Evolving Spectral Triples

Evolving Spectral Triples Francesco D’Agostino October 30, 2025 Abstract In this article we extend Connes’ spectral triple formalism to include time evolution by introducing a strongly continuous automorphism group Φtthat acts coherently on the algebra, Hilbert space, and Dirac operator. This yields a family of time-dependent spectral triples (A(t), H(t), T (t)) that remain unitarily equivalent to the original structure. We define a dynamic spectral distance that is well-behaved and covariant under the flow. By coupling the operator evolution to geometric curvature through ˙ T(t)=−λR(T, t), we obtain a spectral flow formula relating analytic variations to topological invariants. The K-theory and K-homology classes, the Chern–Connes character, and index pairings are all preserved under evolution. When boundary data is included, we prove a bulk–boundary correspondence: the variation of interaction energy equals the topological boundary index. This framework unifies spectral dynamics with index theory and cohomological stability in noncommutative geometry. 1 Introduction Noncommutative geometry, as formulated by Connes [1], encodes geometric and topological information through a spectral triple (A, H, D): a C∗-algebra Aon a Hilbert space Hwith a self-adjoint Dirac operator D. We extend this framework to include time dependence via a strongly continuous automorphism group Φt, producing an evolving family (A(t), H(t), T(t)) that preserves the spectral triple structure. We generalize the spectral distance to this dynamic setting and introduce a curvature-driven evolution ˙ T(t) = −λR(T, t), where R(T, t) = [[T(t),˙ T(t)], T(t)]. Under this evolution, all topological invariants are preserved: K-theory classes, K-homology, and the Chern–Connes character remain constant. We prove a spectral flow formula linking analytic variations to topological flux, and establish a bulk–boundary correspondence where energy variation equals the boundary index. This provides a dynamic framework for spectral geometry while maintaining the core correspondence between analysis, topology, and cohomology. 2 Structure-Preserving Time Evolution To establish a coherent foundation for this dynamic system, we begin by recalling that we work within the standard setting of spectral triples as introduced by Connes [1]. The classical triple (A, H, D) provides the algebraic and analytic structure of noncommutative geometry, and we adopt it here without modification. We therefore take Aand Hexactly as in the original formulation in [1] for all analytic details and structural assumptions. Our focus will not be on altering the static framework but on introducing a consistent notion of time evolution acting upon it. The Dirac-type operator T(instead of the standard D) will serve as the generator of this evolution, giving rise to a time-dependent family T(t) that carries the geometric and topological dynamics to be analyzed in the following sections. At present we concentrate exclusively on the formal introduction of time dependence. To describe this evolution uniformly across all components, we introduce an external automorphic flow that governs their transformation in time. The central idea is that time acts not separately on each component but through a single, coherent reversible process. We formalize this as follows. Definition 1 (Time Evolution of Spectral Triples).Let (A, H, T)be a spectral triple. A time evolution of (A, H, T)is a strongly continuous one-parameter group of morphisms Φt: (A, H, T)7−→ (A(t), H(t), T(t)), t ∈R,(1) satisfying the group properties Φt+s= Φt◦Φs,Φ0= id.(2) 1 The image (A(t), H(t), T(t)) := Φt(A, H, T)is called the dynamic spectral triple at time t. The action of Φtis determined by A(t)=αt(A), H(t)=U(t)H, T(t)=U(t)T U(t)−1,(3) where: I. αt:A→Ais a strongly continuous one-parameter group of ∗-automorphisms; II. U(t) : H→His a strongly continuous unitary group implementing αton H, i.e. U(t)π(a)U(t)−1=π(αt(a)), a ∈A; (4) III. the domain of Tis invariant under the evolution: U(t)Dom(T) = Dom(T), t ∈R; (5) IV. the map t7→ Φtis continuous in the strong operator topology on each component (A(t)in the strong-∗topology, and T(t), U(t)in the strong operator topology). Having introduced the evolution mechanism, we now verify that the fundamental properties defining a spectral triple are preserved along the flow. The following result