Trace-Based Spectral Definitions

• Trace-based spectral definitions are mathematical frameworks that use trace functionals to sum and reduce spectral data into scalars, ensuring robustness under perturbations.
• They underpin methods in index theory and spectral flow analysis, linking operator perturbations and quantum invariants through precise trace-based formulations.
• These techniques are widely applied in spectral graph theory, noncommutative geometry, and quantum models to extract geometric and dynamical information from operator spectra.

A trace-based spectral definition refers to any framework in spectral theory, functional analysis, quantum theory, or noncommutative geometry where the trace (or a trace-like functional) is used as a primary tool to define, characterize, or compute spectral invariants, spectral flows, dynamical quantities, or spectral statistics. Such definitions leverage the fact that traces are inherently spectral: they sum (possibly weighted) spectral data, reduce operator-theoretic information to scalars, and are robust under perturbations and invariance properties. Trace-based spectral definitions arise in broad domains, including topological phases, operator algebras, index theory, noncommutative geometry, statistical mechanics of Markov operators, and spectral graph theory.

1. Abstract Trace-Form Evaluators and Axiomatic Frameworks

A foundational approach to trace-based spectral definitions is provided by the theory of trace-form evaluators on classes of self-adjoint operators or spectral geometries. Given a separable Hilbert space $H$, a family $\mathcal{C}$ of bounded self-adjoint operators is selected, subject to:

• closure under the functional calculus (if $X\in\mathcal{C}$ and $g$ is bounded Borel, then $g(X)\in\mathcal{C}$),
• closure under orthogonal sums,
• closure under monotone strong limits.

An evaluator $\mathcal{E}$: $\mathcal{C}\to[0,\infty]$ is a functional satisfying:

• unitary invariance,
• extensivity on orthogonal sums,
• projector-locality,
• dominated-convergence continuity,
• and linear normalization on scalar multiples of the identity.

The main theorem (trace-form representation; (Homer, 12 Dec 2025)) asserts: on the trace-class envelope $\mathcal{C}_1=\mathcal{C}\cap \mathcal{S}_1(H)$, every such evaluator must be of trace form

$\mathcal{E}(f(D)) = c \ \operatorname{Tr}[h(f(D))]$

for some nondecreasing profile $h$ (unique up to affine rescaling), and normalization constant $c>0$. This establishes that any "reasonable" spectral scalar obtained from the spectrum of bounded f(D) is rigidly given by a trace of a profile applied through the functional calculus, revealing a rigidity principle and placing all such trace-based spectral definitions within a unified analytic backbone (Homer, 12 Dec 2025).

2. Trace-Based Index Theory and Spectral Flow

Trace-based spectral definitions play a fundamental role in modern index theory, particularly for paths of operators and non-Fredholm situations. The core object is the spectral shift function (SSF), originally arising from the Lifshitz–Krein trace formula:

$$\operatorname{Tr}\bigl(f(A)-f(A_0)\bigr) = \int_{-\infty}^{+\infty} f'(\lambda)\,\xi(\lambda;A,A_0) d\lambda$$

for suitable function $f$ and self-adjoint $A,A_0$ whose difference is trace-class or satisfies a relative trace-class property (Carey et al., 2016). The spectral shift function $\xi(\lambda;A,A_0)$ encodes how the spectrum shifts under perturbation, and the Fredholm index, spectral flow, or Witten index of a Dirac-type operator $D_A=(d/dt)+A(t)$ is then obtained via trace formulas involving $\xi$:

• For operator pairs with essential spectrum, the Witten index and spectral flow generalize as

$$W_r(D_A) = \lim_{z\to 0^+} z\,\operatorname{Tr}\left[(D_A^*D_A + z)^{-1} - (D_A D_A^* + z)^{-1}\right] = \xi(0;A_+,A_-)$$

tying index theory directly to spectral traces (Carey et al., 2016).

Similarly, trace index in entanglement spectra is defined for free-fermion topological insulators by tracking jumps in the trace of the reduced one-body correlation matrix as an external parameter (momentum or flux) is varied. The trace index counts the number of entanglement eigenvalues crossing a reference level (Fermi level), and is proven to be exactly equal to the bulk Chern number:

$$\mathcal{A}_{U(1)} = \sum_{k_c} \left[ n_A(k_c^+) - n_A(k_c^-) \right] = C_1$$

providing an entanglement-based spectral characterization of topological invariants (Alexandradinata et al., 2011).

