:$D^2$ 有一个升层位置 $s = \partial M$(边界),它的 ρ-signature 是 $\rho_{\chi, s} = 1$——边界释放了 $\chi$ 的全部值。内部几何余项为零(平直),所有 $\chi$ 信息从边界余项流出。这与 §6.4 的 readout-first principle 一致:$\chi$ readout 在 $D^2$ 平直情形下完全由边界 $k_g$ 项决定,而非由内部曲率决定。

A.1.3 环面 $T^2$ 与连通和 $T^2 \# T^2$

环面 $T^2$ 选平坦度规($K \equiv 0$),无边界。Ledger 给出 $0 = \chi(T^2)$,即 $\chi(T^2) = 0$。完美闭合。$T^2$ 的 ρ-signature 是平凡的——既无内部余项也无边界余项,亦无升层位置。

亏格 2 曲面 $\Sigma_2 = T^2 \# T^2$ 选双曲度规($K \equiv -1$,面积 $4\pi$,因 Gauss-Bonnet)。Ledger:

$$\frac{1}{2\pi}\int_{\Sigma_2} K\, dA = \frac{-4\pi}{2\pi} = -2 = \chi(\Sigma_2).$$

闭合。注意此处 $\chi = -2 = 2 - 2g$ 验证 Gauss-Bonnet 在高亏格上仍然作为 ledger 工作。

A.1.4 锥奇点情形(Class B 示例)

考虑带单一锥奇点的曲面:$M = $ 平直 $\mathbb{R}^2$ 在原点开了一个总角 $\theta_0 \in (0, 2\pi)$ 的圆锥点。除奇点外光滑平直,$K \equiv 0$ 在 $M \setminus \{0\}$ 上。

按 §6.3 定义的 Gauss-Bonnet 余项缺口

$$\Delta_\rho(0) = 2\pi - \theta_0 \in (0, 2\pi).$$

若 $\theta_0 = \pi$(直角锥),$\Delta_\rho = \pi \in 2\pi \cdot (1/2) \subset 2\pi\mathbb{Q}$——Class B,signature 落在有理兑换扩展上。

若 $\theta_0 = 2\pi/n$($n \in \mathbb{Z}_+$,orbifold 锥),$\Delta_\rho = 2\pi(1 - 1/n) \in 2\pi\mathbb{Q}$——同样 Class B。

所有标准锥奇点(orbifold 类型、有理倾斜锥)的 ρ-signature 都落在 Class B。Class C 奇点要求 $\theta_0 / 2\pi$ 为超越数,标准几何构造尚不能自然产生这种奇点——这是 §6.3 问题 6.4 的具体落地形式。

A.2 Ricci flow surgery 的 ρ-signature 框架例

A.2.1 标准 neckpinch 简介

考虑三维流形上的 Ricci flow,经历标准 neckpinch surgery。Hamilton-Perelman 程序中的典型场景:流形 $M_t$ 形成一个收缩到零体积的圆柱形颈区,在奇点时刻 $t_s$ 切除奇异邻域、在两侧补上半球冠,得到拓扑变化后的流形 $M_{t_s^+}$。

设 $M_{t_s^-}$ 在 surgery 前是连通的紧致 3-流形,$M_{t_s^+}$ 在 surgery 后分裂为两个连通分量(典型的 neckpinch 把一个流形分成两个"哑铃头")。则在不变量族 $\mathcal{I} = \{\chi, b_0\}$($b_0$ 是连通分量数)下:

$$\Delta\chi_s = \chi(M_{t_s^+}) - \chi(M_{t_s^-})$$

$$\Delta b_0 = b_0(M_{t_s^+}) - b_0(M_{t_s^-}) = 1$$

这两个具体值依赖于 surgery 的具体类型(例如,3-球面型 surgery 与 lens space 型 surgery 给出不同的 Δχ),由 surgery 的局部模型(neckpinch、degenerate neckpinch 等)精确决定。

