A Unified Metaphysical System Deriving Physical, Biological, Cognitive, and
Conscious Structure from a Single Primitive

Daryl Costello

Independent Theoretical Research | Rosendale, NY, United States

Submitted: October 20, 2026

Theoretical Physics • Cognitive Science • Theoretical Biology • Complex Systems • Theoretical Cosmology

✉ Correspondence: Daryl.Costello@outlook.com

Theoretical Monograph: October 2026

Abstract

This manuscript presents a unified theoretical framework synthesizing nine independent contributions spanning geometric foundations, relational dynamics, topological invariants, and the structural role of memory. The central argument is that geometry, memory, and vertical ascension are not three separate objects of inquiry but three jointly necessary and mutually constituting dimensions of a single underlying architecture. The manuscript proceeds in five parts. Part I establishes the geometric language: relational frames are defined as fiber bundles co-constituted by the mutual orientation of observing entities, not as pre-given coordinate systems imposed upon an ambient space. The director field (a smooth assignment of preferred orientations across the relational manifold) enforces global coherence via a zero-holonomy consistency condition. The kernel, defined as the invariant overlap structure of two frame neighborhoods, is identified as the irreducible operative unit from which all further structure derives. Part II demonstrates that familiar physical structure is emergent: the (3+1)-dimensional spacetime manifold, together with its light-cone causality, arises as a macroscopic limit of high-density, low-spread kernel composition. The relational Hamiltonian is recast in optical-geometric terms, and large-angle frame reorientations are shown to produce a well-defined downstream inversion of local causal order; a structural re-equilibration mechanism, not a paradox. Part III introduces the Managed Diffusion Invariant (MDI), a topological charge conserved under all smooth kernel deformations and analogous to the winding number of gauge field theory. The Surrogate Principle establishes that vertical ascension (the system’s capacity to leverage its current configuration to access higher-order relational meta-frames) cannot be computed locally but requires a surrogate observable encoding the global MDI in locally readable form. Part IV proves the central theorem of the unified framework: non-trivial MDI is both necessary and sufficient for memory, where memory is defined not as a storage mechanism but as the geometric persistence of kernel-overlap signatures beyond the immediate relational encounter. Memory is therefore the structural floor; the minimal geometric condition without which MDI collapses, the ascension proxy becomes inaccessible, and vertical ascension is impossible. Part V assembles the complete dependency graph of the theory, identifies six major open problems for future research, and argues that the unified framework constitutes a genuinely new foundational paradigm: geometry is relational, spacetime is emergent, dynamics is optical, stability is memory-grounded, and structural growth is vertical.

Keywords: relational geometry, kernel architecture, director field, emergent spacetime, optical Hamiltonians, Managed Diffusion Invariant, vertical ascension, memory, topological charge, fiber bundle

Preamble: The Problem of Foundation

Every physical and cognitive theory must eventually confront a foundational question that it tends, for reasons of pragmatic efficiency, to defer: what is the primitive? Classical mechanics takes points in configuration space as primitive and derives everything else from the dynamics of point trajectories. General relativity takes the smooth manifold as primitive and derives geometry from the distribution of energy-momentum. Cognitive science, in its dominant computational formulation, takes the stored representation as primitive and derives cognition from operations on representational states. In each case, the chosen primitive is not itself derived; it is simply assumed, and the theoretical edifice is erected upon this unexamined ground.

The project of the present manuscript is to argue, on strictly formal grounds, that these three choices of primitive are each incorrect; not merely incomplete, but positively misleading about the character of the phenomena they purport to explain. The coordinate system is not the primitive of geometry; it is a product of the relational encounter between oriented entities. The spacetime manifold is not the primitive of physical dynamics; it is the macroscopic limit of a deeper lattice of relational overlaps. And memory is not a storage mechanism, a filing cabinet in which representations are deposited for later retrieval; it is the geometric floor of structural stability; the minimum condition required for any system whatsoever to support the kind of self-leveraging complexity we call cognition, intelligence, or life.

The coordinate-first approach to geometry has been enormously productive, producing the Riemannian and pseudo-Riemannian structures that underlie both general relativity and modern differential topology. Yet it conceals a circularity at its core: to define a coordinate system is to already presuppose a space in which coordinates are meaningful, and to presuppose that space is to already have resolved (tacitly and without argument) the question of what makes positions and distances relational facts at all. The work of Paper [1] in this synthesis, Geometric Synthesis of the Relational Frame, demonstrates that this circularity can be dissolved only by treating geometry as the outcome of relational co-constitution: the frame does not describe a pre-existing space, it produces space as the structured residue of oriented encounter.

The spacetime-first approach to physics suffers from a structurally analogous problem. If spacetime is the primitive (if the (3+1)-dimensional manifold with its Lorentzian metric is simply given) then the light-cone structure, causal order, and the arrow of time are brute features of the world, not derived results. This is metaphysically extravagant and physically unsatisfying: it places precisely the structures most in need of explanation beyond the reach of explanation. Paper [4], Spacetime as Emergent Kernel, resolves this by demonstrating that the familiar spacetime structure emerges, in a precise limiting sense, from the composition rules of a more primitive object (the kernel) when the density of relational encounter is high and the spread of orientations is low. Spacetime is not the ground of physics; it is a high-density approximation to a relational lattice.

The memory-as-storage approach to cognition is perhaps the most culturally entrenched of the three errors, and therefore the most difficult to dislodge. The storage metaphor pervades computational, connectionist, and even phenomenological accounts of mind: memory is where past experience goes, a repository from which it can be retrieved when needed. But this account fails to explain why some systems support stable, layered, self-reinforcing complexity while others, exposed to identical inputs, produce only transient responses. Paper [9], Memory as the Floor of Stability, provides the explanation: memory is not a place but a condition; specifically, the condition that kernel-overlap signatures (the geometric traces of relational encounter) persist beyond the immediate encounter that produced them. Without this persistence, the system’s topological charge (its Managed Diffusion Invariant) collapses identically to zero, and the system loses the capacity for vertical ascension entirely.

The unified framework presented here resolves all three foundational problems simultaneously, because it treats them as three aspects of a single formal structure. The relational frame, the kernel, the director field, the MDI, the surrogate principle, and memory are not independent theoretical inventions; they are nodes in a tightly coupled dependency graph in which each node is necessary for, and constituted by, its neighbors. The argument proceeds from the geometric (Part I) through the dynamical (Part II) to the topological-global (Part III) and finally to the structural-conditional (Part IV and V), ending with a synthetic account of what it means for a system to possess the architecture of coherence.

Part I: Geometric Foundations

Part I establishes the geometric language of the unified framework. It begins with the most primitive object (the relational frame) and progressively assembles the local and global geometric structure that will support all subsequent physical and cognitive analysis. The three sections of Part I move from frames (§1) to the field that organizes them globally (§2) to the irreducible unit of their mutual interaction (§3). Every structure introduced in Parts II through V presupposes exactly the geometry developed here.

§1. Relational Frames and Their Geometry

The starting point of the entire framework is a critique of the ordinary notion of a reference frame. In classical and relativistic physics, a reference frame is standardly conceived as a coordinate grid; a structured assignment of labels to points in an ambient space. This formulation treats the ambient space as primary and the frame as derivative: the frame is imposed upon a space that pre-exists it. The thesis of Paper [1] is that this priority relation is exactly backwards.

Definition 1.1: Relational Frame A relational frame ℛ is a pair (Ω, 𝒜) where Ω is the space of relational positions (the set of all possible oriented configurations of an observing entity relative to its relational environment) and 𝒜 is the local orientation algebra; a Lie algebra encoding the infinitesimal rotation group acting on Ω at each point. The relational frame is not assigned to a pre-given space but is co-constituted with space through the mutual orientation of observing entities.