3. Trace Formulas in Spectral Theory and Quantum Models

Trace-based formulas are central in both direct and inverse spectral theory, encoding extensive spectral or geometric information. For finite Hermitian matrices, the entire spectrum is encoded in trace-based expansions:

• The eigenvalue counting function can be expressed as

$$\mathcal{N}(\lambda) = \frac{N\lambda - \operatorname{Tr} H}{2\pi} + \frac{N}{2} + \frac{1}{\pi}\Im\sum_{n=1}^{\infty}\sum_{s=0}^{\infty} e^{-i\frac{\pi}{2}s} \frac{\operatorname{Tr} H^s}{s!} n^{s-1} e^{i n\lambda}$$

and equivalent periodic-orbit expansions connect spectral data to combinatorial graph walks (Gnutzmann et al., 2019).

In geometric settings (e.g., hyperbolic manifolds, quantum geometry), the Selberg trace formula equates a spectral sum involving the trace of functions of the Laplacian to a sum over geometric invariants (closed geodesics), providing spectral lower/upper bounds and connections to geometric and topological invariants (Bonifacio et al., 2023). Modern trace-based techniques enable the derivation of sharp spectral gap and systole bounds by optimizing trace formulas over families of test functions.

In quantum field theory and noncommutative geometry, the spectral action principle posits

$S(\Lambda) = \operatorname{Tr} f(|D|/\Lambda)$

with heat-trace or zeta-function expansions expressing $S(\Lambda)$ in terms of spectral (Wodzicki) residues, curvature invariants, and dimension spectrum. Exact and non-asymptotic closed-form expressions for spectral actions on quantum homogeneous spaces are computed using trace-based methods (Eckstein et al., 2013), and canonical traces extend this to a broad class of operators of non-integer order (Paycha, 2010).

4. Trace Formulas in Operator Integrals and Perturbation Theory

Trace-class properties fundamentally delimit the class of spectral definitions for operator integrals and perturbations. For self-adjoint operators with compact resolvent, functionals like $\operatorname{Tr}\,f(D+A)$ (the spectral action) are expanded via Taylor or Gâteaux derivatives, each coefficient written as a sum over products of divided differences evaluated on the spectrum and matrix elements of $A$ (Suijlekom, 2010). This expansion is fully trace-based and determines the behavior of spectral functionals under bounded perturbations.

In higher-order perturbation theory, trace formulas generalize to multiple operator integrals (MOIs), with each term controlled by the trace of multilinear or factorized kernels, and are tightly linked to the existence of absolutely continuous spectral shift measures for various classes of operators (contractions, dissipative, unitary, or self-adjoint pairs) (Chattopadhyay et al., 2024).

The theory of triple operator integrals identifies precise Hilbert-space factorization conditions under which a kernel $\phi$ yields a mapping from Hilbert–Schmidt operators to the trace class. A function $\phi$ gives a trace-class operator integral exactly when it factors as

$$\phi(t_1, t_2, t_3) = \langle a(t_1,t_2), b(t_2,t_3)\rangle_{\mathcal{H}}$$

with suitable $L^\infty$-boundedness (Coine et al., 2017).

5. Trace-Based Spectral Definitions in Graph Theory and Markov Processes

Trace-based spectral definitions underpin algorithms in spectral graph sparsification, where the efficacy of a sparsifier is measured via trace functionals on graph Laplacians. Specifically, the trace reduction metric,

$$\Delta \mathrm{Tr} = \operatorname{Tr}(L_S^{-1} L_G) - \operatorname{Tr}(L_{S'}^{-1} L_G)$$

quantifies the impact of adding edges on the spectral alignment, providing a computationally tractable spectral surrogate for the condition number. This metric is efficiently approximated using sparse matrix techniques and neighborhood truncations, enabling fast, scalable graph sparsification preserving spectral features (Liu et al., 2022).

In Markov chain Monte Carlo and stochastic processes, the spectral gap (the difference between the largest and second-largest eigenvalues) governs convergence rates and is estimated via trace-based quantities:

$$s_k = \sum_{i=0}^\infty \lambda_i^k = \mathrm{Tr}(K^k)$$

where the bounds on the second eigenvalue are constructed directly from these power sums, which admit integral representations suitable for direct Monte Carlo estimation and variance analysis (Qin et al., 2017).