A.2.2 ρ-signature 在 W-泛函上的层级形式

按 §6.5 猜想 6.6,Perelman W-泛函在 surgery 时刻的突变 $\Delta\mathcal{W}_s$ 应当是被 $(\Delta\chi_s, \Delta b_0, \tau_s, \ldots)$ 控制的离散值。本附录不给出 $\Delta\mathcal{W}_s$ 的具体数值——这需要在现有 Bamler entropy framework 内做完整计算,涉及具体维度归一化与 surgery 类型分类,超出本文承担。本附录给出的是结构性 ledger 形式:

$$\mathcal{W}(g_{t_s^-}) = \mathcal{W}(g_{t_s^+, \mathrm{reg}}) + \Delta\mathcal{W}_s + \varepsilon_\mathcal{W}.$$

在 SAE 框架下,猜想 6.6 主张 $\varepsilon_\mathcal{W} = 0$(strict equality, 量子化形式),而 Bamler 现有结果给出 $\varepsilon_\mathcal{W} \neq 0$ 但有界(不等式形式)。这两种主张的真正差异,正是本猜想留给未来工作可证伪检验的内容:在合理归一化下,$\Delta\mathcal{W}_s$ 是连续值还是被拓扑数据量子化。

A.2.3 与 RCD 静态层级的对照

按 §6.6 猜想 6.7,几何流的 ρ-signature 应当对偶于统计流形相变的 ρ-signature。在 RCD 静态层级,Lott-Villani-Sturm 已建立 $\mathrm{Ric}(g) \geq K \iff K\text{-凸性}$ 的精确等价。在动力学层级,本猜想预测:

$$\Delta\mathcal{W}_s^{\mathrm{Ricci}} \longleftrightarrow \Delta S_s^{\mathrm{phase}}$$

即 Ricci flow 一侧 W-泛函突变与统计相变一侧熵跃变之间存在对偶。这一对偶的具体形式(函子、自然变换、derived 对应)有待形式化,但其结构性预测已经可以被 P2 流体力学借用作为湍流爆发与统计相变的统一框架。

A.3 Heisenberg 群上反常测地线的 ρ-signature 框架

A.3.1 Heisenberg 群作为最简单的非平凡次黎曼几何

三维 Heisenberg 群 $\mathbb{H}^3$ 是次黎曼几何最经典的非平凡例子。它是李群 $\mathbb{R}^3$ 上配以 rank-2 左不变分布:在标准坐标 $(x, y, z)$ 下,分布 $\mathcal{D}$ 由两个向量场张成,

$$X = \partial_x - \frac{y}{2}\partial_z, \quad Y = \partial_y + \frac{x}{2}\partial_z.$$

注意 $[X, Y] = \partial_z$,即 $z$-方向"不在分布中"但通过 Lie bracket 可达——这是 Heisenberg 群被研究最多的特征(Hörmander 条件 / bracket-generating)。次黎曼度规由 $X, Y$ 正交规范决定,只有沿 $\mathcal{D}$ 的曲线有定义长度,沿 $\partial_z$ 方向的瞬时位移没有定义长度(因为不在分布中)。

Heisenberg 群上的最短测地线问题已被 Brockett、Hsu、Beals-Gaveau-Greiner 等系统研究,标准 Hamiltonian 测地线由 Hamilton 方程的横截极值原理给出。问题在于:Heisenberg 群上所有最短测地线都来自 normal Hamiltonian 极值原理,在标准 Heisenberg 群上没有反常测地线。

A.3.2 反常测地线在更高 step 次黎曼结构上的出现

反常测地线出现在 step 更高(即 bracket-generating 需要更多层 Lie bracket)的次黎曼结构上。Liu-Sussmann 1995 在 Engel 分布 / Goursat 分布(rank-2 分布在 4 维流形上,bracket-generating step 3)上构造了具体的严格最短反常测地线例子。这些例子的具体性质(平滑性、唯一性、稳定性)至今仍部分开放。