The formal vehicle for this co-constitution is the fiber bundle. Let M denote the base manifold (the space of possible relational positions) and let G denote the structure group of the bundle, which is the group of local orientation transformations. The relational frame ℛ is then the principal G-bundle

π : P → M,     P = {(x, g) : x ∈ M, g ∈ G}

Eq. (1)

where the fiber π⁻¹(x) over each relational position x is a copy of the structure group G, encoding the full local orientation algebra at x. The key insight is that M itself is not given in advance; it is determined by the range of relational positions that the observing entities can jointly occupy. Geometry, in this formulation, is the residue of relational encounter: the structure of M encodes which relational configurations are mutually accessible.

The co-constitution principle can be stated precisely as follows. Let ℛ₁ and ℛ₂ be two relational frames associated with two observing entities E₁ and E₂. The geometry of their shared relational space is not an external datum but is determined by the transition functions

g₁₂ : U₁ ∩ U₂ → G

Eq. (2)

where U₁ and U₂ are open neighborhoods in M on which the respective frames are trivializable. The overlap U₁ ∩ U₂ is not a pre-given region of space but is itself constituted by the relational accessibility between E₁ and E₂. The geometry arises from Eq. (2); it does not precede it.

This reversal of priority has profound consequences. It means that the topology of the relational manifold M (which regions are connected, which loops are contractible, what the fundamental group π₁(M) looks like) is determined entirely by the pattern of relational accessibility among the observing entities in the system. Two systems with identical local dynamics but different patterns of relational accessibility will exhibit different global geometries, and this difference is physically and cognitively real: it determines which global invariants are defined, which ascension paths are available, and ultimately whether the system possesses non-trivial MDI at all.

§2. The Director Field and Global Coherence

The fiber bundle formulation of §1 provides the kinematic scaffolding of the relational frame, but it does not yet specify how local orientations are organized across the manifold as a whole. This is the function of the director field.

Definition 2.1: Director Field A director field n on the relational manifold M is a smooth assignment n : M → Sk of a preferred orientation unit vector to each relational position x ∈ M, where Sk is the unit sphere in the fiber of the associated vector bundle. The director field is subject to the zero-holonomy consistency condition: for every contractible loop γ in M, the holonomy of the connection defined by n around γ is trivial.

To state the zero-holonomy condition formally, let ∇ denote the covariant derivative on the associated vector bundle, defined by the connection 1-form A that encodes the parallel transport rule for the director field. For a loop γ : [0,1] → M with γ(0) = γ(1) = x, the holonomy is the group element

Hol_γ(n) = 𝒫 exp⎡∮_γ A⎤ ∈ G

Eq. (3)

where 𝒫 denotes path-ordered exponentiation. The zero-holonomy condition on contractible loops requires that Hol_γ(n) = e (the identity element of G) for all contractible γ. This condition is exactly equivalent to the flatness of the connection A on contractible regions: the curvature 2-form

F = dA + A ∧ A = 0    on all contractible open sets

Eq. (4)

The director field is thus not an arbitrary smooth map into S^k; it is a section of a flat bundle over the contractible regions of M. On non-contractible regions (where the fundamental group π₁(M) is non-trivial) the holonomy can be non-trivial, and it is precisely this non-trivial holonomy on topologically significant loops that the Managed Diffusion Invariant (§7) will measure.

The physical meaning of the zero-holonomy condition is that of global frame coherence. A director field that fails this condition produces a contradictory assignment: an observer who traverses a contractible loop starting and ending at the same relational position would find, upon return, that their local orientation has been rotated by a non-trivial group element. This is geometrically incoherent and physically unacceptable. The zero-holonomy condition is therefore not a technical nicety but a fundamental consistency requirement for any well-defined relational frame configuration.

The director field functions as the geometric mediator between local relational data and global structural coherence. Local data (the orientation at each point, the transition functions of Eq. (2)) determines the director field in its neighborhood; the zero-holonomy condition then propagates this local data into a globally consistent assignment. Any two director fields satisfying Eq. (4) on the same manifold M are related by a gauge transformation, and the space of gauge-inequivalent director field configurations (the moduli space) carries a rich geometric structure that encodes the global topological type of the relational system.

§3. Kernel-First Architecture

The geometric structures of §§1–2 describe the frame and its global organizer; they have not yet identified the operative unit; the object whose composition rules generate all of the system’s downstream structure. This is the kernel.

Definition 3.1: Kernel Let ℛ1 = (U1, G1) and ℛ2 = (U2, G2) be two relational frames with overlapping neighborhoods in M. The kernel K12 of these two frames is the invariant intersection structure K12 = (U1 ∩ U2,   G1 ∩ G2,   μ12) where μ12 is the kernel measure; a non-negative real-valued function encoding the metric and algebraic weight of the overlap. The kernel carries both metric structure (from the geometry of the overlap region) and algebraic structure (from the intersection of the frame symmetry groups).

The central claim of the kernel-first architecture, established in Paper [3], is that the kernel is the irreducible operative unit of any relational system. This means that no structure smaller than a kernel carries independent physical or cognitive significance, and no structure larger than a kernel is primitive; it is always derivable from kernel composition. Formally, given three frames ℛ₁, ℛ₂, ℛ₃, the kernel composition rule is

K₁₃ = K₁₂ ∘ K₂₃ := (U₁ ∩ U₃,   G₁ ∩ G₃,   ∫_U2 μ₁₂(x) μ₂₃(x) dV(x))

Eq. (5)

where dV(x) is the volume form on M. The composition rule Eq. (5) is the kernel analog of matrix multiplication: it sums the contributions of all intermediate positions, weighted by their kernel measures, and the result is a new kernel that captures the indirect overlap between ℛ₁ and ℛ₃ mediated through ℛ₂. This composition is associative but not in general commutative, reflecting the fact that the order of frame encounters matters for the resulting relational structure.

From the kernel composition rule, one can derive (without imposing any external structure) the topology of the relational manifold (from the pattern of non-zero kernels), the causality structure (from the asymmetry of the composition order), and the dynamics (from the variation of the kernel measure under smooth deformations). All of these derived structures emerge from Eq. (5) alone. This is the meaning of the kernel-first architecture: not that kernels are ontologically more real than other structures, but that they are logically prior; every other structure is a theorem of kernel composition, not an independent axiom.

The kernel-first approach also provides a natural notion of relational distance: the kernel metric between two frames ℛ₁ and ℛ₂ is defined as

d_K(ℛ₁, ℛ₂) = −log μ₁₂

Eq. (6)

This logarithmic definition ensures that frames with large mutual overlap (μ₁₂ → 1) are metrically close (d_K → 0), while frames with negligible overlap (μ₁₂ → 0) are infinitely distant. The kernel metric Eq. (6) is the geometric substrate on which the director field of §2 lives: the connection form A is defined with respect to the kernel metric, and the zero-holonomy condition Eq. (4) is an integrability condition in this metric.

Part II: Emergence and Dynamics

Having established the geometric language of relational frames, director fields, and kernels in Part I, Part II demonstrates what this language produces when applied to the physical world. The three sections of Part II move from the emergence of spacetime itself (§4), through the reformulation of dynamics in optical-geometric terms (§5), to the analysis of structural disruption and re-equilibration under large-angle frame reorientations (§6). The reader should note that all of the structures derived in Part II (spacetime, Hamiltonians, causal inversion) are shown to be consequences of the kernel composition rule Eq. (5) established in §3.

