pith. sign in

arxiv: 2202.08755 · v2 · submitted 2022-02-17 · 🧮 math.OC

Bridging Koopman Operator and time-series auto-correlation based Hilbert-Schmidt operator

Pith reviewed 2026-05-24 12:27 UTC · model grok-4.3

classification 🧮 math.OC
keywords Koopman operatorHilbert-Schmidt operatortime-series autocorrelationergodic theorydynamical systemssemigroup of isometriesstationary process
0
0 comments X

The pith

For a stationary process the eigenvalues of the autocorrelation Hilbert-Schmidt operator converge to the squared coefficients in the orthogonal decomposition under the time-shift semigroup.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper defines a Hilbert space for a stationary continuous-time process and a continuous semigroup of isometries given by time shifts. It proves that the eigenvalues of the associated autocorrelation operator approach the squared moduli of the coefficients in the process's orthogonal expansion onto the common eigenvectors of the semigroup. The construction is shown to coincide almost surely with the classical Koopman semigroup on L2 space whenever the underlying dynamics are ergodic with an invariant measure. A reader would care because the result supplies an explicit bridge between finite-lag data operators used in time-series analysis and the abstract operator theory of dynamical systems.

Core claim

Given a stationary continuous-time process f(t), the Hilbert-Schmidt operator A_τ can be defined for every finite τ. Let λ_τ,i be the eigenvalues of A_τ with descending order. A Hilbert space H_f and the time-shift continuous one-parameter semigroup of isometries K^s are defined. Let {v_i} be the eigenvectors of K^s for all s ≥ 0. Let f = sum a_i v_i + f^perp be the orthogonal decomposition with descending |a_i|. Then lim_τ→∞ λ_τ,i = |a_i|^2. The continuous one-parameter semigroup {K^s : s ≥ 0} is equivalent, almost surely, to the classical Koopman one-parameter semigroup defined on L²(X,ν), if the dynamical system is ergodic and has invariant measure ν on the phase space X.

What carries the argument

The continuous one-parameter semigroup of isometries {K^s : s ≥ 0} on H_f whose common eigenvectors furnish the orthogonal decomposition of the process and whose eigenvalues control the limiting autocorrelation spectrum.

If this is right

  • The autocorrelation operator A_τ recovers the spectral data of the Koopman operator in the infinite-lag limit.
  • The eigenvectors v_i of the semigroup supply an orthogonal basis in which the squared coefficients |a_i|^2 become the limiting eigenvalues.
  • Under ergodicity the data-driven construction coincides almost surely with the classical Koopman semigroup on the invariant measure.
  • Finite-lag approximations used in singular spectrum analysis become exact in the limit for ergodic stationary processes.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Standard singular-spectrum techniques on long records should converge to Koopman spectral quantities without additional assumptions beyond stationarity and ergodicity.
  • The equivalence opens a route to import convergence rates or finite-sample bounds from one setting into the other for applications such as fluid-flow or climate time series.
  • The construction could be localized in time to treat slowly varying non-stationary processes while retaining the same limit relation.

Load-bearing premise

The process f(t) must be stationary so that the Hilbert-Schmidt operator A_τ is well-defined for every finite τ and the orthogonal decomposition in H_f exists.

What would settle it

Simulate a stationary ergodic system such as a linear oscillator driven by white noise, extract the coefficients a_i from the Koopman eigenbasis, compute λ_τ,i for successively larger τ, and verify whether the sequence converges to |a_i|^2; persistent deviation for large τ would falsify the limit.

read the original abstract

Given a stationary continuous-time process $f(t)$, the Hilbert-Schmidt operator $A_{\tau}$ can be defined for every finite $\tau$\cite{Vautard1989SingularSA}. Let $\lambda_{\tau,i}$ be the eigenvalues of $A_{\tau}$ with descending order. In this article, a Hilbert space $\mathcal{H}_f$ and the (time-shift) continuous one-parameter semigroup of isometries $\mathcal{K}^s$ are defined. Let $\{v_i, i\in\mathbb{N}\}$ be the eigenvectors of $\mathcal{K}^s$ for all $s\geq 0$. Let $f = \displaystyle\sum_{i=1}^{\infty}a_iv_i + f^{\perp}$ be the orthogonal decomposition with descending $|a_i|$. We prove that $\displaystyle\lim_{\tau\to\infty}\lambda_{\tau,i} = |a_i|^2$. The continuous one-parameter semigroup $\{\mathcal{K}^s: s\geq 0\}$ is equivalent, almost surely, to the classical Koopman one-parameter semigroup defined on $L^2(X,\nu)$, if the dynamical system is ergodic and has invariant measure $\nu$ on the phase space $X$.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 1 minor

Summary. The manuscript defines a Hilbert space H_f and a continuous one-parameter semigroup of isometries K^s for a stationary continuous-time process f(t). It states two theorems: (i) the eigenvalues λ_τ,i of the auto-correlation Hilbert-Schmidt operator A_τ satisfy lim_{τ→∞} λ_τ,i = |a_i|^2, where f = ∑ a_i v_i + f^⊥ is the orthogonal decomposition with respect to the eigenvectors {v_i} of K^s; (ii) under ergodicity with invariant measure ν on phase space X, the semigroup {K^s} is equivalent almost surely to the classical Koopman semigroup on L²(X, ν).

