[SOURCE-FIDELITY NOTICE: This content was extracted from a PDF. Two-dimensional equation layout may be lost, including fraction bars, numerator/denominator grouping, superscripts, roots, matrices, and stacked limits. Treat arithmetic or algebraic discrepancies as inconclusive unless they can be verified against an equation-preserving source.]

Gauge Freedom and the Born Rule Extending the Unitary Refinement Program to Generalized Quantum Measurements Michael Tolbert Version 3 | September 2026 Abstract Recent work established a representation theorem for finite projective events, showing that equal-norm refinement, primitive equivalence, cardinal composition, and refinement invariance uniquely induce the Born measure on homogeneous branch spaces. That construction, however, applies only to pure states and idealized projection-valued measurements. This paper develops a generalized refinement framework for mixed states and positive operator-valued measures (POVMs). By combining purification and Naimark dilation, arbitrary finite-dimensional open-system descriptions are embedded into enlarged Hilbert spaces in which the original cardinality representation theorem applies. The central result is a Representation Invariance Theorem 4.5: physically equivalent purification-dilation realizations induce identical abundance measures despite differing ancillary constructions. The result resolves the representa- tion problem rather than merely importing an already-assumed probability rule into the enlarged space. This invariance reveals a representational structure analogous to gauge freedom. Ancillary refinements, purifications, and measurement dilations generate equivalence classes of mathematical realizations whose variation leaves operational typicality unchanged. The generalized Born measure T(E | ρ) = Tr(ρE) is obtained by first applying the projective refinement theorem as a pre-probabilistic counting construction in an enlarged space and only then proving that the trace rule is its compressed, representation-independent form on the operational pair (ρ, E). A Lipschitz bound expressed through trace distance and operator norm shows that small perturbations of states and effects produce controlled changes in typicality. For a finite family of normalized records with pairwise overlaps bounded byε, Löwdin symmetric orthogonalization, rank-one subspace perturbation geometry, and the operational Lipschitz estimate yield an explicit dimension-dependent error bound that vanishes with ε. Keywords: Born rule; Everett interpretation; POVM; Naimark dilation; purification; quantum information; typicality; gauge freedom; representation invariance; decoherence. Draft status. This manuscript develops the exact finite-dimensional representation theorem and an explicit finite-overlap robustness theorem. The approximate result is dimension-dependent and applies effect-wise to normalized rank-one record operators; a measurement-level theorem also controls canonical normalization and all coarse-event sums for approximately normalized finite positive families; extensions deriving such hypotheses from general higher-rank or continuous record architectures remain open. 1

Gauge Freedom and the Born RuleMichael Tolbert 1 Introduction Closed quantum systems evolve unitarily, while operational quantum theory assigns outcome probabilities through the Born rule. For a density operatorρand a finite POVM{E i }satisfying E i ≥0 and P i E i =I, the probability assigned to outcomeiisTr(ρE i ). The formula is indispensable to quantum information theory, but its mathematical success does not by itself explain why this particular measure should govern branch-relative records in an Everettian theory. The preceding unitary-refinement program separated two questions that are often conflated. The measure problem asks which quantitative functional represents the relative size or typicality of a branch event. The credence problem asks why that measure should govern rational expectation, self-location, or decision. For finite projective decompositions of pure states, the earlier repre- sentation theorem addressed the first question by constructing equal-norm ancillary refinements. Permutation symmetry rendered the descendants primitive-equivalent; cardinal composition fixed relative abundance on the homogeneous representative space; common-refinement independence removed denominator dependence; and regularity extended the rational construction to arbitrary squared norms. That result was intentionally limited. A pure state together with a sharp projection-valued measure is an idealized description. Open systems are represented by mixed density operators, and general measurements are represented by POVMs whose effects need not be mutually orthogonal or idempotent. A realistic extension must therefore answer two distinct representation questions. First, does the derived measure depend on how a mixed state is purified? Second, does it depend on how a POVM is realized as a projective measurement in a larger space? The central proposal of this paper is that these questions should be answered together. A purification and a Naimark dilation convert an operational pair (ρ,{E i }) into a pure state and a projective decomposition in an enlarged Hilbert space. The earlier finite-abundance theorem can then be applied in that enlarged space. This transfer alone is not the main result. The essential task is to prove that the induced typicality is independent of the auxiliary realization and therefore descends to the operational system. We call this requirement representation invariance. Ancillary realizations form equivalence classes analogous to gauge orbits: distinct mathematical representatives encode the same operational state-measurement pair and must yield the same coarse typicality. The analogy is deliberately limited. No spacetime gauge field, connection, or curvature is assumed. The relevant structure is a redundancy of representation under unitary or isometric transformations of auxiliary degrees of freedom. The paper establishes four claims. First, purification and Naimark dilation provide a unified projective realization of every finite-dimensional pair (ρ,{E i }). Second, the cardinality representation theorem induces the generalized Born form in any such realization. Third, the Representation Invariance Theorem proves that the induced value is independent of the chosen minimal realization and extends to nonminimal realizations after restriction to their generated subspaces. Fourth, operational Lipschitz bounds and the finite-overlap theorem show that small perturbations of states, effects, and finite coarse record families produce controlled changes in typicality. 2

Gauge Freedom and the Born RuleMichael Tolbert 2 Operational Data and Realizations 2.1 Mixed states and generalized measurements LetH S be a finite-dimensional complex Hilbert space. A state is a positive operatorρonH S with Tr(ρ) = 1. A finite generalized measurement is a POVM E ={E i } n i=1 with E i ≥ 0 and X i E i = I S . An event associated with a set A of outcome labels is represented by the effect E A

X i∈A E i . Unlike a projective event,E A need not satisfyE 2 A =E A . The objective is to construct a typicality functional T (A| ρ,E) without assuming at the outset that it equals Tr(ρE A ). 2.2 Purification A purification of ρ is a pair (H R ,|φ⟩) with |φ⟩∈H S ⊗H R such that Tr R (|φ⟩⟨φ|) = ρ. If two purifications use minimal reference spaces, they are related by a unitary on the purifying factor. More generally, purifications in larger spaces are related on the support generated by the state through an isometry or partial isometry. This freedom is the first auxiliary redundancy that a physical abundance measure must ignore. 2.3 Naimark realization A Naimark realization ofEis a triple (H K ,W,{P i }) in whichW:H S →H K is an isometry and {P i } is a projective decomposition of the identity on H K satisfying W † P i W = E i for every i. The projectors are exact and mutually orthogonal: P i P j = δ ij P i , X i P i = I K . A concrete canonical realization may be constructed on H S ⊗ C n by W|ψ⟩ = X i p E i |ψ⟩⊗|i⟩ and P i = I S ⊗|i⟩⟨i|. The theorem below does not depend on choosing this canonical model. 3