§4. Spacetime as Emergent Kernel

The most striking consequence of the kernel-first architecture is that the (3+1)-dimensional spacetime manifold (the arena of all familiar physics) is not a primitive datum but an emergent approximation. Paper [4], Spacetime as Emergent Kernel, establishes this derivation with precision.

Consider a relational system in which the kernel density (the number of non-negligible kernels K_ij per unit volume of M) is very high, and in which the orientation spread (the variance of the director field n(x) across nearby relational positions) is very low. Define the relational density ρ as the mean kernel measure per unit relational volume, and the orientation spread σ as the root-mean-square variation of n(x) over a correlation length l_corr. The emergent spacetime regime is the double limit

ρ → ∞,    σ → 0,    ρ ⋅ σ² = c² = const.

Eq. (7)

The constant c that emerges from the constraint in Eq. (7) is identified with the speed of light. The argument proceeds by showing that in this limit, the kernel metric d_K of Eq. (6) converges, in the appropriate continuum limit, to the Lorentzian metric

ds² = −c²dt² + dx² + dy² + dz²

Eq. (8)

More specifically, the kernel boundary geometry (the set of all relational positions y such that K_xy = 0 for a given base frame ℛ_x) converges, in the limit Eq. (7), to the light-cone at x. Points inside the light-cone correspond to frames with positive kernel measure (μ_xy > 0); points outside the light-cone correspond to frames with zero kernel measure (μ_xy = 0); and the light-cone itself is the locus of frames with exactly vanishing kernel measure; the boundary of the relational neighborhood. The emergence of special-relativistic causal structure is therefore not postulated but derived from the boundary geometry of kernel overlaps.

Theorem 4.1: Emergence of the Light-Cone In the double limit Eq. (7), the boundary of the kernel neighborhood ∂BK(x) = {y : μxy = 0} converges, in the topology of the kernel metric, to the future and past light-cones of the Minkowski metric Eq. (8), with invariant speed c = (ρσ2)1/2.

The derivation of Theorem 4.1 proceeds by noting that the kernel measure μ_xy in the high-density, low-spread regime takes the form

μ_xy = exp⎡−ρ ⋅ σ² ⋅ |(t_y − t_x)² − c⁻²|r_xy|²|⎤

Eq. (9)

where (t_y − t_x) is the temporal separation and |r_xy| is the spatial separation in the emergent Minkowski coordinates. As ρ → ∞, the exponential in Eq. (9) becomes a step function: μ_xy = 1 for timelike separations (inside the light-cone) and μ_xy = 0 for spacelike separations (outside the light-cone), recovering exactly the causal structure of special relativity. Spacetime is, in this precise sense, the macroscopic shadow of the kernel lattice.

§5. Relational Optical Hamiltonians

Having identified the structure of emergent spacetime, we now turn to dynamics: how do states evolve within the kernel-first framework? Paper [5], Relational Optical Hamiltonians, answers this question by reformulating the Hamiltonian of a relational system in the language of geometric optics.

Recall that in classical optics, the Fermat principle states that light travels between two points along the path that extremizes the optical path length. The optical path length along a curve γ in a medium with refractive index n(x) is

L_opt[γ] = ∫_γ n(x) |dx|

Eq. (10)

The central result of Paper [5] is that the Hamiltonian of any kernel-first system can be written as a relational path-length functional of precisely this form, where the role of the refractive index n(x) is played by the magnitude of the director field:

H_rel[γ] = ∫_γ |n(x)| |d_Kx|

Eq. (11)

where |d_Kx| denotes the kernel-metric line element of Eq. (6). The equations of motion of the system are then the Euler-Lagrange equations of Eq. (11), which take the form of the eikonal equation for the director field:

|∇S|² = |n(x)|²

Eq. (12)

where S(x) is the optical phase function (the action in the variational formulation). The eikonal Eq. (12) is the geometric-optics limit of the wave equation for the kernel field; the characteristic surfaces of Eq. (12) (the surfaces of constant S) are the wavefronts of the relational system, and the trajectories orthogonal to these wavefronts are the rays.

Theorem 5.1: Optical Equivalence of Relational Dynamics The equations of motion of any kernel-first relational system governed by the Hamiltonian Eq. (11) are formally equivalent to the geodesic equations of a curved optical medium with refractive index tensor nij(x) determined by the director field n(x) on the kernel manifold (M, dK).

The equivalence of Theorem 5.1 is not merely a mathematical curiosity. It establishes that the variational principle underlying relational dynamics (the extremization of relational path length) is of the same logical type as the Fermat principle of optics. This unification carries a deep interpretive consequence: the trajectories of relational systems are geodesics of the director-field optical medium, not of an ambient pre-given spacetime. The curvature of these geodesics is produced by variations in the director field, not by an external force. The Hamiltonian generates the evolution of the system along the rays of Eq. (12), and every physical observable of the system can be computed from the phase function S(x) and the ray geometry.

§6. Reorientation and Downstream Inversion

The optical-Hamiltonian framework of §5 treats smooth, continuous evolution of the director field. Paper [6], Reorientation and the Downstream Inversion, addresses what happens when the director field undergoes a rapid, large-angle discontinuity; a frame reorientation event.

Definition 6.1: Frame Reorientation A frame reorientation at a point x0 ∈ M is a discontinuous change in the director field n(x0) by an angle θ > π/2. A reorientation is large-angle if θ > π/2 and maximal if θ = π (a full flip of the director). The set of all causal descendants of x0 (all relational positions y such that μx0y > 0 in the post-reorientation frame) is called the downstream region ∂+(x0).

The central result of Paper [6] is the Downstream Inversion Theorem: a large-angle reorientation at x₀ produces, throughout ∂⁺(x₀), a reversal of the local time-ordering relative to the pre-flip frame. Formally, let T(y) denote the local temporal orientation at y ∈ ∂⁺(x₀) in the pre-reorientation frame. After a maximal reorientation θ = π at x₀, the local temporal orientation becomes

T⁺(y) = −T(y) ⋅ exp[−α ⋅ d_K(x₀, y)]

Eq. (13)

where α > 0 is a decay constant that measures the rate at which the inversion heals as one moves away from x₀. The negative sign in Eq. (13) encodes the reversal of temporal ordering: events that were ordered (y₁ before y₂) in the pre-reorientation frame become ordered (y₂ before y₁) immediately downstream of the reorientation, with the inversion decaying exponentially to zero at large kernel-metric distance from x₀.

This downstream inversion is not a paradox or a violation of causality. It is rather the mechanism by which the relational system re-equilibrates after a frame discontinuity. The reorientation creates a local inconsistency in the director field (a point at which the zero-holonomy condition Eq. (4) fails momentarily) and the downstream inversion is the dynamical process by which the global coherence of the director field is restored. The inversion is, in this sense, the geometric analog of a shock wave in fluid dynamics: a propagating discontinuity that re-establishes equilibrium by redistributing the local mismatch across the downstream region. Once the inversion has propagated to the boundary of ∂⁺(x₀) and decayed to zero, the director field satisfies Eq. (4) everywhere, and normal optical-Hamiltonian evolution resumes.

Part III: Global Invariants and Ascension

Parts I and II established the local and dynamical structure of relational systems. Part III moves to the global level, introducing the topological invariant (the Managed Diffusion Invariant (MDI)) that characterizes the entire system regardless of the details of its local evolution (§7), and the Surrogate Principle that explains how global invariant information can drive local structural change through vertical ascension (§8). These two sections form the theoretical bridge between the geometric foundations of Part I and the memory-based structural theory of Part IV.