expresses that the evolution acts by structure-preserving equivalence rather than deformation. Proposition 1 (Preservation of Spectral Triple Structure).If Φtis a strongly continuous one-parameter automorphism group acting as above, then for every t∈Rthe triple (A, H, T)satisfies all defining conditions of a spectral triple. In particular: I. For every a∈A, the commutator [T(t), π(αt(a))] is bounded on H. II. The operator (1 + T(t)2)−1is compact. III. The triples (A(t), H(t), T(t)) and (A(t), H(t), T (t)) are unitarily equivalent. Proof. Let (A, H, T ) be a spectral triple [1], and let Φt: (A, H, T )7→ (A(t), H(t), T(t)) be a strongly continuous one-parameter automorphism group acting as specified in Definition 1. Recall that Φtacts on the individual components by A(t) = αt(A), H(t) = H, T (t) = U(t)T U(t)−1,(6) where (αt)t∈Ris a strongly continuous group of ∗-automorphisms of Aand (U(t))t∈Ris a strongly continuous unitary group on Hwith Dom(T) invariant under all U(t), ensuring T(t) is self-adjoint. Implementing αt, that is, U(t)π(a)U(t)−1=π(αt(a)), a ∈A. (7) We verify the three required properties separately. (i.) Fix a∈A. Note that we use the same representation π:A→B(H) throughout, with π(αt(a)) denoting the representation of the evolved algebra element αt(a)∈A(t) on the original Hilbert space H. Using (7) and the definition T(t)=U(t)T U(t)−1, we compute [T(t), π(αt(a))] = U(t)T U(t)−1U(t)π(a)U(t)−1−U(t)π(a)U(t)−1U(t)T U(t)−1 =U(t)Tπ(a)−π(a)TU(t)−1=U(t) [T, π(a)] U(t)−1.(8) Since [T, π(a)] is bounded by the defining property of a spectral triple, the conjugate U(t)[T, π(a)]U(t)−1 is also bounded, because ∥U(t)[T, π(a)]U(t)−1∥=∥[T, π(a)]∥.(9) Therefore [T(t), π(αt(a))] is bounded on Hfor all a∈Aand t∈R. (ii.) We next show that (1 + T(t)2)−1is compact. Using the unitary conjugation definition, T(t)2= (U(t)TU(t)−1)(U(t)TU(t)−1)=U(t)T2U(t)−1,(10) and hence 1+T(t)2=U(t)(1 + T2)U(t)−1.(11) 2 Since the functional calculus is preserved under unitary conjugation, (1+T(t)2)−1=U(t)(1 + T2)−1U(t)−1.(12) Because (1 + T2)−1is compact by assumption and compactness is invariant under unitary conjugation, (1+T(t)2)−1is compact for every t∈R. (iii.) Define the map Ψt:H−→ H(t),Ψt(ψ) = U(t)ψ. (13) This map is unitary by construction, and it satisfies Ψtπ(a) Ψ−1 t=U(t)π(a)U(t)−1=π(αt(a)), a ∈A, (14) and ΨtTΨ−1 t=U(t)TU(t)−1=T(t).(15) Hence the representation and the Dirac-type operator of (A(t), H(t), T (t)) are obtained from those of (A, H, T) by conjugation with the unitary Ψt. This establishes that the two triples are unitarily equivalent. Explicitly, Φt(A, H, T) = (αt(A), U(t)H, U(t)TU(t)−1)≃(A(t), H(t), T(t)),(16) where ≃denotes unitary equivalence of spectral triples. Since properties (i.)–(iii.) are all verified, it follows that for every t∈Rthe triple (A(t), H(t), T(t)) satisfies the defining conditions of a spectral triple. All transformations depend continuously on t: strong continuity of αtand U(t) by assumption directly implies strong continuity of T(t)=U(t)TU(t)−1and of the representation π(αt(a)) on H(t) for each a∈A. 3 Dynamic Structure Having established the structural integrity of the dynamic spectral triple, we now extend the metric content of noncommutative geometry to the time-dependent framework. In the static case, Connes’ spectral distance provides a metric on the state space of the algebra. Our objective is to generalize this construction to a smoothly evolving family of spectral triples Φt(A, H, T) = (A(t), H(t), T(t)). Definition 2 (Dynamic Spectral Distance).Let Φtbe the time evolution defined in Definition 1, and let Adenote the C∗-algebra norm closure of Ain B(H). Let S(A)denote the space of states (positive linear functionals of norm one) on A. For each fixed t∈Rand for φ, ψ ∈ S(A), define the dynamic spectral distance by dt(φ, ψ) = sup |φ(αt(a)) −ψ(αt(a))|:a∈A, ∥[T(t), π(αt(a))]∥ ≤ 1,(17) where π:A→B(H)denotes the canonical representation from the base spectral triple. Remark 1. (i) The supremum in (2) may take the value +∞, as in the static case of Connes’ spectral metric. (ii) At t= 0, one recovers the usual spectral distance associated with (A, H, T): d0(φ, ψ) = sup |φ(a)−ψ(a)|:a∈A, ∥[T, π(a)]∥≤1.