Gauge Freedom and the Born RuleMichael Tolbert 2.4 Unified purification-dilation realization Choose a purification |φ⟩∈H S ⊗H R and a Naimark isometry W :H S →H K . Define |Ψ⟩ = (W ⊗ I R )|φ⟩∈H K ⊗H R , Q i = P i ⊗ I R . Then {Q i } is a PVM on the unified space and ⟨Ψ|Q i |Ψ⟩ =⟨φ|(E i ⊗ I R )|φ⟩ = Tr(ρE i ). At this stage the right-hand trace is an algebraic identity describing compression of the projective realization. It is not yet used as the definition of typicality. Typicality will be constructed on the projective realization by the finite-abundance theorem and only then shown to equal this compressed expression. 3 Extended Finite-Abundance Construction To keep the present extension self-contained, Table 1 summarizes the assumptions and results inherited from Paper 1. The table is not a new axiomatization and does not replace the proofs established there. Its purpose is to identify the precise structure imported into the purified and dilated projective carrier. The present paper retains the original sequence from primitive equivalence and cardinal abundance to induced event measure, while extending representation independence from common refinements to equivalent purifications and Naimark dilations. Table 1: Logical Structure Inherited from Paper 1 StagePaper 1 premise or resultRole in the present paper Pointer-defined events Decoherence supplies a finite coarse algebra of stable record events, idealized by orthogonal projectors. The purified and dilated description provides the projective carrier on which the inherited construction operates. Equal-norm refinement Rational squared norms admit ancillary refinements containing equal-norm orthogonal descendants. Outcome projectors in the enlarged space are refined into homogeneous representative sets. Primitive equivalence Permutation-related equal-norm descendants are primitive-equivalent units of branch abundance. The same premise applies within each purification-dilation realization. Cardinal composition Abundance over disjoint collections of primitive representatives composes cardinally. This remains the source of finite additivity; additivity is not assumed directly on POVM effects. Cardinality representation Normalized cardinality is the unique finite-abundance representation on a homogeneous representative space. It fixes abundance within each enlarged projective realization. Common- refinement independence Equivalent rational refinements of the same parent event yield identical abundance assignments. Paper 2 generalizes this description independence to equivalent purifications and Naimark dilations. Regular extension Continuity or monotonicity uniquely extends the rational result to arbitrary real squared norms. The same extension applies after the generalized measurement is represented projectively. 4

Gauge Freedom and the Born RuleMichael Tolbert StagePaper 1 premise or resultRole in the present paper Event measureThe constructed abundance induces T(P | ψ) =∥Pψ∥ 2 on finite projective events. Paper 2 shows the induced measure descends to T (E | ρ) = Tr(ρE). Credence boundaryAdditional epistemic principles are required to identify event measure with rational credence. Paper 2 preserves this separation. 3.1 Rational projective geometry in the enlarged space For the unified state |Ψ⟩ and projectors {Q i }, define x i =∥Q i Ψ∥ 2 . Ifx i =m i /Mis rational, the original equal-norm refinement theorem supplies an ancillary isometry that decomposes outcomeiintom i mutually orthogonal descendants of squared norm 1/M. The descendants form an M-element homogeneous representative set Ω M . 3.2 Abundance axioms •Normalization: the empty representative event has abundance zero and Ω M has abundance one. •Permutation invariance: permutations of primitive-equivalent descendants preserve abun- dance. • Disjoint cardinal composition: the abundance of disjoint collections is the sum of their abundances. • Uniform-refinement invariance: replacing each representative by the same number of equivalent descendants preserves every coarse-event ratio. These conditions are imposed on homogeneous representative spaces, not assumed directly as probability axioms on POVM effects. They uniquely give normalized cardinality a M (A) = |A| M . Common refinement removes dependence on the rational denominator, and continuity or monotonicity extends the assignment to arbitrary x i . Regular extension to irrational weights. Letx i =∥Q i Ψ∥ 2 be irrational. The monotonicity invoked here is inherited from cardinal inclusion on finite homogeneous representative spaces: whenever representative eventsAandBsatisfyA⊆ B, normalized cardinality givesT(A)≤ T(B). Choose rational sequencesq k increasing tox i ands k decreasing tox i . For each rational approximant, Paper 1’s equal-norm refinement theorem supplies a finite exact refinement, although the ancillary 5

Gauge Freedom and the Born RuleMichael Tolbert dimension may depend on the approximating denominator. No single fixed finite refinement space is required to realize the irrational value exactly. q k ≤ T (x i )≤ s k . Takingk →∞forcesT(x i ) =x i . Equivalently, continuity gives the same result from any rational sequence converging tox i . Because purification and Naimark dilation preserve the squared norm being approximated, the extension applies unchanged in the enlarged projective carrier. This step imports Paper 1’s regular-extension theorem; it does not assert the existence of an exact finite refinement for an irrational squared norm. Theorem 3.1 (Generalized refinement representation). For every finite-dimensional density operator ρ, finite POVME={E i }, and purification-dilation realization (|Ψ⟩,{Q i }), the homogeneous finite- abundance construction assigns outcome i the value T R (i) =∥Q i Ψ∥ 2 . Consequently, T R (i) = Tr(ρE i ). Proof.The projectors{Q i }form an exact finite projective decomposition of the normalized pure state|Ψ⟩. The earlier cardinality theorem therefore applies on rational squared-norm vectors and uniquely yields normalized descendant cardinalityx i . Common-refinement invariance makes the rational value representation-independent within the refinement class, and regular extension gives T R (i) =∥Q i Ψ∥ 2 for arbitraryx i . Compression ofQ i throughWand reduction overH R gives ∥Q i Ψ∥ 2 = Tr(ρE i ). The theorem transfers the projective construction to generalized measurements, but it does not yet show that a different purification or dilation produces the same abundance construction. That is the purpose of the next section. 3.3 Diagnostic countermodels and the role of the assumptions The assumptions inherited from Paper 1 perform logically distinct tasks. The following finite countermodels are not alternative physical theories. They are diagnostic constructions showing that recognizable pathologies remain when particular structural requirements are removed. Accordingly, the analysis establishes functional separation of the assumptions rather than a complete model- theoretic independence theorem. Proposition 3.2 (Diagnostic countermodels). On finite homogeneous representative spaces, the following assignments demonstrate that normalization alone does not secure the cardinality repre- sentation. 1.If permutation invariance is removed, label-dependent weighted assignments distinguish primitive-equivalent descendants. 2. If disjoint cardinal composition is removed, nonlinear functions of normalized cardinality satisfy normalization and permutation invariance but fail finite additivity. 3.If uniform-refinement invariance is removed, abundance may change under a physically idle uniform splitting of every representative. 6

Gauge Freedom and the Born RuleMichael Tolbert 4.If common-refinement independence is removed, equivalent rational presentations of the same coarse weight may receive denominator-dependent values. 5.If regular extension is removed, agreement on every rational weight does not uniquely determine values at irrational weights. Proof. Let Ω M ={1,...,M}. Permutation invariance. Choose positive numbers w 1 ,...,w M that are not all equal and define a (w) M (A) := P j∈A w j P M j=1 w j . This assignment is normalized and additive on disjoint subsets, but a permutation exchanging representatives with unequal weights changes the value. Thus normalization and additivity do not by themselves express primitive equivalence. Disjoint cardinal composition. Define a (2) M (A) :=  |A| M  2 . The assignment is normalized, permutation invariant, and unchanged when every representative is uniformly replaced byLdescendants, because|A|/M=L|A|/(LM). For two disjoint nonempty events A and B, however, a (2) M (A∪ B) =  |A| +|B| M  2 ̸= a (2) M (A) + a (2) M (B). Hence refinement stability and symmetry alone do not yield finite additivity. Uniform-refinement invariance. Define a size M (A) :=        0,A = ∅, |A|/(M + 1), A̸= ∅, A̸= Ω M , 1,A = Ω M . The assignment is normalized and permutation invariant. Under uniformL-fold refinement, a proper nonempty event changes from|A|/(M+ 1) toL|A|/(LM+ 1). The construction therefore detects a splitting that leaves every coarse ratio unchanged. Common-refinement independence. For a rational presentation x = m/M, define b ε (m,M ) :=          0,m = 0, 1,m = M, m M