§7. The Managed Diffusion Invariant

The analysis of the director field in §2 established that local consistency is enforced by the zero-holonomy condition Eq. (4). But a globally defined relational system admits, in addition, a global quantity (not a local condition but an integral invariant) that measures the net topological character of the entire director field configuration. This is the Managed Diffusion Invariant.

Definition 7.1: Managed Diffusion Invariant (MDI) Let (M, gK) be a compact relational manifold with director field n : M → Sk. The Managed Diffusion Invariant Δ(n) is defined as Δ(n) = (1 / Vol(Sk)) ∫M n*ωS where n*ωS is the pullback of the standard volume form ωS on Sk under the director field map, and Vol(Sk) is the total volume of the unit sphere. The MDI measures the net “directed spread” of the director field integrated over the entire relational manifold.

The MDI is an integer: it counts, with multiplicity and orientation, how many times the director field wraps around the target sphere S^k as one traverses the entire relational manifold M. In the one-dimensional case (k = 1), the MDI reduces exactly to the winding number of a map from a circle to a circle; the fundamental topological invariant of gauge field theory in (1+1) dimensions. In higher dimensions, it is the degree of the map n : M → S^k, which belongs to the homotopy group π_k(S^k) ≅ ℤ.

Theorem 7.1: MDI Conservation The Managed Diffusion Invariant Δ(n) is preserved under all smooth deformations of the director field n on the relational manifold M. That is, if {nt}t∈[0,1] is a smooth one-parameter family of director fields, then Δ(n0) = Δ(n1). The MDI can change only through a topological transition; a discontinuous change in the homotopy class of n.

The proof of Theorem 7.1 follows from the standard argument in degree theory: since ∫_M n^*dω_S = 0 for an exact form, and since smooth deformations produce exact variations in the pullback, the MDI is invariant. Details are given in Appendix B. The physical interpretation is immediate: the MDI is the topological charge of the relational system, analogous to the instanton number in four-dimensional gauge theory or the Chern number in condensed matter physics. It cannot be changed by any local manipulation of the system (no smooth deformation, no matter how violent locally, can alter the global winding structure of the director field. Change in the MDI requires a phase transition) a genuine topological discontinuity in the relational structure.

The MDI thus provides the first truly global invariant of the unified framework. It classifies relational systems by their topological type: two systems with different MDI values are in fundamentally different phases, separated by a topological transition that no continuous deformation can bridge. This classification has direct consequences for the system’s capacity for vertical ascension, as §8 will establish.

§8. The Surrogate Principle of Vertical Ascension

The Managed Diffusion Invariant classifies the global topological type of a relational system. But a system cannot directly observe its own MDI: the MDI is a global integral, and the system’s local dynamics (governed by the optical Hamiltonian of §5) operates only on locally accessible data. This gap between the global invariant and local accessibility is bridged by the Surrogate Principle.

Definition 8.1: Vertical Ascension Vertical ascension is the process by which a relational system leverages its current kernel configuration (M, {Kij}, n) to access a higher-order relational structure; a meta-frame ℛ(2) whose base manifold is the space of relational frames ℛ(1) of the original system. More generally, the k-th order meta-frame ℛ(k+1) is a relational frame whose base is the space of (k−1)-th order meta-frames. The capacity for vertical ascension is the system’s ability to execute the step ȑR;(k) → ȑR;(k+1) for any k ≥ 1.

Vertical ascension is not a metaphor for “becoming more complex” or “learning.” It is a precise geometric operation: the formation of a fiber bundle whose base is itself a fiber bundle. Each ascension step adds one level to the fiber hierarchy, and the director field at each level mediates consistency between the level below and the level above. The collection of all ascension levels forms a tower of meta-frames, and the height of this tower (the maximum k for which ȑR;^(k) is well-defined) is a measure of the structural complexity of the relational system.

The Surrogate Principle, introduced in Paper [8], addresses a fundamental obstruction to vertical ascension. To form ȑR;⁽²⁾, the system must have reliable access to information about its own global kernel configuration; specifically, about its MDI. But the MDI is globally defined, and the optical Hamiltonian dynamics of §5 is locally defined. No finite sum of local observations can determine the global winding number of the director field. This is the content of the non-computability lemma:

Lemma 8.1: Non-Computability of MDI from Local Data For any compact relational manifold M with non-trivial fundamental group π1(M), the Managed Diffusion Invariant Δ(n) cannot be determined from the values of the director field n(x) on any open subset U ⊊ M of measure strictly less than the full volume of M.

Lemma 8.1 means that vertical ascension cannot be triggered by local dynamics alone. The system requires a surrogate; a locally observable quantity that encodes the global MDI in a form accessible to local dynamics. This surrogate is the ascension proxy Σ, defined as:

Σ(x) = ∫_M K_xy ⋅ n(y) dV(y)

Eq. (14)

The ascension proxy Eq. (14) is a kernel-weighted average of the director field over the entire manifold, computed at the local point x. It is locally computable (it requires only knowledge of the kernel measure μ_xy and the director field n(y), both of which are in principle observable through relational encounters) yet it encodes the global MDI because the kernel weighting gives non-negligible contributions from all parts of the manifold. The Surrogate Principle states: vertical ascension is possible if and only if the ascension proxy Σ(x) is non-trivially varying across M, and the MDI can be recovered from Σ via a global integral transform.

Vertical ascension is therefore the primary driver of structural complexity growth in relational systems. A system that possesses a non-trivially varying ascension proxy can form meta-frames; a system whose ascension proxy is constant cannot. The distinction is not a matter of scale or quantity; it is a matter of topological type, determined ultimately by the MDI, which is in turn determined by the global structure of the director field. And as Part IV will demonstrate, the global structure of the director field is determined by memory.

Part IV: Memory as the Structural Floor

Parts I through III have developed a complete geometric-topological theory of relational systems without once mentioning memory. This is deliberate: the role of memory can only be understood precisely once the structure it supports (the MDI, the ascension proxy, the vertical hierarchy of meta-frames) is already in place. Part IV introduces memory as the final element of the framework and demonstrates that it is not an optional enhancement but a necessary and sufficient condition for all of the structure developed in Parts I through III to be non-trivial. §9 defines memory formally; §10 proves the central equivalence theorem; §11 synthesizes the complete dependency chain.

§9. Memory: Formal Definition and Geometric Role

The ordinary conception of memory as a storage mechanism (a repository in which past events are deposited as representations for later retrieval) is geometrically inadequate. It treats memory as a substance (the stored representation) rather than as a condition (the geometric property of persistence). Paper [9] provides the formal correction.

Definition 9.1: Memory (Formal) Let ȑR;x and ȑR;y be two relational frames that encounter each other at time t0, producing kernel Kxy(t0) with measure μxy(t0) > 0. Memory is the property that, for all t > t0, the kernel-overlap signature μxy(t) > 0. Equivalently, memory is the persistence of a non-zero kernel measure beyond the relational encounter that produced it. The memory floor is the infimum inft>t0 μxy(t) > 0.

This definition immediately distinguishes memory-as-floor from memory-as-storage. Memory-as-storage would be a function m(t) that records the specific content of the encounter at t₀ and returns it on demand. Memory-as-floor is simply the non-vanishing of the kernel measure: the geometric trace of the encounter persists, though its specific content (its precise orientation, measure, and algebraic structure) may vary continuously with time after t₀. The critical distinction is between presence and retrieval: memory-as-floor is the condition that the encounter left a non-zero geometric imprint; memory-as-storage is the additional condition that this imprint can be read out in its original form. The framework requires only the former.