(18) (iii) The definition measures how fixed states on the original algebra separate under the time-evolved geometry determined by T(t)and αt. Remark 2. Although dtdepends explicitly on the time parameter t, Proposition 1 implies that the distance is actually time-invariant when the states are appropriately transformed. More precisely, if we define φt=φ◦α−1 tand ψt=ψ◦α−1 t, then dt(φ, ψ)=d0(φt, ψt)by the unitary equivalence of the triples. The value of introducing dtis that it allows us to track how fixed states on the original algebra Aseparate as the geometry evolves. 3 The definition follows the same variational form as Connes’ spectral distance, with the replacement of the static algebra elements a∈Aand Dirac operator Tby their time-evolved counterparts αt(a)∈A(t) and T(t). The supremum is taken over all a∈Asuch that the evolved commutator [T(t), π(αt(a))] has operator norm at most one, ensuring that only “Lipschitz elements” with respect to the evolved geometry contribute to the distance. Proposition 2 (Well-Definedness and Continuity).Let Φtbe a strongly continuous one-parameter automorphism group acting on the spectral triple (A(t), H(t), T (t)). Then: I. For each fixed t∈R,dtdefines a metric on S(A)taking values in [0,∞), and dt(φ, ψ)=0if and only if φ=ψ. II. The map t7→ dt(φ, ψ)is continuous on Rfor every pair of states φ, ψ ∈ S(A). III. For the trivial evolution Φt= id, one recovers the standard spectral distance: dt(φ, ψ)=d0(φ, ψ) = sup |φ(a)−ψ(a)|:a∈A, ∥[T, π(a)]∥≤1.(19) IV. (Unitary Invariance) For all s, t ∈R, dt(φ◦αs, ψ ◦αs)=dt−s(φ, ψ).(20) Proof. Let (A(t), H(t), T(t)) and Φtbe as in Definition 1, so that A(t)=αt(A) and T(t)=U(t)T U(t)−1 for a strongly continuous one-parameter unitary group U(t) implementing the ∗-automorphism group αt.(i.) Fix t∈Rand φ, ψ ∈ S(A). For any a∈A,αt(a)∈Asince αtis an automorphism. Because (A(t), H(t), T(t)) satisfies the defining properties of a spectral triple (Proposition 1), the commutator [T(t), π(αt(a))] is bounded, hence the admissible set Lt={a∈A:∥[T(t), π(αt(a))]∥ ≤ 1}(21) is nonempty. The evaluation map a7→ |φ(αt(a)) −ψ(αt(a))|is norm-continuous on Abecause φand ψare continuous on the C∗-closure A. Since the seminorm Lt(a) = ∥[T(t), π(αt(a))]∥vanishes only on scalar multiples of the identity (as in the standard Connes framework), Ltis bounded modulo scalars, and hence the supremum over Ltis finite. Positivity and symmetry follow directly from the absolute value. If φ=ψ, then dt(φ, ψ) = 0 trivially. Conversely, if dt(φ, ψ) = 0, then |φ(αt(a)) −ψ(αt(a))|= 0 for all a∈Aby density of the domain defining the supremum, hence φ=ψon A. Finally, the triangle inequality follows from linearity of φ, ψ and the usual argument for Connes’ metric based on the triangle inequality for the absolute value. Thus dtdefines a genuine metric on S(A). (ii.) Fix φ, ψ ∈ S(A) and t0∈R. We show that limt→t0dt(φ, ψ)=dt0(φ, ψ). For each a∈A, strong continuity of t7→ αt(a) and t7→ T(t) implies that t7→ |φ(αt(a)) −ψ(αt(a))|is continuous. To ensure the same for t7→ ∥[T(t), π(αt(a))]∥, we assume—as part of the regularity of the evolution—that t7→ [T(t), π(αt(a))] is continuous in operator norm for each a∈A. This is satisfied, for instance, whenever U(t) and αtare norm–continuous on bounded sets. Let ε>0. Choose a0∈Asuch that |φ(αt0(a0)) −ψ(αt0(a0))|≥dt0(φ, ψ)−ε, ∥[T(t0), π(αt0(a0))]∥≤1.(22) By continuity, there exists δ > 0 such that for all |t−t0|< δ, ∥[T(t), π(αt(a0))]∥ ≤ 1 and |φ(αt(a0)) −ψ(αt(a0))|−|φ(αt0(a0)) −ψ(αt0(a0))|< ε. (23) Therefore, for |t−t0|< δ,a0is admissible in the definition of dt, and dt(φ, ψ)≥ |φ(αt(a0)) −ψ(αt(a0))| ≥ dt0(φ, ψ)−2ε. (24) This gives lim inft→t0dt(φ, ψ)≥dt0(φ, ψ).For the reverse inequality, let ε > 0 and |t−t0|< δ (to be determined). For any a∈Awith ∥[T(t), π(αt(a))]∥ ≤ 1, the assumed norm–continuity of [T(t), π(αt(a))] implies that there exists δ(a)>0 such that ∥[T(t0), π(αt0(a))]∥ ≤ 1 whenever |t−t0|< δ(a). By uniform continuity of the maps a7→ ∥[T(t), π(αt(a))]∥and a7→ |φ(αt(a)) −ψ(αt(a))|on bounded subsets of A, we may select a common δ > 0 (independent of a) such that the following estimate holds: sup a:∥[T(t),π(αt(a))]∥≤1 |φ(αt(a)) −ψ(αt(a))| ≤ sup a:∥[T(t0),π(αt0(a))]∥≤1 |φ(αt0(a)) −ψ(αt0(a))|+ 2ε. (25) 4 Hence lim supt→t0dt(φ, ψ)≤dt0(φ, ψ). Since ε>0 is arbitrary, continuity follows. (iii.) If Φt= id for all t, then αt= id and U(t) = id, so A(t) = Aand T(t) = T. The definition of dt then reduces to dt(φ, ψ) = sup{|φ(a)−ψ(a)|:∥[T, π(a)]∥≤1}=d0(φ, ψ),(26) which coincides exactly with the standard spectral distance of Connes. (iv.) For any s, t ∈Rand a∈A, the automorphism property of the flow gives αt(αs(a))=αt+s(a). Then for states φ, ψ ∈ S(A), dt(φ◦αs, ψ ◦αs) = sup a∈A|φ(αs(αt(a))) −ψ(αs(αt(a)))|:∥[T(t), π(αt(a))]∥ ≤ 1 = sup a∈A|φ(αt+s(a)) −ψ(αt+s(a))|:∥[T(t), π(αt(a))]∥ ≤ 1.