• ε m(M − m) M 2 gcd(m,M )− 1 gcd(m,M ) , 0 < m < M, where 0< ε≤1. This assignment preserves the endpoints and remains in [0,1]: for 0< m < M, the correction is nonnegative and at most m(M − m) M 2 ≤ M − m M = 1− m M . 7

Gauge Freedom and the Born RuleMichael Tolbert For a reduced presentation,gcd(m,M) = 1 and henceb ε (m,M) =m/M. Under a nontrivial common scaling (m,M )7→ (Lm,LM ), however, b ε (Lm,LM ) = m M

• ε m(M − m) M 2 L− 1 L , which generally differs from the reduced value. Thus the same rational coarse weight can acquire a denominator-dependent value unless common refinements are identified. Regular extension. Define f : [0, 1]→ [0, 1] by f (x) := ( x,x∈ Q, 1− x, x /∈ Q. The function agrees with normalized cardinality on every rational weight produced by finite exact refinement, but it is not fixed at irrational weights and is neither continuous nor monotone. Therefore the rational construction alone does not uniquely determine its extension. Proposition 3.2 clarifies the division of labor. Primitive equivalence removes label sensitivity; cardinal composition supplies finite additivity; uniform-refinement and common-refinement invariance remove dependence on artificial descendant multiplicity and rational presentation; and regularity fixes the irrational completion. The proposition does not show that these conditions follow from unitary dynamics. It shows why each condition addresses a distinct ambiguity in the abundance construction. 4 Representation Invariance 4.1 Continuity of the Paper 1 Framework Paper 1 established a conditional sequence with four distinct stages. First, symmetry identifies equal- norm descendants as primitive-equivalent representatives. Second, disjoint cardinal composition assigns normalized abundance to collections of those representatives. Third, common-refinement independence and regular extension induce a measure on the coarse projective event algebra. Fourth, separate epistemic principles may identify that measure with branch-relative credence. The present paper preserves this order. Purification and Naimark dilation do not replace the homogeneous representative construction. They supply a larger pure projective carrier in which the Paper 1 construction can operate for mixed states and generalized measurements. Primitive descendants remain the units of abundance; abundance induces an operational event measure; and no claim about single-case credence follows without an additional epistemic bridge. Accordingly, the term typicality in this paper denotes the induced event measure unless credence is explicitly discussed. 4.2 Operational equivalence and realization classes Definition 4.1 (Operational equivalence). A purification-dilation realizationRof an operational pair (ρ,E) consists of a purifying space H A R , a purification |φ R ⟩∈H S ⊗H A R satisfying Tr A R (|φ R ⟩⟨φ R |) = ρ, 8

Gauge Freedom and the Born RuleMichael Tolbert a Naimark dilation (H K ,W R ,{P R i }) satisfying W † R P R i W R = E i , the unified state |Ψ R ⟩ = (W R ⊗ I A R )|φ R ⟩, and the outcome projectors Q R i = P R i ⊗ I A R . Two realizationsRandR ′ are operationally equivalent, writtenR ∼ R ′ , when both compress to the same density operator ρ and the same outcome-labeled POVM E ={E i }. Outcome labels are held fixed; a relabeling represents the same operational measurement only when the relabeling is explicitly carried through the event identification. Definition 4.2 (Generated realization subspace). Define the reduced state on the purifying subsystem by τ R := Tr S (|φ R ⟩⟨φ R |). For a realization R, define the system-generated Naimark subspace K gen R := span{P R i W R |χ⟩ :|χ⟩∈H S , i = 1,...,n}, and define the corresponding unified generated subspace G R :=K gen R ⊗ supp(τ R ). Heresupp(τ R ) is the supported purifying subspace associated withρand has dimensionrank(ρ). In finite dimensions the closure symbol is unnecessary; in an infinite-dimensional extension the spans would be norm-closed. The vectors |Ψ R ⟩ and Q R A |Ψ R ⟩ lie in G R for every coarse event A, where Q R A := X i∈A Q R i . The abundance construction is evaluated on this generated subspace. State-inaccessible orthogonal sectors are not part of the minimal operational realization. The generated subspace is not itself the object of cardinal evaluation. It is the carrier space in which the equal-norm descendants and homogeneous representative sets of Section 3 are constructed. Cardinal abundance remains assigned to disjoint collections of primitive-equivalent descendants, exactly as in Paper 1. Criterion 4.3 (Operational Representation Criterion). An abundance assignment intended to characterize an operational pair (ρ,{E i }) should be constant under equivalences of auxiliary repre- sentation that preserveρ, preserve the labeled effectsE i , and intertwine the corresponding generated outcome subspaces. The Operational Representation Criterion is the generalized analogue of the description-independence requirement expressed by Common Refinement Independence in Paper 1. It is not an additional premise used to derive the finite-dimensional equality in Theorem 4.5. Rather, it states the operational admissibility condition that a candidate abundance assignment should satisfy. Lemma 4.4 and Theorem 4.5 show that the abundance assignment inherited from Paper 1 satisfies the criterion for exact finite-dimensional purification-dilation realizations. The criterion remains programmatically useful in approximate, infinite-dimensional, or operator-algebraic settings where exact unitary equivalence may no longer be available in the same form. 9

Gauge Freedom and the Born RuleMichael Tolbert Lemma 4.4 (Constructive equivalence on generated subspaces). LetRandR ′ be finite-dimensional purification-dilation realizations of the same outcome-labeled operational pair (ρ,{E i }). After restriction to G R and G R ′ , there exists a unitary V :G R →G R ′ such that V|Ψ R ⟩ =|Ψ R ′ ⟩, V Q R i = Q R ′ i V for everyi. On unreduced auxiliary spaces containing inaccessible orthogonal sectors, the same correspondence extends by zero to a partial isometry whose initial and final spaces areG R andG R ′ . Proof. First construct the Naimark intertwiner. On the algebraic generated span define U K,0

X i P R i W R |χ i ⟩ ! := X i P R ′ i W R ′ |χ i ⟩. For all outcome labels i,j and all |χ⟩,|η⟩∈H S , D P R i W R χ,P R j W R η E =⟨χ|W † R P R i P R j W R |η⟩ = δ ij ⟨χ|W † R P R i W R |η⟩ = δ ij ⟨χ|E i |η⟩. The same inner product is obtained inR ′ because both realizations compress to the same labeled effects E i . Consequently,

X i P R i W R |χ i ⟩

2

X i P R ′ i W R ′ |χ i ⟩

2 . ThusU K,0 is well defined: every representation of the zero vector maps to zero. It preserves inner products and is therefore an isometry. Its range isK gen R ′ by definition, so in finite dimensions it extends uniquely to a unitary U K :K gen R →K gen R ′ , satisfying U K W R = W R ′ , U K P R i = P R ′ i U K on the generated subspaces. Moreover, dimK gen R = dimK gen R ′

X i rank(E i ), where a zero effect contributes zero dimension. Next construct the purification intertwiner. Letr=rank(ρ). Since|φ R ⟩and|φ R ′ ⟩purify the same density operator, choose Schmidt decompositions over a common eigenbasis {|a⟩ S } r a=1 of supp(ρ): |φ R ⟩ = r X a=1 p λ a |a⟩ S |a⟩ A R , |φ R ′ ⟩ = r X a=1 p λ a |a⟩ S |a⟩ A R ′ . 10