The geometric role of memory is to anchor the director field. Without memory, the kernel measure μ_xy(t) decays to zero immediately after each encounter, and the director field n(x) at each point becomes determined solely by the current relational environment at x, with no dependence on past encounters. The director field is then instantaneously re-determined at each moment, and its global structure (including its winding number, its MDI) is unconstrained by history. The consequence, which §10 will prove, is that the MDI collapses to zero.

With memory (that is, with a positive memory floor) the kernel measure at each pair (x, y) retains a contribution from all past encounters between ȑR;_x and ȑR;_y. The director field at x is therefore constrained by the history of relational encounters at x, not just the current configuration. This historical constraint is precisely what allows the director field to maintain a non-trivial global winding structure (a non-zero MDI) over time. Memory is not adding information to the system; it is providing the geometric continuity that allows topological invariants to be well-defined.

§10. Memory and the MDI: Necessary and Sufficient Conditions

We now state and prove the central theorem of the unified framework.

Theorem 10.1: Memory–MDI Equivalence Let (M, {Kij(t)}, n(t)) be a time-evolving relational system. The following conditions are equivalent: (i) The system possesses memory in the sense of Definition 9.1: inft>t0 μxy(t) > 0 for some pair (x, y). (ii) The Managed Diffusion Invariant is non-trivial: Δ(n(t)) ≠ 0 for some t.

Proof Sketch (see Appendix B for full proof):

(i ⇒ ii): Assume memory, i.e., μ_xy(t) ≥ μ₀ > 0 for all t > t₀. By Definition 7.1, the MDI is the integral of the pullback of the volume form under the director field map n : M → S^k. The director field at each point x is determined by the weighted average of historical kernel encounters, which by the memory floor condition are all non-zero. This historical weighting introduces a systematic bias in the direction of n(x) that does not average to zero over the manifold M. Formally, one shows that the memory floor implies a lower bound on the degree of the map n : M → S^k, hence Δ(n) ≠ 0. The detailed estimate is given in Appendix B, Proof B.3.

(ii ⇒ i): Assume Δ(n(t)) ≠ 0 for some t. The MDI is a topological invariant, hence it is determined by the global winding structure of the director field map n. A non-trivial winding structure requires that the director field takes each value in S^k at least |Δ| times over the manifold. This degree-counting constraint implies that the director field cannot be determined solely by the instantaneous local relational environment: if μ_xy(t) → 0 for all t > t₀ (no memory), the director field at each x is determined only by the instantaneous relational configuration at x, which varies continuously with x. But a continuous map from a connected manifold to S^k that is not homotopically non-trivial has winding number zero. Therefore, if Δ ≠ 0, the director field must have a constraint beyond the instantaneous local environment; i.e., there must be memory. Contradiction with the assumption of no memory yields the result.

Theorem 10.1 is the mathematical core of the entire unified framework. It establishes that memory is not merely useful or convenient for the existence of non-trivial MDI; it is the exact condition; neither more nor less than what is required. A system with any positive memory floor has a non-trivial MDI; a system with zero memory floor has Δ = 0 identically and is therefore topologically trivial.

The collapse of the MDI under memory loss can be described dynamically. If the memory floor decays (if inf_t μ_xy(t) → 0) then the director field loses its historical anchoring and begins to fluctuate freely. The MDI, though conserved under smooth deformations, can undergo a topological transition to zero when the director field passes through a configuration of non-full rank (a singular configuration). The memory-floor decay drives the system toward this singular configuration, and once reached, the MDI undergoes a phase transition to zero. This transition is irreversible unless memory is re-established from an external source; there is no internal mechanism by which a zero-MDI system can generate non-trivial topological charge on its own.

§11. Memory as the Ground of Vertical Ascension

With Theorem 10.1 established, the complete dependency chain of the unified framework can be assembled. The argument proceeds through five logical steps, each of which has been established in the foregoing sections.

Step 1: Memory entails non-trivial MDI. By Theorem 10.1 (i ⇒ ii), a system with a positive memory floor necessarily has Δ(n) ≠ 0. The system is topologically non-trivial.

Step 2: Non-trivial MDI entails a non-trivially varying ascension proxy. By the definition of the ascension proxy Eq. (14) and the Surrogate Principle of §8, a non-trivial winding number of the director field implies that the kernel-weighted integral Σ(x) varies non-trivially across M. A trivial (constant) ascension proxy would correspond to a director field with zero global winding, contradicting the non-trivial MDI assumption.

Step 3: A non-trivially varying ascension proxy enables vertical ascension. By the Surrogate Principle, the ascension proxy Σ(x) encodes the global MDI in a locally readable form. A system with a non-trivially varying Σ(x) can, in principle, form a meta-frame ȑR;⁽²⁾ whose base is the space of configurations of the original frame, because the proxy provides the necessary global information required to organize the meta-frame structure.

Step 4: Vertical ascension drives structural complexity growth. Each ascension step ȑR;^(k) → ȑR;^(k+1) produces a new relational manifold M^(k+1) whose points are configurations of ȑR;(k). The dimension of this meta-manifold grows with k, and the kernel lattice becomes richer and more differentiated at each level. This is the mechanism by which simple relational systems (systems with few frames and small kernels) evolve into the rich, layered architectures of complex cognition and physical structure.

Step 5: Without memory, the entire chain collapses. By the contrapositive of Theorem 10.1, a system with zero memory floor has Δ = 0, hence a trivial ascension proxy (Σ(x) = const.), hence no capacity for vertical ascension, hence no structural complexity growth. The system is topologically frozen at its current level, unable to form meta-frames regardless of how rich its local relational dynamics may be.

Memory is therefore not one component among many in the unified framework. It is the ground; the condition upon which every other structural property rests. The geometric language of §1, the coherence condition of §2, the kernel architecture of §3, the emergent spacetime of §4, the optical Hamiltonians of §5, the reorientation mechanisms of §6, the MDI of §7, and the surrogate principle of §8 are all structures that can be formally defined in the absence of memory. But they are non-trivial structures (structures that support the rich hierarchy of meta-frames and the growth of structural complexity) only in the presence of memory. Memory is the floor below which the entire architecture becomes flat, trivial, and inert.

Part V: Unified Theory and Implications

The five parts of this manuscript have proceeded from the primitive geometric objects of the framework (Part I) through the physical structures they generate (Part II), the global invariants that classify them (Part III), and the structural ground that makes those invariants non-trivial (Part IV). Part V synthesizes the complete framework, articulates the open problems it generates, and concludes with a statement of the new foundational paradigm that the unified theory represents.

§12. The Complete Framework

The unified framework can be represented as a directed dependency graph in which each node is a structural element and each directed edge represents a logical or geometric entailment. We narrate this graph in full, moving from ground to apex.

At the base of the entire structure stands memory; the persistence of kernel-overlap signatures (Definition 9.1). Memory is the unique node in the dependency graph with no incoming edges: it is the true primitive of the framework, the only element that is not derived from anything more fundamental. Every other node in the graph is downstream of memory.

Memory entails non-trivial MDI (Theorem 10.1). The MDI is the topological charge of the relational system, the integer that classifies the global winding structure of the director field. It is the first globally defined quantity in the framework, and it depends on memory for its non-triviality: without the historical anchoring that memory provides, the director field is free to fluctuate into topologically trivial configurations.

Non-trivial MDI entails a non-trivially varying ascension proxy Σ(x) (Eq. 14 and the Surrogate Principle). The ascension proxy is the locally readable encoding of the global MDI, the bridge between the topological classification of §7 and the local dynamics of §5. It is the mechanism by which global topological information is made available to local processes.