(27) To identify the corresponding admissible sets, set b=αs(a) so that a=α−s(b) and αt(a)=αt(α−s(b)) = αt−s(b). Using the covariance of the Dirac-type operators under the evolution T(t) = U(t)TU(t)−1, we have [T(t), π(αt(a))] = [T(t), π(αt−s(b))] = U(s) [T(t−s), π(αt−s(b))] U(s)−1.(28) Since conjugation by a unitary preserves the operator norm, ∥[T(t), π(αt(a))]∥=∥[T(t−s), π(αt−s(b))]∥.(29) Hence the admissible set at time t, Lt={a∈A:∥[T(t), π(αt(a))]∥≤1},(30) is carried bijectively by αsonto Lt−s={b∈A:∥[T(t−s), π(αt−s(b))]∥ ≤ 1}. Under this change of variables, the evaluation term transforms as |φ(αt+s(a)) −ψ(αt+s(a))|=|φ(αt(b)) −ψ(αt(b))|.(31) Therefore, dt(φ◦αs, ψ ◦αs) = sup b∈A|φ(αt(b)) −ψ(αt(b))|:∥[T(t−s), π(αt−s(b))]∥ ≤ 1 =dt−s(φ, ψ), (32) which proves the unitary covariant invariance of the dynamic spectral distance. All four statements are therefore verified, completing the proof. 3.1 Dynamic Curvature and Operator Flows Having established the metric aspect of the evolving geometry, we now turn to its differential structure. The goal of this section is to introduce an operator-theoretic notion of curvature that captures how the spectral data bends and twists as the geometry evolves in time. In the classical setting, curvature arises from the commutator of covariant derivatives; here, commutators with the Dirac-type operator play the analogous role. We formalize the infinitesimal rate of change of the geometry and the associated curvature operator as follows. Definition 3 (Time Derivative of the Dirac-Type Operator).Let (A(t), H(t), T (t)) be a spectral triple and Φtits time evolution as in Definition 1, with T(t) = U(t)T U(t)−1. Assume that t7→ T(t)is strongly differentiable on a dense invariant domain Dom(T)⊂H, so that the following strong limit in B(H) exists: ˙ T(t) = dT(t) dt = lim h→0 U(t+h)TU(t+h)−1−U(t)TU(t)−1 h,on Dom(T).(33) If the unitary group (U(t))t∈Ris generated by a self-adjoint operator Xon Hsatisfying U(t)=eitX , then differentiation yields ˙ T(t) = i[X, T(t)],(34) which expresses the infinitesimal evolution of the Dirac-type operator as a commutator with the generator of time translations. 5 This quantity measures how the local differential structure of the geometry changes under the automorphic flow. To capture the geometric obstruction to flat evolution—analogous to curvature in classical differential geometry, we introduce an operator-valued two-form built from commutators of T(t) and its derivative. Definition 4 (Dynamic Curvature Operator).Let T(t)and ˙ T(t)be as in Definition 3. Assume that T(t)and ˙ T(t)share a common dense invariant domain Dom(T)⊂Hon which both commutators below are well-defined and closable. The dynamic curvature operator associated with the evolving spectral triple is the (a priori possibly unbounded) operator on Hgiven by R(T, t) = [[T(t),˙ T(t)], T(t)],(35) whenever this iterated commutator is well-defined on Dom(T). This operator encodes the second-order response of the noncommutative geometry to its own time evolution. In analogy with the curvature twoform in classical geometry, R(T, t)measures the noncommutativity of successive infinitesimal evolutions of the differential structure determined by T(t). The operator R(T, t) thus acts as a curvature measure in the dynamic setting: when [T(t),˙ T(t)] = 0, the flow is said to be flat, representing a geometry that evolves without internal distortion. The next step is to establish the analytical properties of R(T, t) and to derive the operator flow equations governing the evolution of T(t). Proposition 3 (Analytic and Structural Properties of the Dynamic Curvature).Let (A(t), H(t), T(t)) be a spectral triple and let Φtbe its time evolution as in Definition 1, with T(t) = U(t)T U(t)−1for a strongly continuous unitary group U(t) = eitX generated by a self-adjoint operator Xon Hthat leaves Dom(T)invariant. Assume that Tis self-adjoint and that there exists a common dense invariant core D ⊂ Dom(T)such that [X, T],[T, [X, T ]] are densely defined and closable on D,(36) and that Dis invariant under U(t)for all t∈R. The algebraic commutator identities [X, T(t)]=U(t)[X, T]U(t)−1,[[T(t),˙ T(t)], T(t)]=U(t)[[T, [X, T ]], T]U(t)−1,(37) hold on D. Define ˙ T(t) = i[X, T(t)] and R(T, t) = [[T(t),˙ T(t)], T(t)].