Gauge Freedom and the Born RuleMichael Tolbert The map U A |a⟩ A R =|a⟩ A R ′ is unitary from supp(τ R ) to supp(τ R ′ ) and satisfies (I S ⊗ U A )|φ R ⟩ =|φ R ′ ⟩. Set V := U K ⊗ U A . ThenVis unitary fromG R toG R ′ , maps|Ψ R ⟩to|Ψ R ′ ⟩, and intertwines every labeled outcome projector. Although the unreduced ambient spaces may have different dimensions, the generated initial and final spaces have equal dimensions. ExtendingVby zero onG ⊥ R produces a partial isometry with initial projection ontoG R and final projection ontoG R ′ . No unitary equivalence of the entire unreduced ambient spaces is asserted. Theorem 4.5 (Representation Invariance). LetRandR ′ be finite-dimensional purification-dilation realizations of the same operational pair (ρ,{E i }). Apply the homogeneous finite-abundance con- struction of Section 3 separately in each generated subspace. Then, for every outcomeiand every coarse event A⊆{1,...,n}, T R (i) = T R ′ (i) and T R (A) = T R ′ (A). Proof. For a realization R, the projective abundance assignment is T R (A) =

Q R A Ψ R

2 , Q R A := X i∈A Q R i . By Lemma 4.4, V Q R A = Q R ′ A V, V|Ψ R ⟩ =|Ψ R ′ ⟩. Therefore,

Q R ′ A Ψ R ′

2

Q R ′ A V Ψ R

2

V Q R A Ψ R

2

Q R A Ψ R

2 . The equal-norm refinements in the two realizations thus represent the same coarse norm-squared value, and the cardinality representation with regular extension assigns the same abundance. The outcome statement is the special case A ={i}. Corollary 4.6 (Operational descent). For each finite-dimensional operational pair (ρ,E), there exists a well-defined induced abundance functional T (ρ,E) on coarse events, defined by T (ρ,E) (A) := T R (A) for any realizationRof (ρ,E). Its uniqueness within the present framework follows from the cardinality representation and regular-extension results inherited through Section 3. Explicitly, T (ρ,E) (A) =

Q R A Ψ R

2 = Tr(ρE A ), E A := X i∈A E i . The result is exact, finite-dimensional, and outcome-label preserving. It does not identify realizations of different POVMs or remove the substantive finite-abundance assumptions inherited from the projective theory. Approximate departures from orthogonality are treated separately in Section 5. 11

Gauge Freedom and the Born RuleMichael Tolbert 4.3 Nonminimal realizations A nonminimal dilation may contain ancillary sectors never reached from the embedded system state. Restricting to the cyclic subspace generated by{P i Wψ:ψ ∈ H S }isolates the minimal Naimark core; the corresponding unified construction is then restricted to the generated purification-dilation subspace used in Lemma 4.4. Orthogonal unused sectors are representational surplus. Because the partial isometry is unitary on the generated initial and final subspaces, the norm-squared and cardinal-abundance identities remain unchanged. 4.4 Preparation independence The purification proof also resolves ensemble ambiguity. A mixed state may admit many convex decompositions ρ = X a p a |ψ a ⟩⟨ψ a |. The construction does not count the vectors in a chosen ensemble. It acts on a purification whose reduced state isρ, and Representation Invariance removes dependence on the choice of purification. Thus preparations represented by the same density operator receive the same typicality for every POVM effect. No preferred ensemble decomposition is introduced. 4.5 Conservative gauge interpretation At the present finite-dimensional stage, gauge freedom denotes an algebraic and representational redundancy: distinct auxiliary realizations correspond to the same operational pair and induce the same event measure. The operational pair (ρ,E) is the invariant object. Purifications, Naimark spaces, ancillary bases, and uniform refinements are representatives. Intertwining unitaries and support-preserving isometries map one representative to another. Typicality is a well-defined function on the resulting equivalence class because it is constant along that class. 5 Approximate Representation Invariance 5.1 Why an approximate theorem is needed Exact Naimark dilation is available for exact POVMs, so generalized measurements do not themselves require nonorthogonal outcome projectors in the enlarged model. Approximation enters elsewhere: experimentally reconstructed states and effects are imperfect; two implementations may realize nearby rather than identical operational pairs; and decohered macroscopic records may be only approximately orthogonal before an idealized coarse event algebra is selected. A robustness theorem should separate these sources of error rather than treating all “fuzziness” as one parameter. 5.2 Stability under perturbations of state and effect 5.2.1 Typicality of an arbitrary effect For every effect 0≤ X ≤ I S , let E X :={X, I S − X} 12

Gauge Freedom and the Born RuleMichael Tolbert be the associated binary POVM and define T ρ (X) := T (ρ,E X ) ({1}). By Corollary 4.6,T ρ (X) =Tr(ρX). The value depends only on the pair (ρ,X), not on whetherXis also presented as one outcome of a larger POVM. In particular, each normalized rank-one operator R i =|r i ⟩⟨r i |and each orthogonalized projector Π i =|e i ⟩⟨e i |is a valid effect, even when the family {R i }is not collectively normalized. All later uses ofT(R i | ρ) andT(Π i | ρ) may equivalently be read as T ρ (R i ) and T ρ (Π i ). Let (ρ,E) and (σ,F) be two operational descriptions with effectsE i andF i . Once the representation theorem has identifiedT(E i | ρ) with the induced projective abundance, the difference between the two assignments satisfies the elementary decomposition |T (E i | ρ)− T (F i | σ)|≤|Tr[(ρ− σ)E i ]| +|Tr[σ(E i − F i )]|. Using trace-norm and operator-norm duality, together with ∥E i ∥ ∞ ≤ 1 and ∥σ∥ 1 = 1, gives |T (E i | ρ)− T (F i | σ)|≤∥ρ− σ∥ 1 +∥E i − F i ∥ ∞ . With the standard trace distance D(ρ,σ) = 1 2 ∥ρ− σ∥ 1 , one has∥ρ− σ∥ 1 = 2D(ρ,σ). Trace distance is the standard operational metric for quantum-state distinguishability. The perturbation estimate therefore takes the explicitly quantum-information form |T (E i | ρ)− T (F i | σ)|≤ 2D(ρ,σ) +∥E i − F i ∥ ∞ . This Lipschitz bound shows that representation invariance does not shatter under small imperfections: nearby states and measurement effects produce small, linearly controlled variations in the derived typicality. It is a robustness result for nearby operational models, not yet a theorem converting a bound on pairwise pointer-state overlaps into a universal constant. Proposition 5.1 (Operational perturbation bound). For density operatorsρandσand effects 0≤ E,F ≤ Ion the same finite-dimensional Hilbert space, the induced refinement typicalities satisfy |T (E | ρ)− T (F | σ)|≤∥ρ− σ∥ 1 +∥E− F∥ ∞ = 2D(ρ,σ) +∥E− F∥ ∞ . Hence exact representation invariance extends Lipschitz-continuously to approximately equivalent operational pairs. 5.3 Robustness of Gauge Descent under Approximate Decoherence LetH S be finite-dimensional and let{|r i ⟩} n i=1 be normalized, linearly independent record vectors. Define their Gram matrix by G ij =⟨r i | r j ⟩ and assume |G ij |≤ ε for i̸= j. By the row-sum bound for Hermitian matrices, ∥G− I∥ ∞ ≤ (n− 1)ε =: δ. Assumeδ <1. ThenGis positive definite and its spectrum lies in [1− δ,1 +δ]. The condition is explicitly dimension-dependent: pairwise overlap alone is not sufficient unless (n−1)εremains below 13