6. Trace-Based Invariants, Dimension Spectrum, and Noncommutative Integrals

Trace-based spectral definitions are essential for defining invariants in noncommutative geometry and operator ideals:

• In noncommutative geometry, spectral triples $(\mathcal{A}, H, D)$ assign dimension spectrum and residues to the poles of functions like $\zeta_P(s)=\operatorname{Tr}(P |D|^{-s})$. The finite part of this zeta regularization defines canonical traces on abstract pseudodifferential operators, independent of the regulator and vanishing on commutators:

$$\tau(A) = \underset{s=0}{\mathrm{fp}}\,\operatorname{Tr}(A|D|^{-s})$$

This construction extends classical notions (Kontsevich–Vishik trace) and supports noncommutative local index theory (Paycha, 2010).

• For operator ideals, traces are characterized as spectral if they depend only on the eigenvalue sequence. Sukochev–Zanin showed that a positive trace is spectral if and only if it is monotone under logarithmic submajorization; in trace-class ideals closed under this order, every trace possesses a Lidskii-type formula, guaranteeing its value is fully determined by the eigenvalues (Sukochev et al., 2013).
• In noncommutative physics and ergodic theory, the Dixmier trace provides a universal, trace-based framework for constructing the noncommutative integral, density-of-states measures, and index invariants. For suitable weight operators $M_w$ and self-adjoint $H$ admitting an averaged density of states, the Dixmier trace formula

$\operatorname{Tr}_\omega(f(H)M_w) = \int f\,d\nu_H$

can be established for strongly disordered systems, open manifolds, or quantum systems (Hekkelman, 27 Jun 2025).

7. Applications: Inverse Spectral Theory, Sum Rules, and Integrable Systems

Trace-based spectral sum rules, often referred to as trace formulas, provide "if and only if" characterizations of coefficient classes for self-adjoint operators (generalized indefinite strings, Krein strings, canonical systems, Dirac operators). Fundamental relative trace formulas relate weighted integrals of coefficients to sums and integrals of spectral data:

$$\int_0^x \!F[w(s),v(s)] ds = \frac{1}{\pi} \int_{\mathbb{R}} K(\xi)\ln|a(\xi,x)|d\xi + \sum_{\kappa\in Z}F_1(\kappa)-\sum_{\eta\in P}F_1(\eta)$$

where $a(\xi,x)$ is a relative perturbation determinant (Wronskian ratio), and the right side encodes both continuous and discrete spectral data. These trace-based characterizations underlie modern Szegő-type theorems and Lieb–Thirring inequalities, identifying the admissible classes of spectral measures for given operator classes, and are crucial in the analysis of integrable flows such as the conservative Camassa–Holm equation (Eckhardt et al., 2023).

---

References:

• "Spectral-Operator Calculus (Part A): Trace-Form Evaluators and Spectral Growth Taxonomy" (Homer, 12 Dec 2025)
• "Trace Index and Spectral Flow in the Entanglement Spectrum" (Alexandradinata et al., 2011)
• "Trace Formulas for a Class of non-Fredholm Operators: A Review" (Carey et al., 2016)
• "Heat trace and spectral action on the standard Podles sphere" (Eckstein et al., 2013)
• "Trace formulas for general Hermitian matrices: Unitary scattering approach and periodic orbits on an associated graph" (Gnutzmann et al., 2019)
• "Which traces are spectral?" (Sukochev et al., 2013)
• "Higher-Order Trace Formulas for Contractive and Dissipative Operators" (Chattopadhyay et al., 2024)
• "When do triple operator integrals take value in the trace class?" (Coine et al., 2017)
• "Perturbations and operator trace functions" (Suijlekom, 2010)
• "Spectral Bounds on Hyperbolic 3-Manifolds: Associativity and the Trace Formula" (Bonifacio et al., 2023)
• "Trace formula for contractions and it's representation in $\mathbb{D}$" (Chattopadhyay et al., 2021)
• "Trace Formulas in Noncommutative Geometry" (Hekkelman, 27 Jun 2025)
• "A Canonical Trace Associated with Certain Spectral Triples" (Paycha, 2010)
• "Pursuing More Effective Graph Spectral Sparsifiers via Approximate Trace Reduction" (Liu et al., 2022)
• "Estimating the spectral gap of a trace-class Markov operator" (Qin et al., 2017)
• "Trace formulas and inverse spectral theory for generalized indefinite strings" (Eckhardt et al., 2023)