(38) Then: (I.) (Well-definedness) R(T, t)is densely defined on Dand closable on H. (II.) (Skew-adjointness) R(T, t)∗=−R(T, t)for all t∈R. In particular, iR(T, t)is self-adjoint. (III.) (Covariance) R(T, t)=U(t)R(T, 0) U(t)−1, so the dynamic curvature transforms covariantly under the unitary evolution. Proof. (i.) Since U(t) is a strongly continuous unitary group with generator Xself-adjoint and leaving Dom(T) invariant, the family T(t)=U(t)T U(t)−1consists of self-adjoint operators with Dom(T(t)) = U(t)Dom(T). By unitary covariance, Dom(T2) is mapped onto itself and remains a common invariant core for T(t) and for its time derivative ˙ T(t) = i[X, T(t)]. Hence both T(t) and ˙ T(t) are densely defined and closable on Dom(T2), and the commutator [T(t),˙ T(t)] is well defined on this same dense core. Consequently, the iterated commutator R(T, t) = [[T(t),˙ T(t)], T(t)] (39) is densely defined on Dom(T2). Because each commutator entering the definition is closable on the common invariant domain Dom(T2), their finite algebraic composition R(T, t) is also closable on H. (ii.) Because T(t) is self-adjoint and ˙ T(t) = i[X, T(t)] is skew-adjoint, we have [T(t),˙ T(t)]∗=−[T(t),˙ T(t)].(40) 6 Therefore, R(T, t)∗= ([[T(t),˙ T(t)], T(t)])∗ = (T(t)[T(t),˙ T(t)] −[T(t),˙ T(t)]T(t))∗ = [T(t),˙ T(t)]∗T(t)∗−T(t)∗[T(t),˙ T(t)]∗ = [T(t),˙ T(t)]∗T(t)−T(t)[T(t),˙ T(t)]∗ = [T(t),[T(t),˙ T(t)]∗] = [T(t),−[T(t),˙ T(t)]] =−[[T(t),˙ T(t)], T(t)]=−R(T, t). (41) Hence R(T, t) is skew-adjoint, and consequently iR(T, t) is self-adjoint. (iii.) Since U(t) = eitX commutes with its generator X, we have U(t)XU(t)−1=X, and therefore ˙ T(t) = i[X, T(t)] = i[X, U(t)TU(t)−1]=U(t)i[X, T]U(t)−1.(42) Consequently, R(T, t) = [[U(t)T U(t)−1, U(t)(i[X, T])U(t)−1], U(t)TU(t)−1] =U(t) [[T, i[X, T]], T]U(t)−1=U(t)R(T, 0)U(t)−1,(43) where we used that conjugation by unitaries distributes over commutators. The curvature operator captures how the noncommutative geometry deforms under its own evolution. In classical geometric flows—Ricci flow being the canonical example—the metric evolves according to its curvature. We now establish the analogous mechanism in our setting: a flow equation that governs the time evolution of the Dirac operator in terms of the dynamic curvature R(T, t). Definition 5 (Curvature-Driven Evolution).A time evolution Φtof a spectral triple (A(t), H(t), T(t)) with generator Xis called curvature-driven with parameter λ∈Rif ˙ T(t) = −λR(T, t),(44) or equivalently, if Xand Tsatisfy [X, T]=−iλ[[T, [X, T ]], T] (45) on Dom(T2). Remark 3. This condition establishes a self-consistent relationship between the evolution generator and the geometry: the rate of change of the Dirac operator is proportional to its own curvature. The parameter λcontrols the coupling strength between dynamics and geometry. When λ= 0, the evolution is curvature-free. Example 1 (Trivial Curvature-Driven Evolution).If [X, T]=0, then ˙ T(t)=0and R(T, t)=0, so the evolution is trivially curvature-driven for any λ. This represents a ”frozen” geometry where T(t) = T for all t. Proposition 4 (Basic Properties).If Φtis curvature-driven with parameter λ, then: (I.) The curvature remains covariant: R(T, t)=U(t)R(T, 0)U(t)−1. (II.) The constraint is time-independent: if it holds at t= 0, it holds for all t∈R. (III.) The parameter λis invariant under unitary conjugation. Proof. (i.) This is immediate from Proposition 3(iii), which establishes R(T, t)=U(t)R(T, 0)U(t)−1for any strongly continuous unitary evolution, independent of whether the evolution is curvature-driven. (ii.) Assume that at t= 0 the relation [X, T]=−iλ[[T, [X, T ]], T] holds on Dom(T2). Since U(t)=eitX commutes with X, we have for any t∈R: i[X, T(t)]=i[X, U(t)T U(t)−1]=U(t)i[X, T]U(t)−1 =−λ U(t) [[T, [X, T]], T]U(t)−1(applying the t= 0 constraint) =−λ[[U(t)TU(t)−1, U(t)[X, T]U(t)−1], U(t)T U(t)−1] (distributing conjugation) =−λ[[T(t),[X, T(t)]], T(t)] =−λ[[T(t),˙ T(t)], T(t)]=−λ R(T, t). (46) 7 Thus ˙ T(t) = −λR(T, t) holds for all t∈R, establishing that the curvature-driven constraint is preserved under time evolution. (iii.) Let Vbe any unitary operator on H, and define ˜ T=V TV ∗and ˜ X=V XV ∗. Since conjugation by unitaries preserves commutator structure, we have [˜ X, ˜ T] = V[X, T]V∗=−iλ V [[T, [X, T]], T ]V∗=−iλ [[ ˜ T, [˜ X, ˜ T]],˜ T].