Gauge Freedom and the Born RuleMichael Tolbert one. Herendenotes the number of record vectors retained in the finite coarse event family, not the dimension of the environment or a microscopic census of branches. Environmental redundancy therefore does not by itself forcento increase. For a sequence of models with retained record numbern k and overlap boundε k , the present estimate remains informative whenever (n k − 1)ε k → 0. Whether a particular decoherence model satisfies that condition depends on its system-environment dynamics and is not assumed here. 5.3.1 Step 1: Löwdin symmetric orthogonalization Let R : C n →H S be the synthesis operator Rc = X i c i |r i ⟩, so that R † R = G. Define the Löwdin orthogonalization operator U := RG −1/2 , |e i ⟩ := U|i⟩ = X j (G −1/2 ) ji |r j ⟩. Because U † U = G −1/2 R † RG −1/2 = I, the vectors{|e i ⟩}are orthonormal and span the same subspace as the records. The relevant displacement is controlled directly through the polar factor: ∥R− U∥ ∞

U (G 1/2 − I)

∞

G 1/2 − I

∞ . Since the spectrum of G is contained in [1− δ, 1 + δ], functional calculus gives

G 1/2 − I

∞ ≤ 1− √ 1− δ = δ 1 + √ 1− δ =: b(δ). Consequently, for every outcome label i, ∥|r i ⟩−|e i ⟩∥≤ b(δ). This bound corrects the formal route suggested by expandingG −1/2 : the synthesis operatorR contributes to the vector displacement, whereas the polar-factor identity controls the complete transformation without dropping that contribution. 5.3.2 Step 2: Rank-one subspace perturbation Define the normalized rank-one record operatorsR i :=|r i ⟩⟨r i |and the orthogonalized projectors Π i :=|e i ⟩⟨e i |. For unit vectors, the operator-norm distance between the associated rank-one projections equals the sine of their canonical angle: ∥R i − Π i ∥ ∞

q 1−|⟨r i | e i ⟩| 2 ≤∥|r i ⟩−|e i ⟩∥≤ b(δ). This is the rank-one instance of the subspace-rotation geometry associated with Davis–Kahan bounds. 1 No separate spectral-gap denominator is needed here because both subspaces are one- dimensional and the projector-distance identity is exact. 1 The rank-one projector estimate used here is the one-dimensional specialization of Davis–Kahan-style subspace perturbation theory. For higher-rank or spectrally clustered record sectors, the full Davis–Kahan framework supplies bounds on rotations of invariant subspaces in terms of perturbation size and spectral separation. 14

Gauge Freedom and the Born RuleMichael Tolbert 5.3.3 Step 3: Lipschitz integration Fix a density operator ρ. Proposition 5.1, with the state held fixed, gives |T (R i | ρ)− T (Π i | ρ)|≤∥R i − Π i ∥ ∞ . Combining Steps 1 and 2 therefore yields an explicit finite-overlap estimate. The result is effect-wise: a nonorthogonal family{R i }need not sum to the identity and is not automatically a POVM. Each R i is nevertheless a valid effect, and the estimate applies whenever the realistic and idealized record effects are embedded in operational measurements being compared. Theorem 5.2 (Finite-Overlap Approximate Invariance). Let{|r i ⟩} n i=1 be normalized, linearly independent record vectors with |⟨r i | r j ⟩|≤ ε for i̸= j, and set δ := (n− 1)ε < 1. Let {|e i ⟩} be their Löwdin symmetric orthogonalization, R i :=|r i ⟩⟨r i |,Π i :=|e i ⟩⟨e i |. Then, for every density operator ρ and every i, |T (R i | ρ)− T (Π i | ρ)|≤ b(δ), where b(δ) := 1− √ 1− δ = δ 1 + √ 1− δ . In particular, |T (R i | ρ)− Tr(ρΠ i )|≤ b((n− 1)ε), and the bound tends to zero as ε→ 0 for fixed n. Proof.The Gram-matrix estimate gives∥G− I∥ ∞ ≤ δ <1, soGis positive definite. Löwdin orthogonalization defines the isometry U = RG −1/2 . The polar decomposition R = UG 1/2 implies ∥R− U∥ ∞

G 1/2 − I

∞ . Spectral localization of G gives

G 1/2 − I

∞ ≤ 1− √ 1− δ = b(δ), hence∥|r i ⟩−|e i ⟩∥ ≤ b(δ). The exact rank-one projector formula then gives∥R i − Π i ∥ ∞ ≤ b(δ). Proposition 5.1 with identical states yields |T (R i | ρ)− T (Π i | ρ)|≤∥R i − Π i ∥ ∞ . Finally, Π i is an exact projective event, so the representation theorem givesT(Π i | ρ) =Tr(ρΠ i ). Combining the inequalities proves the claim. 15

Gauge Freedom and the Born RuleMichael Tolbert 5.4 Normalization and coarse-event robustness for finite positive families The effect-wise estimate can be lifted to an entire finite measurement whenever a positive family is already close to normalized. Let {F i } n i=1 be positive operators and write S := n X i=1 F i . If S is invertible, the canonical normalization E i := S −1/2 F i S −1/2 forms a POVM. This normalization is the operator-valued analogue of passing from a frame to its canonical Parseval normalization; the relation between frames and POVMs is discussed in [1]. Theorem 5.3 (Finite-POVM normalization and coarse-event robustness). Let{F i } n i=1 be positive operators on a finite-dimensional Hilbert space, let S = P i F i , and suppose ∥S− I∥ ∞ ≤ η < 1. Define E i := S −1/2 F i S −1/2 . Then {E i } n i=1 is a POVM. For every coarse event A⊆{1,...,n}, with F A := X i∈A F i , E A := X i∈A E i , one has ∥E A − F A ∥ ∞ ≤ κ(η), where κ(η) :=

1− p 1− η
1 + p 1 + η
. Consequently, for density operators ρ and σ, |T (E A | ρ)− Tr(σF A )|≤ 2D(ρ,σ) + κ(η). HereT(E A | ρ) is well defined becauseE A is an effect. The raw operatorF A need not be an effect, so its comparison term is written explicitly as the trace functionalTr(σF A ) rather than asT(F A | σ). In particular, for a fixed state ρ, the bound is uniform over all coarse events: sup A |T (E A | ρ)− Tr(ρF A )|≤ κ(η). Moreover, κ(η) = η + O(η 2 ) as η → 0. Proof. The hypothesis implies (1− η)I≤ S ≤ (1 + η)I, so S is positive and invertible. Positivity of each E i is preserved under congruence by S −1/2 , and X i E i = S −1/2 SS −1/2 = I. 16

Gauge Freedom and the Born RuleMichael Tolbert Thus {E i } is a POVM. Set B := S 1/2 . For every A, E A = S −1/2 F A S −1/2 , F A = BE A B. Because 0≤ F A ≤ S, congruence byS −1/2 gives 0≤ E A ≤ I, and hence∥E A ∥ ∞ ≤1. The spectral inclusion spec(B)⊆ [ p 1− η, p 1 + η] implies ∥B− I∥ ∞ ≤ 1− p 1− η, ∥B∥ ∞ ≤ p 1 + η. Using F A = BE A B, ∥E A − F A ∥ ∞ =∥E A − BE A B∥ ∞ ≤∥(I− B)E A ∥ ∞ +∥BE A (I− B)∥ ∞ ≤∥B− I∥ ∞ ∥E A ∥ ∞