A.3.3 反常测地线的 ρ-signature 框架解读

按 §6.7 SAE 重读 6.8,反常测地线在 ρ-defect ledger 框架下被读为约束升层(从 rank-$n$ 全黎曼到 rank-$k$ 次黎曼)的 ρ-signature 在 geodesic structure 不变量上的非零分量。具体:

设 $X_t = $ 全黎曼对象($\mathcal{D} = TM$),$X_\infty = $ Engel/Goursat 次黎曼对象($\mathcal{D} \subsetneq TM$)。选 $\mathcal{I} = \{\text{normal geodesics}, \text{abnormal geodesics}, \text{cut locus}, \text{minimizer set}\}$。Ledger:

$$\text{geodesic structure}(X_t) = \text{normal geodesics}(X_\infty) + \text{abnormal geodesics}(X_\infty) + \varepsilon$$

反常测地线就是该升层的 ρ-signature 在 geodesic structure 上的非零分量,Heisenberg 群($\varepsilon = 0$ 因为没有反常测地线)与 Engel/Goursat 分布($\varepsilon \neq 0$ 但被显式承担为反常测地线项)的差异,正是 SAE 框架下"约束升层步数"决定 ρ-signature 是否具有反常分量的具体显形。

这一框架不解决 Sussmann 难题(反常测地线的具体性质仍开放),但它为既有研究提供一个统一坐标:从 SAE 视角,反常测地线不是病态例外,而是约束升层在 step 高于 2 时必然产生的合法 ρ-signature 分量。Heisenberg 群之所以"没有反常测地线",不是它的特殊性使然,而是 step 2 约束升层不足以产生反常 ρ 显形;step 3 及以上的约束升层则产生它,作为更深层约束的合法记账。

具体的反常 ρ-signature 计算(在 Engel 分布上明确写出反常测地线的不变量贡献)是一项独立工作,留给次黎曼几何专项。本附录给出框架,具体 calibration 留给后续。

---

参考文献

[1] B. Riemann, Über die Hypothesen, welche der Geometrie zu Grunde liegen, Habilitationsvortrag, Göttingen, 1854; published posthumously in Abhandlungen der Königlichen Gesellschaft der Wissenschaften zu Göttingen, 13 (1868), 133–150.

[2] H. Qin, Self-as-an-End, Paper 1: 道德律的本体论奠基, Zenodo, 2025. DOI: 10.5281/zenodo.18528813.

[3] H. Qin, Self-as-an-End, Paper 2: 凿构循环与方法论奠基, Zenodo, 2025. DOI: 10.5281/zenodo.18666645.

[4] O. Bonnet, Mémoire sur la théorie générale des surfaces, Journal de l'École Polytechnique, 19 (1848), 1–146.

[5] S. S. Chern, A simple intrinsic proof of the Gauss-Bonnet formula for closed Riemannian manifolds, Annals of Mathematics, 45 (1944), 747–752.

[6] R. Bott and L. W. Tu, Differential Forms in Algebraic Topology, Graduate Texts in Mathematics 82, Springer, 1982.

[7] F. Lindemann, Über die Zahl π, Mathematische Annalen, 20 (1882), 213–225.

[8] M. Kontsevich and D. Zagier, Periods, in Mathematics Unlimited—2001 and Beyond, Springer, 2001, 771–808.

[9] M. F. Atiyah and I. M. Singer, The index of elliptic operators on compact manifolds, Bulletin of the American Mathematical Society, 69 (1963), 422–433.

[10] R. Apéry, Irrationalité de ζ(2) et ζ(3), Astérisque, 61 (1979), 11–13.

[11] R. S. Hamilton, Three-manifolds with positive Ricci curvature, Journal of Differential Geometry, 17 (1982), 255–306.