(47) Therefore, the conjugated pair ( ˜ X, ˜ T) satisfies the curvature-driven relation with the same parameter λ, showing that λis a unitary invariant of the spectral triple. 4 Topological Structure of the Evolving Algebra We now turn to the algebraic foundation underlying the dynamic framework. In the setting of noncommutative geometry, the algebra Areplaces the role of the algebra of continuous functions on a topological space and therefore encodes the topological structure of the noncommutative space. To understand how topology behaves under time evolution, we examine how the family of algebras A(t)=αt(A), induced by the automorphic flow Φt, relates to the original algebra A. This analysis establishes that the evolution acts by topological equivalence at the algebraic level, preserving the K-theoretic content of the system. Definition 6 (Topological Equivalence of Evolving Algebras).Let (A(t), H(t), T(t)) be a spectral triple and let Φtbe the strongly continuous one-parameter automorphism group defined in Definition 1, with A(t)=αt(A)and α0= id. We say that the evolution is topologically equivalent if for each t∈R, the induced map αt:A−→ A(t) (48) isa∗-isomorphism extending continuously to the C∗-closures Aand A(t), and such that αtinduces group isomorphisms on the K-theory groups (αt)∗:K∗(A)−→ K∗(A(t)).(49) Because αtis by construction a ∗-automorphism of A, this condition is automatically satisfied. In particular, the extensions of αtto the C∗-closures Aand A(t) yield isomorphic C∗-algebras, preserving all topological information encoded in the algebraic structure. The corresponding invariance result follows. Proposition 5 (Algebraic Topological Invariance).Let Φtbe the time evolution of the spectral triple (A(t), H(t), T(t)) as in Definition 1. Each αtextends continuously to a *-isomorphism between the C∗-closures Aand A(t). Then for all t∈R, K∗(A(t)) ∼ =K∗(A),(50) with the isomorphism induced by αt. In particular, the K-theory class of any projection or unitary in A is preserved under time evolution: [p(t)] = [αt(p)] = [p],[u(t)] = [αt(u)] = [u].(51) Proof. Since αt:A→A(t)isa∗-automorphism, it extends to an isometric isomorphism of C∗-algebras A→A(t). K-theory for C∗-algebras is invariant under ∗-isomorphisms (see, for example, [2], [3]): if f:B1→B2is a ∗-isomorphism, then f∗defines a canonical group isomorphism K∗(B1)→K∗(B2). Applying this to f=αt, we obtain (αt)∗:K∗(A)→K∗(A(t)) as an isomorphism. Thus K∗(A(t)) ∼ = K∗(A) for all t. For a projection p∈Mn(A) representing a class in K0(A), αt(p)∈Mn(A(t)) is again a projection and defines the same class under the induced map. The same argument applies to unitaries in K1(A). Hence both K0and K1are preserved under Φt. This result establishes that the time evolution acts by topological equivalence: the algebraic backbone of the noncommutative space remains unchanged, even as the geometric operator T(t) evolves dynamically. In classical terms, the “underlying space” of the noncommutative geometry is topologically static, while its metric and curvature structures evolve through T(t). This distinction between topological invariance and geometric evolution forms the starting point for the next layers of analysis, where K-homological and cohomological structures will be examined. 8 Theorem 1 (Emergent Topological Stability of the Evolving Algebra).Let (A(t), H(t), T(t)) be a spectral triple and let Φtbe the strongly continuous one-parameter automorphism group of Definition 1. Assume that Φtacts by ∗-automorphisms αton Aand by unitaries U(t)on H, with A(t)=αt(A), T(t) = U(t)TU(t)−1.(52) Then the following statements hold: (I.) The C∗-algebras Aand A(t)are ∗-isomorphic, and hence their K-theory groups coincide: K∗(A(t)) ∼ =K∗(A).(53) (II.) Each Dirac-type operator T(t)defines a Fredholm module (πt, H, T (t)) representing a K-homology class [T(t)] ∈K∗(A(t)), and the family t7→ [T(t)] is continuous in the topology of K-homology. (III.) The pairing between K-theory and K-homology, ⟨[p],[T(t)]⟩= Index(pT(t)p),(54) is independent of t. In particular, ⟨[p],[T(0)]⟩=⟨[p],[T(0)]⟩,(55) so that the analytic evolution induced by Φtpreserves the topological content of the noncommutative geometry. Proof. (i.) By Definition 1, Φtacts on Avia the *-automorphism αt:A→A(t), implemented by unitary conjugation αt(a)=U(t)aU(t)−1. Since unitary conjugation preserves the operator norm, αtextends to a *-isomorphism A→A(t). By Proposition 5, this induces a group isomorphism (αt)∗:K∗(A)→ K∗(A(t)), establishing K∗(A(t)) ∼ =K∗(A). (ii.) By Proposition 1, (A(t), H(t), T (t)) is a spectral triple for each t∈R, hence (πt, H, T (t)) defines a Fredholm module over A(t) and a K-homology class [T(t)] ∈K∗(A(t)). Since U(t) is strongly continuous, T(t) = U(t)TU(t)−1is continuous in the strong resolvent topology [4], assuming U(t)Dom(T) = Dom(T). By functional calculus, the bounded transform F(t) = T(t)(1 +T(t)2)−1/2is then strongly continuous in the operator topology. In Kasparov’s KK-theory, Fredholm modules (H, F(t)) connected by a strongly continuous path (with uniformly compact commutators [F(t), a] for a∈A(t)) represent the same Khomology class [5]. Since conjugation by the unitary U(t) preserves the compactness of commutators— that is, [T(t), αt(a)] = U(t)[T, a]U(t)−1is compact whenever [T, a] is compact—the family F(t) satisfies the conditions for operator homotopy equivalence. Therefore [T(t)] = [T(t0)] in K∗(A(t)) for all t, t0∈R. The evolution thus preserves the topological data encoded in [T]. (iii.) Let p∈Mn(A) be a projection representing [p]∈K0(A). By Proposition 5, p(t)=αt(p) represents (αt)∗[p] in K0(A(t)). Since T(t) = U(t)TU(t)−1and αtis implemented by U(t) (Definition 1), we have p(t)T(t)p(t) = U(t)pTp U(t)−1.(56) By unitary invariance of the Fredholm index, ⟨[p(t)],[T(t)]⟩= Index(p(t)T(t)p(t)) = Index(pTp)=⟨[p],[T(0)]⟩.(57) For u∈Mn(A) unitary, the K1-pairing is ⟨[u],[T]⟩= Index(PuP), where Pis the positive spectral projection of T. Unitary conjugation by U(t) preserves this index as well. Hence the index pairing is time-independent for all K-theory classes, establishing that the topological content is preserved under Φt. 5 Cohomological and Spectral Stability Having established that the K-theory and K-homology classes remain invariant under the automorphic evolution Φt, we now pass to the dual cohomological picture. The Chern–Connes character provides the canonical map from the analytic data of a spectral triple to the cyclic cohomology of its underlying algebra, encoding the same topological information in a differential form–like setting. We start by showing that the Chern–Connes character associated with the evolving Dirac-type operator T(t) remains cohomologically constant under time evolution. 9 and its derivative satisfies d dt⟨[p],[T(t)]⟩=1 πTr p˙ T(t)(1 + T(t)2)−1.(110) Differentiating the spectral action S(t) = Tr(f(T(t))) with respect to tand using the same trace-class assumptions yields d dtS(t) = Tr f′(T(t)) ˙ T(t),(111) which coincides with the variation formula for Eint(t) when f(x) = (1 + x2)−1/2and V=f′(T(t))T(t) up to normalization. Hence, the analytic derivative in (108) represents precisely the index pairing of the boundary class δ([p∂]) with the evolving Fredholm module [T(t)], up to the constant factor λ: d dtEint(t)=λ⟨δ([p∂]),[T(t)]⟩.(112) Equation (106) ensures that the connecting map δis covariant under the evolution, so that the righthand side of (112) is invariant under conjugation by U(t). If the evolution is purely unitary, T(t) = U(t)TU(t)−1and V= 0, then R(T, t) = U(t)R(T, 0)U(t)−1and the integrand in (108) is a total trace of a commutator: Tr [V, T (t)](1 + T(t)2)−3/2= 0.(113) Therefore d dt Eint(t) = 0 in this case. Finally establishing the identity (104) and the stated vanishing property for unitary evolution. Corollary 1 (Energy Conservation and Quantized Boundary Flux).Under the hypotheses of Theorem 5, the following are equivalent: (I.) The boundary K-class is invariant under α∂ t. (II.) The interaction energy Eint(t)is constant in time. (III.) The spectral flow between T(0) and T(t)equals the boundary index variation: sf(T(s)) = ⟨δ([p∂]),[T(t)] −[T(0)]⟩.(114) In particular, if δ([p∂]) is nontrivial, the change in interaction energy is quantized and coincides with the integer index carried by the bulk–boundary map. Proof. (i.) By Theorem 5, the boundary map δ:K∗(A∂)→K∗−1(J) satisfies the covariance relation δ◦(α∂ t)∗= (αJ t)∗◦δ. (115) Since α∂ tand αJ tare ∗-isomorphisms (Proposition 5), they induce group isomorphisms on K-theory. Hence the class δ([p∂]) is invariant under the evolution if and only if [p∂] is. If α∂ tacts trivially on K0(A∂), then [p∂(t)] = [p∂] for all t, and therefore δ([p∂(t)]) = δ([p∂]).