1 +∥B∥ ∞
≤

1− p 1− η
1 + p 1 + η
. BecauseE A is an effect, Corollary 4.6 givesT(E A | ρ) =Tr(ρE A ). AlthoughF A need not be an effect, its trace pairing with a state is well defined. Therefore |T (E A | ρ)− Tr(σF A )| =|Tr(ρE A )− Tr(σF A )| ≤|Tr[(ρ− σ)E A ]| +|Tr[σ(E A − F A )]| ≤∥ρ− σ∥ 1 ∥E A ∥ ∞ +∥σ∥ 1 ∥E A − F A ∥ ∞ ≤ 2D(ρ,σ) +∥E A − F A ∥ ∞ , which proves the stated bounds without treating the raw operatorF A as an operational effect. The asymptotic expansion follows from the Taylor expansions of the square roots at η = 0. Corollary 5.4 (Coarse-event convergence). Let{F (k) i } n i=1 be finite positive families with frame operators S k

P i F (k) i and ∥S k − I∥ ∞ → 0. Let E (k) i := S −1/2 k F (k) i S −1/2 k . Then, for every fixed density operator ρ, sup A⊆{1,...,n}    T (E (k) A | ρ)− Tr  ρF (k) A     −→ 0. The theorem controls normalization and every coarse-event sum simultaneously, without an error factor proportional to the number of outcomes inA. It compares the normalized operational typicality with the raw positive-family trace functional and does not assert that every unnormalized F A is itself an effect. It does not by itself convert pairwise record-overlap bounds into a bound on ∥S− I∥ ∞ for arbitrary higher-rank architectures; that remains a separate model-dependent step. 17

Gauge Freedom and the Born RuleMichael Tolbert 5.5 Interpretation and limits Theorem 5.2 establishes a proved, dimension-dependent robustness statement. It identifies a complete continuity chain: microscopic pairwise overlap controls the Gram perturbation; Löwdin orthogonalization controls record-vector displacement; rank-one subspace geometry controls effect displacement; and Proposition 5.1 controls the induced measure. For fixedn, the bound is asymptotically b((n− 1)ε) = (n− 1)ε 2

• O(ε 2 ). The theorem does not claim a dimension-free estimate for arbitrary record families, nor does it imply that the raw nonorthogonal rank-one family{R i }is itself a normalized POVM. A full measurement-level theorem must additionally control normalization, coarse graining, higher-rank record sectors, and possibly changes in state preparation. Those extensions now begin from an explicit finite-overlap theorem rather than an unspecified target. Theorem 5.3 resolves the collective normalization step for finite positive families whose frame operator is close to the identity and supplies a uniform bound for all coarse-event sums. The remaining open problem is to derive such near-normalization bounds directly from concrete nonorthogonal record models, especially for higher-rank sectors, varying outcome number, and continuous outcome structures. 6 Structural Consequences 6.1 A hierarchy of representation independence The generalized program now contains three levels of redundancy. Uniform and common refinement remove dependence on the number of equal-norm representatives. Purification invariance removes dependence on an ensemble or reference-system realization ofρ. Naimark invariance removes dependence on the projective realization of a POVM. Together they support a single principle: a physical branch measure must be constant under transformations that preserve the operational event represented. Across these levels, the conceptual hierarchy remains unchanged: primitive descendants are the representative units; cardinal composition supplies abundance; representation independence and regular extension induce measure on coarse operational events; and any identification of that measure with credence requires a separate epistemic argument. 6.2 Why this is not merely the Born rule by dilation Naimark dilation guarantees that POVM statistics can be represented projectively once a probability rule is already available. By itself, dilation therefore does not derive the Born rule. The present argument reverses the logical order. It first invokes the projective finite-abundance construction as a pre-probabilistic counting mechanism on homogeneous representatives in the enlarged space, where additivity enters through cardinal composition of disjoint descendants. It then proves that the resulting abundance is invariant under changes of purification and dilation. Only after that construction is complete does the trace formula appear as the compressed representation of the derived measure. The substantive assumptions remain visible: primitive equivalence, disjoint cardinal composition, refinement invariance, and regular extension. 18

Gauge Freedom and the Born RuleMichael Tolbert 6.3 Relation to Gleason-type representation theorems and noncontextuality The trace form obtained in Corollary 4.6 places the present construction in the same mathematical neighborhood as Gleason-type representation theorems. The logical route, however, is different. It is therefore important to distinguish three questions:

1. What assumptions characterize the trace functional on projectors or effects?

What structure supplies the additive event measure to which such a characterization applies? 3.Why should the resulting measure be independent of the auxiliary realization used to represent a state and measurement? Gleason’s theorem addresses the first question for probability measures on projections. In its standard form, one begins with a nonnegative, normalized assignment on the projection lattice that is additive over orthogonal decompositions. For complex Hilbert spaces of dimension at least three, the assignment must have the form μ ρ (P ) = Tr(ρP ) for a density operatorρ[2]. The theorem is a representation theorem for an already additive measure. It does not derive the additivity assumption from the unitary dynamics of a measurement interaction. Busch’s generalized Gleason theorem extends this characterization from projections to the full effect space E (H S ) :={X : 0≤ X ≤ I S }. A generalized probability measure on effects is assumed to satisfy positivity, normalization at the identity, and additivity whenever a finite or countable sum of effects remains an effect. Under those conditions, the assignment is represented by μ ρ (X) = Tr(ρX) for a density operatorρ[3]. Because the domain contains all effects rather than projections alone, this formulation also covers the qubit case that lies outside the standard dimension-three restriction of Gleason’s original theorem. The present construction does not compete with these theorems as a different characterization of the same effect functional. Once a normalized additive functional has been defined on all effects, the Gleason–Busch results already explain why the trace representation is unique. The intended contribution here concerns the prior origin and subsequent descent of the measure. Within the inherited refinement framework, finite additivity is not initially postulated on arbitrary projectors or effects. It first enters as disjoint cardinal composition on a finite homogeneous representative space. If Ω M is an M-element set of primitive-equivalent descendants, then a M (A) = |A| M is fixed by normalization, permutation invariance, and cardinal composition. Uniform-refinement invariance and common-refinement independence make the rational assignment independent of the chosen representative denominator. Regular extension then produces the projective event measure T (P | ψ) =∥Pψ∥ 2 . 19

Gauge Freedom and the Born RuleMichael Tolbert Purification and Naimark dilation subsequently extend the domain of the construction, while Theorem 4.5 proves that its value does not depend on the selected auxiliary realization. The relationship among the three approaches may therefore be summarized as follows: Gleason: additive measure on projections ⇓ μ(P ) = Tr(ρP ) Busch: additive measure on effects ⇓ μ(X) = Tr(ρX) whereas the present framework has the conditional structure primitive equivalence + cardinal composition + refinement independence + regular extension =⇒ projective event measure T (P | ψ) =∥Pψ∥ 2 followed by purification and dilation + representation invariance =⇒ T ρ (X) = Tr(ρX). This comparison also clarifies the role of noncontextuality. In an operational framework, two prepa- rations are operationally equivalent when they produce the same statistics for every measurement, and two measurement events are operationally equivalent when they produce the same statistics for every preparation. Generalized noncontextuality requires operationally equivalent procedures to receive identical representations in an ontological model [4]. The present paper does not construct an ontological model of that kind. Its representation invariance is instead a condition on an induced event measure within quantum theory. Equivalent purifications and Naimark dilations represent the same density operator and the same labeled POVM, and the theorem proves that the abundance construction is constant across those auxiliary realizations. This is formally analogous to preparation and measurement noncontextuality because operationally equivalent descriptions receive the same numerical assignment, but the claims should not be identified. More precisely: • Preparation independence. Dependence on a particular ensemble or purification is removed because the induced functional depends only on the reduced density operator ρ. •Measurement-realization independence. Dependence on a particular Naimark dilation is removed because the induced functional depends only on the labeled effects {E i }. •Coarse-event consistency. For a fixed POVM, disjoint unions of outcome labels inherit additivity from the cardinal composition of their projective refinements. •No ontological noncontextuality theorem. The construction does not assign response functions to underlying ontic states and does not prove the existence of a preparation- or measurement-noncontextual ontological model. 20