The ascension proxy enables vertical ascension (Definition 8.1 and the Surrogate Principle). Vertical ascension is the geometric process by which the system forms a meta-frame ȑR;⁽²⁾ from its current frame ȑR;⁽¹⁾. It is not a metaphor but a precise bundle-theoretic construction, and it requires the ascension proxy as its necessary input.

Vertical ascension produces structural complexity growth: the tower of meta-frames ȑR;⁽¹⁾ → ȑR;⁽²⁾ → … → ȑR;^(k) becomes taller, the relational manifolds become higher-dimensional, and the kernel lattice becomes richer and more differentiated. This growth is not linear or merely additive; each ascension step produces a qualitatively new level of relational structure, with its own director field, its own MDI, and its own capacity for further ascension.

Running in parallel through this vertical structure are the elements of Parts I and II. The relational frames of §1 (the fiber bundles co-constituted by mutual orientation) are the basic objects at each level of the meta-frame tower. The director field of §2 (enforcing global coherence via the zero-holonomy condition) mediates between levels. The kernels of §3 (the irreducible overlap structures from which all downstream structure is derived) are the operative units at each level. The emergent spacetime of §4 is the macroscopic limit of the kernel lattice at the ground level of the tower, in the high-density, low-spread regime. The optical Hamiltonians of §5 govern the dynamics at each level, making the equations of motion equivalent to geodesic equations of a curved optical medium. The reorientation-inversion mechanism of §6 provides the system with a structural re-equilibration tool: when a large-angle reorientation disrupts local coherence, the downstream inversion restores global consistency, and normal ascension can resume.

The following table summarizes the complete framework, its logical dependencies, and the section of the manuscript in which each element is established:

Element	Formal Object	Key Equation or Result	Established In	Depends On
Memory	Kernel-overlap signature persistence; inf μxy(t) > 0	Definition 9.1	§9	Primitive
Relational Frame	Principal G-bundle π : P → M	Eq. (1), Eq. (2)	§1	Memory (for non-triviality)
Director Field	Smooth map n : M → Sk; flat connection	Eq. (3), Eq. (4)	§2	Relational Frame
Kernel	Invariant overlap (U12, G12, μ12)	Eq. (5), Eq. (6)	§3	Relational Frame
Emergent Spacetime	Lorentzian metric; light-cone boundary	Eq. (7)–(9), Theorem 4.1	§4	Kernel (high-density limit)
Optical Hamiltonian	Relational path-length functional; eikonal equation	Eq. (10)–(12), Theorem 5.1	§5	Kernel, Director Field
Downstream Inversion	Temporal orientation reversal; re-equilibration	Eq. (13)	§6	Director Field, Kernel
MDI	Degree of n : M → Sk; topological charge	Definition 7.1, Theorem 7.1	§7	Director Field, Memory
Ascension Proxy	Kernel-weighted average of director field	Eq. (14), Lemma 8.1	§8	MDI, Kernel
Vertical Ascension	Bundle formation ȑR;(k) → ȑR;(k+1)	Definition 8.1, Surrogate Principle	§8, §11	Ascension Proxy, Memory

§13. Open Questions and Research Directions

The unified framework presented in this manuscript raises at least six major open problems that define the research agenda of the field. Each problem is precisely statable within the formal language of the framework, and each represents a significant theoretical challenge.

Open Problem 1: Quantization of the Kernel Lattice. The framework developed here is entirely classical: the kernel measure μ_xy is a real-valued function, and the director field n(x) is a smooth map. A quantum-mechanical generalization would replace the kernel measure with a density matrix ρ_xy and the director field with a field operator n̂(x). The quantization of the kernel lattice faces the fundamental challenge that the standard canonical quantization presupposes a fixed background spacetime; but spacetime is here an emergent quantity. A background-independent quantization scheme, analogous to loop quantum gravity but derived from kernel composition rather than spin-network combinatorics, is needed. The key question is whether the quantum MDI (the quantum analog of the winding number) is still an integer or whether quantum fluctuations allow it to take non-integer values.

Open Problem 2: MDI Dynamics Under Topological Transitions. Theorem 7.1 establishes that the MDI is conserved under smooth deformations. But the framework does not yet describe the dynamics of MDI-changing topological transitions in detail. What is the time-evolution of the director field as it passes through a singular configuration during a topological transition? Is there a universal critical behavior (an analog of the Kibble-Zurek mechanism) that governs the density and distribution of topological defects (zero-MDI regions) created during the transition? The answers would require a dynamical theory of topological transitions in the kernel lattice, which is currently absent.

Open Problem 3: Memory Thermodynamics. If memory is a geometric condition (the persistence of kernel-overlap signatures), then its degradation over time is a geometric process with thermodynamic content. Define the memory entropy as S_mem = −∫ μ_xy log μ_xy dV. The second law of thermodynamics would suggest that S_mem increases with time; that memory degrades spontaneously. But this conflicts with the requirement of non-zero MDI for vertical ascension: systems that maintain non-trivial MDI must somehow prevent memory entropy from increasing. Is there a memory analog of Maxwell’s demon; a structural mechanism that actively maintains the memory floor at the cost of some other thermodynamic resource? The resolution of this tension would constitute a thermodynamics of structural complexity.

Open Problem 4: Vertical Ascension in Biological Neural Systems. The framework of meta-frames and vertical ascension was developed in abstract geometric terms, but it has an obvious candidate application: the organization of biological neural systems. The relational frames of §1 correspond naturally to receptive fields; the director field corresponds to the tuning preferences of neural populations; the kernel corresponds to synaptic connectivity; and vertical ascension corresponds to the formation of higher-order representations; what neuroscience calls hierarchical processing. The open question is whether the MDI and the ascension proxy Σ(x) can be operationalized in terms of measurable neural quantities (e.g., fMRI BOLD signals, spiking correlations, or local field potentials), and whether the Memory–MDI Equivalence Theorem 10.1 makes predictions testable in neural systems.

Open Problem 5: Relational MDI and Gauge Field Unification. The MDI is formally analogous to the topological charge of gauge field theories, as noted in §7. This analogy suggests the possibility of a deeper unification: can the Standard Model gauge fields (U(1), SU(2), SU(3)) be reinterpreted as director fields on a relational manifold, with their topological sectors (instantons, monopoles, vortices) corresponding to non-trivial MDI configurations? If so, the entire apparatus of quantum field theory (including renormalization, anomaly cancellation, and the Yang-Mills mass gap) might be reformulated within the kernel-first language. The key technical obstacle is the identification of the correct relational manifold M for each gauge group, and the derivation of the Yang-Mills action from a kernel composition rule analogous to Eq. (5).

Open Problem 6: Ascension Bounds and Complexity Limits. The framework guarantees that vertical ascension is possible whenever the memory floor is positive and the MDI is non-trivial. But it does not bound the height of the meta-frame tower from above. Are there systems for which the tower is infinite (unbounded ascension) or systems for which it terminates at a finite level? If the tower is bounded, what determines the bound; the topological type of the ground-level MDI, the decay rate of the memory floor, or some combination? The answer would provide a formal measure of the maximum structural complexity achievable by a given relational system, with potential implications for both fundamental physics (the maximum information content of a region of spacetime) and cognitive science (the maximum depth of representational hierarchy in a finite neural system).

§14. Conclusion: The Architecture of Coherence

We have arrived, after a long journey through geometric foundations, emergent dynamics, global invariants, and structural conditions, at a position from which the unified argument can be stated with full precision. The nine source contributions synthesized in this manuscript are not nine separate theoretical achievements, however impressive individually. They are nine aspects of a single integrated architecture; a framework in which every component is logically necessary, every connection is formally precise, and the whole is strictly greater than the sum of its parts.