(116) Thus the first point is addressed. (ii.) Assume the hypotheses of Theorem 5. From the identity established there, d dtEint(t)=λ⟨δ([p∂]),[T(t)]⟩,(117) the variation of the interaction energy is expressed through the K-theoretic pairing between the boundary image δ([p∂]) and the evolving Fredholm class [T(t)]. If the boundary K-class is invariant, then by point (I) δ([p∂]) is time-independent. Moreover, Theorem 4(II) guarantees that [T(t)] = [T(0)] in K∗(A) for all t∈R. Since the pairing ⟨·,·⟩ :K0(J)×K∗(A)−→ Z(118) is bilinear and natural under ∗-isomorphisms, the right-hand side of (117) is constant in t. Hence d dt Eint(t) = 0 and therefore Eint(t)=Eint(0) for all t∈R.(119) 16 Conversely, if Eint(t) is constant in time, then differentiating (117) yields d dt Eint(t) = 0, which implies that ⟨δ([p∂]),[T(t)]⟩is constant in t. Since the pairing separates K-classes in this setting, this entails that both δ([p∂]) and [T(t)] remain fixed, thus the boundary class is invariant. Therefore, the constancy of Eint(t) is equivalent to the invariance of the boundary K-class. (iii.) We again assume the hypotheses of Theorem 5, and in particular the curvature-driven evolution law ˙ T(t) = −λR(T, t) together with the trace-class regularity ensuring the validity of the spectral flow formula of Theorem 3. For any continuously differentiable path of self-adjoint Fredholm operators t7→ T(t), that theorem gives sf(T(s)) = λ πZt 0 Tr ˙ T(s)(1 + T(s)2)−1ds. (120) On the other hand, from the proof of Theorem 5 we have the analytic identity linking the interaction energy and the boundary index pairing: d dsEint(s) = λ⟨δ([p∂]),[T(s)]⟩.(121) Integrating (121) from 0 to tyields Eint(t)−Eint(0) = λZt 0 ⟨δ([p∂]),[T(s)]⟩ds. (122) Because δ([p∂]) is constant under the evolution by point (I), the integral reduces to Eint(t)−Eint(0) = λ⟨δ([p∂]),[T(t)] −[T(0)]⟩.(123) Combining this with the trace representation of spectral flow (120) and using the linearity of the pairing, we find sf(T(s)) = 1 πZt 0 Tr ˙ T(s)(1 + T(s)2)−1ds =⟨δ([p∂]),[T(t)] −[T(0)]⟩,(124) which is precisely the claimed equality. This shows that the spectral flow between T(0) and T(t) measures the same integer index as the variation of the boundary K-class under the connecting map δ. The combination of Theorem 5 and Corollary 1 completes the structural picture: the bulk geometry described by (A, H, T) evolves dynamically according to curvature-driven flow, while its topological and cohomological data remain stable. When boundary coupling is introduced, analytic variations of the spectral data manifest as quantized index transfers across the bulk–boundary interface, providing a unified noncommutative expression for geometric interaction energy 7 Conclusion Concluding, we extended Connes’ spectral geometry to time-dependent settings. The evolution ˙ T(t) = −λR(T, t) preserves the spectral triple structure while generating a continuous family of noncommutative geometries. The spectral flow formula relates analytic variation to topological flux, yielding three core invariance results: K-theory of the algebra, K-homology of the Dirac operator, and the Chern–Connes character all remain constant under evolution. For boundary systems, the bulk–boundary correspondence relates energy derivatives to boundary index pairings, giving conservation laws where analytic energy transfer equals integer-valued topological flux. This framework unifies time evolution, curvature, and topology through exact correspondences, providing a foundation for spectral action principles and dynamical curvature equations in noncommutative spaces. References [1] A. Connes, Noncommutative Geometry, Academic Press, 1994. [2] B. Blackadar, K-Theory for Operator Algebras, Cambridge University Press, 1998. [3] M. Rørdam, F. Larsen, and N. J. Laustsen, An Introduction to K-Theory for C∗-Algebras, Cambridge University Press, 2000. 17 [4] M. Reed and B. Simon, Methods of Modern Mathematical Physics, Vol. I–IV, Academic Press, 1980. [5] G. G. Kasparov, Hilbert C∗-Modules: Theorems of Stinespring and Voiculescu, Journal of Operator Theory, 1980. [6] J. Phillips, Self-Adjoint Fredholm Operators and Spectral Flow, Canadian Mathematical Bulletin, 1996. [7] E. Getzler, Cyclic Homology and the Atiyah–Patodi–Singer Index Theorem, Annals of Mathematics, 1993. [8] A. L. Carey and J. Phillips, Spectral Flow in Fredholm Modules, Eta Invariants and the JLO Chern Character, International Journal of Mathematics, 2006. [9] J. Cuntz and D. Quillen, Cyclic Homology and Nonsingularity, Journal of the American Mathematical Society, 1995. 18