Gauge Freedom and the Born RuleMichael Tolbert The comparison identifies both the strength and the limitation of the present result. The strength is that the trace functional is reached through a specified abundance construction and is then shown constructively to descend through purification and dilation freedom. The limitation is that primitive equivalence, cardinal composition, refinement independence, and regular extension remain substantive assumptions of that construction. Accordingly, the present result should not be described as deriving the Born rule from unitary dynamics alone. The appropriate claim is narrower and more precise: conditional on the finite-abundance frame- work, the induced projective measure extends uniquely and representation-independently to finite- dimensional mixed states and finite POVMs. Gleason- and Busch-type theorems provide an independent characterization of the same terminal trace functional, while Theorem 4.5 addresses the distinct question of whether the refinement-based construction is well defined across auxiliary quantum representations. 6.4 Quantum-information relevance The formalism applies to finite generalized measurements, including quantum state and detector tomography, minimum-error and unambiguous state discrimination, noisy measurement models, and ancilla-assisted generalized measurement protocols. These applications share the same structural feature: operational outcomes are represented by POVM effects, while implementations may employ different ancillary systems, couplings, purifications, or projective readouts. The Representation Invariance Theorem ensures that the induced event measure depends on the operational pair (ρ,E), not on the selected realization. In tomography, this provides a realization-independent basis for comparing equivalent reconstructions or detector models once they define the same state and POVM. In state discrimination, it applies directly to optimized nonprojective measurements whose effects cannot generally be represented as orthogonal projectors on the system alone. For noisy detectors and generalized protocols, Proposition 5.1 supplies an operational stability estimate in trace and operator norms, while Theorem 5.2 shows how finite record overlap propagates into a controlled effect-wise deviation. These results do not replace statistical error analysis or full measurement-wise normalization, but they isolate the structural measure that those practical analyses estimate. 6.5 Qubit test case: the trine POVM Consider a qubitH S =C 2 prepared in|ψ⟩=|0⟩. Define three coplanar unit vectors separated by 120 degrees: |φ 1 ⟩ =|0⟩, |φ 2 ⟩ =− 1 2 |0⟩ + √ 3 2 |1⟩, |φ 3 ⟩ =− 1 2 |0⟩− √ 3 2 |1⟩. The effects E i

2 3 |φ i ⟩⟨φ i |, i = 1, 2, 3, are positive and satisfy P i E i = I S . For ρ =|0⟩⟨0|, the operational values are Tr(ρE 1 ) = 2 3 ,Tr(ρE 2 ) = 1 6 ,Tr(ρE 3 ) = 1 6 . 21

Gauge Freedom and the Born RuleMichael Tolbert Canonical realizationR. LetH A =C 3 with orthonormal basis{|1⟩ A ,|2⟩ A ,|3⟩ A }. Define the canonical Naimark isometry and outcome projectors by W|χ⟩ = X i p E i |χ⟩⊗|i⟩ A , Q R i = I S ⊗|i⟩ A ⟨i| A . Since √ E i

p 2/3|φ i ⟩⟨φ i |, the dilated state for |ψ⟩ =|0⟩ is |Ψ R ⟩ = r 2 3 |φ 1 ⟩|1⟩ A − 1 √ 6 |φ 2 ⟩|2⟩ A − 1 √ 6 |φ 3 ⟩|3⟩ A . The three projective branch norms are therefore

Q R 1 Ψ R

2

2 3 ,

Q R 2 Ψ R

2

1 6 ,

Q R 3 Ψ R

2

1 6 . These rational values admit a six-element homogeneous refinement: outcome 1 contains four equal- norm representatives, while outcomes 2 and 3 contain one each. Normalized cardinality gives 4/6, 1/6, and 1/6 before the trace expression is used as a probability rule. Minimal generated realization. The canonical realization acts onH S ⊗ C 3 , which is six dimensional. Because each trine effect has rank one, however, its generated Naimark subspace is the three-dimensional space K gen c = span{|φ i ⟩⊗|i⟩ A : i = 1, 2, 3}. A genuinely different minimal realization acts directly on K m = C 3 . Define W m |χ⟩ := 3 X i=1 r 2 3 ⟨φ i | χ⟩|i⟩, P m i :=|i⟩⟨i|. Then W † m P m i W m = E i . Define U K on the canonical generated subspace by U K (|φ i ⟩⊗|i⟩ A ) :=|i⟩. The three domain vectors are orthonormal because their ancillary labels are orthogonal. Therefore U K is unitary from K gen c onto K m and satisfies U K W c = W m , U K P c i = P m i U K onK gen c . Thus the six-dimensional canonical dilation and the three-dimensional minimal dilation have different ambient dimensions but identical generated operational geometry. Their branch norms, descendant cardinalities, and induced typicalities are identical. This comparison illustrates the generated-subspace equivalence proved in Lemma 4.4 and is not merely a simultaneous conjugation of a state and its projectors within one fixed ancillary space. 7 Principal Objections and Replies 7.1 Does purification merely hide a preferred ensemble? No ensemble decomposition is selected. The construction is performed on a purification, and invariance under changes of purification makes the induced event measure a function of the reduced density operator alone. The primitive units remain equal-norm descendants in a homogeneous refinement of the enlarged projective realization; the purification does not convert ensemble elements into objects of cardinal counting. 22

Gauge Freedom and the Born RuleMichael Tolbert 7.2 Does Naimark dilation assume the Born rule? Naimark dilation is used as an operator-theoretic representation of effects by projectors. The measure is not defined by the trace at the outset. It is constructed on the projective realization by the earlier abundance theorem and only afterward compressed to the system. The construction is pre-probabilistic in a specific and limited sense. It first defines a structural abundance over disjoint collections of primitive-equivalent representatives, analogous to determining the relative multiplicity of homogeneous microstate classes before assigning those multiplicities an epistemic or stochastic role. The analogy does not identify descendants with classical microstates or derive cardinal composition from dynamics; it clarifies that the measure construction logically precedes any interpretation in terms of chance, expectation, or credence. 7.3 Is dilation invariance trivial once the trace formula is written? Equality of trace expressions is algebraically immediate after the Born form is known. The conceptual burden is to show that the pre-probabilistic abundance construction is compatible with unitary equivalence of realizations and ignores unused ancillary sectors. The theorem makes that descent explicit. 7.4 Is “gauge freedom” overstated? No. The present paper establishes a concrete gauge-like structure at the level of quantum-mechanical representation. Equivalent purifications, equivalent Naimark dilations, and equivalent refinement constructions generate classes of mathematically distinct realizations that induce identical event measures. The term gauge is therefore used in its structural sense: a family of redundant descriptions connected by transformations that leave physically meaningful quantities invariant. The present work does not yet construct a local gauge field, connection, curvature operator, or associated dynamics. However, the Representation Invariance Theorem identifies the mathematical architecture from which such structures may emerge. In particular, future extensions involving approximate record overlap, continuous refinement spaces, operator-algebraic formulations, or infinite-dimensional generalizations may naturally promote representation freedom into a richer geometric framework. The current theorem should therefore be understood not as the conclusion of the gauge program, but as its finite-dimensional foundation. 7.5 Does the framework solve approximate decoherence? The framework now proves a finite-dimensional, effect-wise robustness theorem for approximately orthogonal normalized records. Theorem 5.2 combines a Gram-matrix overlap estimate, Löwdin symmetric orthogonalization, rank-one subspace perturbation, and the operational Lipschitz bound to show that the induced measure differs from its idealized projective counterpart by at most b((n− 1)ε) = 1− q 1− (n− 1)ε, provided (n− 1)ε < 1. |T (E | ρ)− T (F | σ)|≤ 2D(ρ,σ) +∥E− F∥ ∞ , 23