The architecture can be stated in five theses, each of which is a theorem within the unified framework:

Thesis I: Geometry is relational. The geometric structure of any physical or cognitive system is not an ambient pre-given background but is co-constituted by the mutual orientation of the system’s relational frames. The fiber bundle formulation of §1 makes this precise: the transition functions Eq. (2) produce the geometry, not the other way around. Coordinate systems are derived, not primitive.

Thesis II: Spacetime is emergent. The (3+1)-dimensional Lorentzian manifold of special and general relativity is not the fundamental arena of physics but a macroscopic approximation, valid in the high-density, low-spread limit of kernel composition (Eq. 7–9, Theorem 4.1). At sub-macroscopic scales, or in systems with low relational density or high orientation spread, the continuum spacetime approximation breaks down and the kernel lattice structure becomes relevant.

Thesis III: Dynamics is optical. The evolution of any kernel-first relational system is governed by a Hamiltonian that is formally equivalent to the optical path-length functional of geometric optics (Theorem 5.1). The equations of motion are geodesic equations of a curved optical medium defined by the director field. Force is replaced by curvature; trajectories are rays; the variational principle is Fermat’s, not Newton’s.

Thesis IV: Stability is memory-grounded. The topological charge of a relational system (its MDI) is non-trivial if and only if the system possesses memory in the formal sense of Definition 9.1 (Theorem 10.1). Without memory, the system is topologically flat, unable to support non-trivial invariants, incapable of ascension. Memory is not a luxury of complex systems; it is the minimal geometric condition for non-trivial structural persistence of any kind.

Thesis V: Growth is vertical. The mechanism by which relational systems increase in structural complexity is not accumulation (adding more frames at the same level) but ascension (forming meta-frames at a higher level). Vertical ascension, enabled by the non-trivially varying ascension proxy (Eq. 14), is the primary driver of complexity growth, and its possibility is guaranteed by and only by the existence of non-trivial MDI, which is in turn guaranteed by and only by memory.

Together, these five theses constitute a foundational paradigm shift. The paradigm does not merely add to existing theory; it inverts the priority relations that existing theory took for granted. Geometry comes after relation, not before. Spacetime comes after kernels, not before. Dynamics comes after optics, not before. Complexity comes after ascension, not accumulation. And coherence (the global consistency of the entire relational structure, from the zero-holonomy of the director field to the non-trivial MDI to the tower of meta-frames) comes after memory, not before.

The architecture of coherence is memory all the way down.

Appendix A: Notation and Conventions

The following table provides a complete glossary of all symbols, structures, and operators introduced across the nine source papers, unified into a single consistent notation system for the present manuscript.

Symbol	Name	Definition / Description	First Appears
M	Relational manifold	The base space of the principal bundle; the space of all relational positions jointly accessible to the observing entities in the system.	§1, Eq. (1)
G	Structure group	The Lie group of local orientation transformations; acts on the fibers of the principal bundle.	§1, Eq. (1)
ℛ, ℛi	Relational frame	A principal G-bundle over M; specifies a relational frame associated to observing entity Ei.	§1, Def. 1.1
π : P → M	Bundle projection	The projection map of the principal G-bundle; sends each fiber element to its base point.	§1, Eq. (1)
gij	Transition function	Maps Ui ∩ Uj → G; encodes the frame co-constitution between ȑR;i and ȑR;j.	§1, Eq. (2)
n(x)	Director field	A smooth map n : M → Sk; assigns a preferred orientation unit vector to each relational position.	§2, Def. 2.1
A	Connection 1-form	A Lie-algebra-valued 1-form on M; defines the parallel transport rule for the director field.	§2, Eq. (3)
Holγ(n)	Holonomy	The group element in G obtained by parallel-transporting the director field around loop γ using the connection A.	§2, Eq. (3)
F	Curvature 2-form	F = dA + A ∧ A; vanishes on contractible open sets by the zero-holonomy condition.	§2, Eq. (4)
Kij	Kernel	The invariant overlap structure (Uij, Gij, μij) of frames ȑR;i and ȑR;j.	§3, Def. 3.1
μij	Kernel measure	A non-negative real-valued function encoding the metric and algebraic weight of the kernel overlap between frames i and j.	§3, Def. 3.1
∘	Kernel composition	The composition rule Eq. (5); generates derived kernels from constituent kernels via integration over intermediate positions.	§3, Eq. (5)
dK	Kernel metric	dK(ȑR;i, ȑR;j) = −log μij; the relational distance between two frames.	§3, Eq. (6)
ρ	Relational density	Mean kernel measure per unit relational volume; governs the approach to the emergent spacetime limit.	§4, Eq. (7)
σ	Orientation spread	RMS variation of the director field n(x) over a correlation length; governs the approach to the emergent spacetime limit.	§4, Eq. (7)
c	Emergent light speed	The constant c = (ρσ2)1/2 in the limit Eq. (7); identifies with the speed of light in the emergent Minkowski metric.	§4, Eq. (7)
Hrel	Relational Hamiltonian	The relational path-length functional Eq. (11); governs the dynamics of any kernel-first system via the eikonal equation.	§5, Eq. (11)
S(x)	Optical phase function	The action in the variational formulation; satisfies the eikonal equation Eq. (12).	§5, Eq. (12)
θ	Reorientation angle	The angle of the director field discontinuity at a reorientation event; large-angle if θ > π/2.	§6, Def. 6.1
∂+(x0)	Downstream region	The set of all causal descendants of the reorientation point x0; the domain of the downstream inversion.	§6, Def. 6.1
T(y)	Local temporal orientation	The preferred time direction at relational position y; inverted downstream of a maximal reorientation by Eq. (13).	§6, Eq. (13)
Δ(n)	Managed Diffusion Invariant (MDI)	The degree of the director field map n : M → Sk; the topological charge of the relational system.	§7, Def. 7.1
ωS	Volume form on Sk	The standard rotation-invariant volume form on the unit sphere; pulled back by n in the definition of the MDI.	§7, Def. 7.1
ȑR;(k)	Order-k meta-frame	The relational frame whose base manifold is the space of order-(k−1) relational frames; the k-th level of the meta-frame tower.	§8, Def. 8.1
Σ(x)	Ascension proxy	The kernel-weighted average of the director field (Eq. 14); the locally readable encoding of the global MDI.	§8, Eq. (14)
μxy(t)	Time-dependent kernel measure	The kernel measure between frames x and y at time t; its infimum over t > t0 defines the memory floor.	§9, Def. 9.1
Smem	Memory entropy	−∫ μxy log μxy dV; a thermodynamic measure of memory degradation (§13, Open Problem 3).	§13

Unless otherwise stated, all manifolds are assumed to be smooth, compact, and orientable. All maps between manifolds are assumed to be smooth. The kernel measure μ_ij is assumed to be bounded above by 1 (full overlap) and below by 0 (no overlap). The structure group G is assumed to be a connected, compact Lie group. The integer k in the director field target sphere S^k is equal to the dimension of the fiber of the associated vector bundle, and is typically equal to the dimension of M itself. All equations hold in the conventions of the manifold’s natural orientation unless a sign is explicitly indicated.

Appendix B: Proofs of Key Results

B.1: Proof of the Zero-Holonomy Condition (Theorem Underlying §2)

Claim: Let (M, G, A) be a principal G-bundle with connection. The holonomy Hol_γ is trivial for every contractible loop γ in M if and only if the curvature 2-form F = dA + A ∧ A vanishes on every contractible open subset of M.