[12] G. Perelman, The entropy formula for the Ricci flow and its geometric applications, arXiv:math/0211159, 2002.

[13] G. Perelman, Ricci flow with surgery on three-manifolds, arXiv:math/0303109, 2003.

[14] G. Perelman, Finite extinction time for the solutions to the Ricci flow on certain three-manifolds, arXiv:math/0307245, 2003.

[15] B. Kleiner and J. Lott, Singular Ricci flows I, Acta Mathematica, 219 (2017), 65–134.

[16] R. H. Bamler, Convergence of Ricci flows with bounded scalar curvature, Annals of Mathematics, 188 (2018), 753–831.

[17] H. Qin, ZFCρ Series, Paper 71: H' 不可消解定理结构性陈述, Zenodo, 2026. DOI: 10.5281/zenodo.20011553.

[18] H. Qin, ZFCρ Series, Paper 72: Spectral Bridge Front II, Zenodo, 2026. DOI: 10.5281/zenodo.20029941.

[19] K. Gödel, Über formal unentscheidbare Sätze der Principia Mathematica und verwandter Systeme I, Monatshefte für Mathematik und Physik, 38 (1931), 173–198.

[20] V. I. Arnold, Mathematical Methods of Classical Mechanics, Graduate Texts in Mathematics 60, Springer, 1989.

[21] S. Amari, Differential-Geometrical Methods in Statistics, Lecture Notes in Statistics 28, Springer, 1985.

[22] S. Amari and H. Nagaoka, Methods of Information Geometry, Translations of Mathematical Monographs 191, American Mathematical Society, 2000.

[23] D. Burago, Y. Burago, and S. Ivanov, A Course in Metric Geometry, Graduate Studies in Mathematics 33, American Mathematical Society, 2001.

[24] A. Connes, Noncommutative Geometry, Academic Press, 1994.

[25] V. Voevodsky, A¹-homotopy theory, Documenta Mathematica, Extra Volume ICM 1998, I, 579–604.

[26] S. Mac Lane and I. Moerdijk, Sheaves in Geometry and Logic: A First Introduction to Topos Theory, Springer, 1992.

[27] The Univalent Foundations Program, Homotopy Type Theory: Univalent Foundations of Mathematics, Institute for Advanced Study, 2013.

[28] S.-T. Yau, On the Ricci curvature of a compact Kähler manifold and the complex Monge-Ampère equation, I, Communications on Pure and Applied Mathematics, 31 (1978), 339–411.

[29] H. Qin, SAE Physics Series: Four Forces Paper 1, Zenodo, 2026. DOI: 10.5281/zenodo.19342106.

[30] H. Qin, SAE Information Theory Series, Paper 4: Black Hole Information via Causal Spectrum, Zenodo, 2026. DOI: 10.5281/zenodo.19880111.

[31] H. Qin, SAE 道德律系列, Paper 1: 基础定理, Zenodo, 2026. DOI: 10.5281/zenodo.20011018.

[32] K. Uhlenbeck, Connections with $L^p$ bounds on curvature, Communications in Mathematical Physics, 83 (1982), 31–42.

[33] J. Cheeger and T. Colding, On the structure of spaces with Ricci curvature bounded below. I, Journal of Differential Geometry, 46 (1997), 406–480.

[34] J. Lott and C. Villani, Ricci curvature for metric-measure spaces via optimal transport, Annals of Mathematics, 169 (2009), 903–991.

[35] K.-T. Sturm, On the geometry of metric measure spaces. I, Acta Mathematica, 196 (2006), 65–131.

[36] F. Otto and C. Villani, Generalization of an inequality by Talagrand and links with the logarithmic Sobolev inequality, Journal of Functional Analysis, 173 (2000), 361–400.

[37] H. J. Sussmann, A cornucopia of four-dimensional abnormal sub-Riemannian minimizers, in Sub-Riemannian Geometry, Progress in Mathematics 144, Birkhäuser, 1996, 341–364.