Gauge Freedom and the Born RuleMichael Tolbert where D(ρ,σ) = 1 2 ∥ρ− σ∥ 1 is the trace distance. This shows that the derived measure degrades continuously and predictably under finite record overlap rather than collapsing under small departures from orthogonality. The proved estimate is dimension-dependent and effect-wise. Extending it to normalized nonorthogonal POVMs, higher- rank macroscopic sectors, continuous outcome structures, and geometric transport across families of record algebras remains an open path toward the broader gauge program. The remaining challenge is no longer the basic overlap-to-error bridge, which Theorem 5.2 supplies for finite rank-one records. The next challenge is to lift that bridge to full measurement architectures and to determine whether smoothly varying orthogonalization choices define connection-like transport on families of equivalent or approximately equivalent realizations. 7.6 Research Program and Future Directions The present manuscript is the second stage of an intentionally sequenced research program. The sequence is cumulative rather than merely thematic: each stage preserves the distinction among primitive equivalence, cardinal abundance, induced event measure, and credence while enlarging the class of admissible representations. StagePrimary domainProgrammatic role Paper 1Finite pure-state PVMsEstablishes equal-norm refinement, primitive equivalence, cardinal composition, common-refinement independence, and the Born measure on finite projective events. Paper 2Mixed states and finite POVMs Extends the construction through purification and Naimark dilation, proves representation invariance, and establishes finite-overlap robustness for rank-one record effects. Paper 3Complete measurement architectures and continuous realization families Planned direction: develop full POVM robustness, including normalization, coarse graining, higher-rank sectors, and perturbations of complete measurements; investigate transport and connection-like structure across continuously varying realizations. Paper 4Operator-algebraic and Type III settings Planned direction: seek an extension beyond finite representative spaces, potentially replacing finite cardinal structure with algebraic weights, measure classes, or modular structures suited to Type III von Neumann algebras. The later stages are not assumed by the present theorems. They identify concrete extension problems opened by the finite-dimensional results: collective robustness of complete POVMs, geometric transport across realization space, and reformulation of primitive abundance where minimal projections and finite cardinal refinements are unavailable. The current paper supplies the representation-invariant foundation on which those questions can be posed precisely. 24

Gauge Freedom and the Born RuleMichael Tolbert 8 Conclusion The unitary-refinement program can be extended from finite pure-state PVMs to finite mixed-state POVMs by combining purification, Naimark dilation, and the original cardinality representation theorem. These enlarged constructions do not replace the Paper 1 abundance framework; they provide the pure projective carrier in which its primitive-equivalent descendants and homogeneous refinements are represented. The enlarged-space construction first fixes cardinal abundance and then induces an event measure represented by∥Q i Ψ∥ 2 . Compression yieldsTr(ρE i ). The central advance is the Representation Invariance Theorem 4.5: equivalent purification-dilation realizations generate the same induced measure, so the measure descends to the operational pair (ρ,{E i }). Ancillary realizations therefore form equivalence classes analogous to gauge orbits. Refinement denominators, purification bases, dilation spaces, and unused ancillary sectors are representational choices rather than new physical weights. The generalized Born measure is constant under those choices. This unifies common-refinement independence, preparation independence, and dilation independence under a broader principle of representation invariance. The exact representation theorem remains conditional on the finite-abundance conception developed in the projective case. Theorem 5.2 now adds an explicit finite-overlap robustness result: for fixed record numbern, approximately orthogonal rank-one records induce effect-wise typicalities converging to the ideal projective values asε →0. The finite-overlap theorem is dimension- dependent, while Theorem 5.3 additionally controls canonical normalization and all coarse-event sums for approximately normalized finite positive families. Deriving the required near-normalization estimate from arbitrary higher-rank or continuous record architectures remains open. Together, the results establish a rigorous bridge from microscopic overlap and collective normalization error to controlled measure error and provide a concrete foundation for later geometric and operator-algebraic extensions. A Proof Architecture at a Glance StageRole Operational inputA density operator ρ and finite POVM {E i }. Unified realizationPurify ρ and dilate {E i } to a pure state |Ψ⟩ and PVM {Q i }. Projective abundance Apply equal-norm refinement and cardinal composition in the enlarged space. Generalized Born formObtain T R (i) =∥Q i Ψ∥ 2 = Tr(ρE i ). Representation descentProve equality across equivalent purifications and dilations. Approximate extensionBound changes under nearby states, effects, and nearly orthogonal records. References [1]Bill Moran, Stephen Howard, and Doug Cochran. Positive operator valued measures: A general setting for frames. In Excursions in Harmonic Analysis, Volume 1. Birkhäuser, 2013. Also 25

Gauge Freedom and the Born RuleMichael Tolbert available as arXiv:1111.1450. [2] Andrew M. Gleason. Measures on the closed subspaces of a hilbert space. Journal of Mathematics and Mechanics, 6:885–893, 1957. [3]Paul Busch. Quantum states and generalized observables: A simple proof of gleason’s theorem. Physical Review Letters, 91:120403, 2003. [4] Robert W. Spekkens. Contextuality for preparations, transformations, and unsharp measure- ments. Physical Review A, 71:052108, 2005. [5]Roberto Beneduci. Notes on naimark’s dilation theorem. Journal of Physics: Conference Series, 1638:012006, 2020. [6] Chandler Davis and William M. Kahan. The rotation of eigenvectors by a perturbation. iii. SIAM Journal on Numerical Analysis, 7(1):1–46, 1970. [7]David Deutsch. Quantum theory of probability and decisions. Proceedings of the Royal Society A, 455:3129–3137, 1999. [8]Lane P. Hughston, Richard Jozsa, and William K. Wootters. A complete classification of quantum ensembles having a given density matrix. Physics Letters A, 183:14–18, 1993. [9]Per-Olov Löwdin. On the non-orthogonality problem connected with the use of atomic wave functions in the theory of molecules and crystals. Journal of Chemical Physics, 18:365–375, 1950. [10] Mark A. Naimark. On a representation of additive operator set functions. C. R. (Doklady) Acad. Sci. URSS (N.S.), 41:359–361, 1943. [11]Michael Tolbert. Typicality, unitary refinement, and the origin of the born rule. Manuscript, 2026. [12] David Wallace. The Emergent Multiverse. Oxford University Press, 2012. [13]Wojciech H. Zurek. Decoherence, einselection, and the quantum origins of the classical. Reviews of Modern Physics, 75:715–775, 2003. 26