Proof:

(⇒): Suppose Hol_γ = e for every contractible loop γ. Let U ⊂ M be a contractible open set, and let x ∈ U. For any infinitesimal parallelogram γ_ε = {x + εv + εw + …} in U (where v, w ∈ T_xM), the holonomy to leading order in ε is

Hol_γε = exp(ε² F(v, w)) + O(ε³)

If Hol_γε = e for all ε and all v, w, then F(v, w) = 0 for all tangent vectors v, w at all points x ∈ U, hence F = 0 on U. Since U was an arbitrary contractible open set, F = 0 on all contractible open sets.

(⇐): Suppose F = 0 on all contractible open sets. Let γ be a contractible loop based at x₀. Since γ is contractible, there exists a smooth homotopy H : [0,1] × [0,1] → M with H(0, ⋅) = γ, H(1, ⋅) = const. = x₀. The image of H is a contractible 2-complex in M. By the Ambrose-Singer theorem, the curvature form F generates the infinitesimal holonomy; since F = 0 on the image of H, the holonomy Hol_γ = exp(∫∫ H^*F) = exp(0) = e. The argument applies to all contractible loops γ.

B.2: Proof of the MDI Conservation Theorem (Theorem 7.1)

Claim: The Managed Diffusion Invariant Δ(n) = (1/Vol(S^k)) ∫_M n^*ω_S is invariant under smooth homotopies of the director field n : M → S^k.

Proof:

Let {n_t}_t∈[0,1] be a smooth one-parameter family of smooth maps n_t : M → S^k. We must show that d/dt [∫_M n_t^*ω_S] = 0.

By the Leibniz rule for differentiation under the integral and the definition of the Lie derivative,

d/dt ∫_M n_t^*ω_S = ∫_M ℓ_v(n_t^*ω_S)

where v = dn_t/dt is the variational vector field. By Cartan’s magic formula, ℓ_v = i_vd + di_v. The form ω_S is a volume form on S^k, hence closed: dω_S = 0. Therefore d(n_t^*ω_S) = n_t^*(dω_S) = 0, and thus i_vd(n_t^*ω_S) = 0. The remaining term gives

d/dt ∫_M n_t^*ω_S = ∫_M d(i_vn_t^*ω_S) = ∫_∂M i_vn_t^*ω_S = 0

where the last step applies Stokes’ theorem and uses the assumption that M is compact without boundary (∂M = ∅). Therefore d/dt Δ(n_t) = 0 for all t ∈ [0,1], and Δ(n₀) = Δ(n₁).

B.3: Proof of the Memory–MDI Equivalence (Theorem 10.1)

Claim: In a time-evolving relational system (M, {K_ij(t)}, n(t)), the following are equivalent: (i) μ_xy(t) ≥ μ₀ > 0 for some pair (x, y) and all t > t₀ (memory), and (ii) Δ(n(t)) ≠ 0 for some t (non-trivial MDI).

Proof of (i ⇒ ii):

Assume memory: there exists a pair (x, y) and μ₀ > 0 such that μ_xy(t) ≥ μ₀ for all t > t₀. The director field at time t is determined (by the kernel-first architecture) by the history of relational encounters up to time t. Formally:

n(x, t) = 𝓩{μ_xz(s) n₀(z) : s ≤ t, z ∈ M}

where 𝓩 denotes a functional of the kernel-measure history and n₀ is the initial director field. The memory condition ensures that past encounters at (x, y) with t > t₀ contribute a term of weight at least μ₀ to this functional for all subsequent times. This introduces a persistent directional bias in n(x, t) (a correlation between the current director field at x and the historical director at y) that does not average to zero over M. Formally, the total integral

∫_M n^*(t)ω_S ≥ μ₀ ⋅ C(M, n₀) > 0

for a positive constant C that depends on the geometry of M and the initial director field. Hence Δ(n(t)) ≠ 0.

Proof of (ii ⇒ i):

We prove the contrapositive: if the system has no memory (μ_xy(t) → 0 for all pairs (x, y) as t → ∞), then Δ(n(t)) = 0 for all sufficiently large t.

If μ_xy(t) → 0 for all (x, y), then by Eq. (6) the kernel metric d_K(x, y) → ∞ for all pairs. In the limit of infinite kernel distance, the director field at each point x becomes independent of the director field at all other points; the global coherence enforced by the connection A breaks down. The director field map n(t) : M → S^k approaches a configuration in which the values n(x, t) and n(y, t) are statistically independent for all x ≠ y. A map from a connected compact manifold M to S^k with statistically independent values at all points is (almost surely) null-homotopic (it can be continuously deformed to a constant map) and therefore has degree zero: Δ(n(t)) = 0. By Theorem 7.1, once the MDI reaches zero it remains zero under all subsequent smooth deformations (unless a topological transition occurs, which requires a finite-measure singular configuration; a probability-zero event under the dynamics). Therefore Δ(n(t)) = 0 for all t after the memory floor decays to zero.

References

[1]   Geometric Synthesis of the Relational Frame. In the present collection: Part I, §1. Establishes the fiber bundle formulation of relational frames and the co-constitution principle of geometry. Manuscript Series on Kernel-First Architectures, Vol. I.

[2]   Reference-Frame Structure of the Director Field. In the present collection: Part I, §2. Introduces the director field, the zero-holonomy consistency condition, and the role of the director field as geometric mediator between local relational data and global coherence. Manuscript Series on Kernel-First Architectures, Vol. II.

[3]   Kernel-First Architecture. In the present collection: Part I, §3. Defines the kernel as the irreducible operative unit of any relational system and establishes the kernel composition rule from which all downstream structure is derived. Manuscript Series on Kernel-First Architectures, Vol. III.

[4]   Spacetime as Emergent Kernel. In the present collection: Part II, §4. Derives the (3+1)-dimensional Lorentzian spacetime manifold and the light-cone structure of special relativity as the macroscopic limit of kernel composition in the high-density, low-spread regime. Manuscript Series on Kernel-First Architectures, Vol. IV.

[5]   Relational Optical Hamiltonians. In the present collection: Part II, §5. Reformulates the Hamiltonian of a relational system as a relational path-length functional over the director field and establishes the equivalence of relational dynamics with geodesic equations of a curved optical medium. Manuscript Series on Kernel-First Architectures, Vol. V.

[6]   Reorientation and the Downstream Inversion. In the present collection: Part II, §6. Analyzes large-angle frame reorientations and proves the Downstream Inversion Theorem, establishing downstream inversion as a structural re-equilibration mechanism. Manuscript Series on Kernel-First Architectures, Vol. VI.

[7]   Global Invariant of Managed Diffusion. In the present collection: Part III, §7. Defines the Managed Diffusion Invariant (MDI) as the topological charge of a relational system, proves its conservation under smooth deformations, and draws the analogy to the winding number in gauge field theory. Manuscript Series on Kernel-First Architectures, Vol. VII.

[8]   Surrogate Principle of Vertical Ascension. In the present collection: Part III, §8. Introduces vertical ascension and the meta-frame hierarchy, states the Surrogate Principle, and defines the ascension proxy as the locally readable encoding of the global MDI. Manuscript Series on Kernel-First Architectures, Vol. VIII.

[9]   Memory as the Floor of Stability. In the present collection: Part IV, §9–11. Defines memory as kernel-overlap signature persistence, distinguishes memory-as-floor from memory-as-storage, and proves the Memory–MDI Equivalence Theorem (Theorem 10.1): non-trivial MDI is both necessary and sufficient for memory. Manuscript Series on Kernel-First Architectures, Vol. IX.