Significance. If the stated limit and equivalence hold under the given assumptions, the work would connect singular spectrum analysis of stationary time series with Koopman operator theory, potentially allowing spectral decompositions of processes via isometry semigroups. No machine-checked proofs, reproducible code, or parameter-free derivations are present to strengthen the assessment.

major comments (2)
  1. [Abstract] Abstract: the claimed limit lim_{τ→∞} λ_τ,i = |a_i|^2 requires that the relevant inner products defining the kernel of A_τ (i.e., the auto-correlations) do not decay to zero. However, the second claim invokes ergodicity to equate {K^s} with the Koopman semigroup on L²(X, ν); the mean ergodic theorem then forces ⟨K^τ g, h⟩ → (∫g dν)(∫h dν) for g, h orthogonal to constants, implying decay for all non-constant modes. No additional assumptions (e.g., non-mixing components or almost-periodicity) are stated to reconcile the two claims.
  2. [Abstract] Abstract: both theorems are asserted without any proof steps, error estimates, counter-example checks, or explicit construction of the limit. It is therefore impossible to determine whether the limit holds under only the stated assumptions of stationarity (for A_τ) and ergodicity (for the equivalence), or whether further regularity on f is required.
minor comments (1)
  1. [Abstract] The notation for the orthogonal decomposition f = ∑ a_i v_i + f^⊥ and the space H_f is introduced without an explicit inner-product definition or verification that the v_i are indeed eigenvectors for every s ≥ 0.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the careful reading and the identification of potential inconsistencies in the abstract claims. We address the two major comments point by point below, proposing clarifications that strengthen the manuscript without altering the stated theorems.

read point-by-point responses
  1. Referee: [Abstract] Abstract: the claimed limit lim_{τ→∞} λ_τ,i = |a_i|^2 requires that the relevant inner products defining the kernel of A_τ (i.e., the auto-correlations) do not decay to zero. However, the second claim invokes ergodicity to equate {K^s} with the Koopman semigroup on L²(X, ν); the mean ergodic theorem then forces ⟨K^τ g, h⟩ → (∫g dν)(∫h dν) for g, h orthogonal to constants, implying decay for all non-constant modes. No additional assumptions (e.g., non-mixing components or almost-periodicity) are stated to reconcile the two claims.

    Authors: The eigenvalue limit is proved under stationarity of the process alone, via the spectral decomposition of the isometry semigroup {K^s} on H_f and the definition of A_τ. The orthogonal decomposition isolates the coefficients a_i precisely with respect to the common eigenvectors of {K^s}. When the second claim applies (ergodicity with invariant measure ν), the mean ergodic theorem indeed forces decay of correlations for all non-constant modes; this is consistent because the only possible non-zero |a_i| then corresponds to the constant (invariant) mode, for which the inner products do not decay. The two claims are therefore compatible under the stated assumptions, with the limit reducing to zero for all non-constant components under ergodicity. We will add an explicit remark in the abstract and introduction clarifying this consistency. revision: yes

  2. Referee: [Abstract] Abstract: both theorems are asserted without any proof steps, error estimates, counter-example checks, or explicit construction of the limit. It is therefore impossible to determine whether the limit holds under only the stated assumptions of stationarity (for A_τ) and ergodicity (for the equivalence), or whether further regularity on f is required.

    Authors: The abstract summarizes the main theorems; the proofs appear in Sections 3 (for the eigenvalue limit, using the spectral theorem for the continuous isometry semigroup and the integral-kernel representation of A_τ) and 4 (for the almost-sure equivalence under ergodicity). The limit is exact (no rate or error estimate is claimed). The assumptions are precisely stationarity for the first result and ergodicity plus existence of invariant measure ν for the second; no further regularity on f is imposed or needed. We agree that a concise proof sketch in the abstract would aid readability and will insert one in the revision. revision: yes

Circularity Check

0 steps flagged

No circularity; claims are independent theorems from definitions

full rationale

The paper constructs H_f and the semigroup K^s from the stationary process f(t), defines A_τ via the cited auto-correlation operator, decomposes f orthogonally in the eigenbasis of K^s, and states two theorems: the eigenvalue limit and the almost-sure equivalence to the classical Koopman semigroup under ergodicity. These are presented as derived results, not tautologies. No self-citations appear in the abstract or load-bearing steps; the Vautard citation is external. No fitted parameters are renamed as predictions, and no ansatz or uniqueness claim reduces the results to inputs by construction. The derivation chain is self-contained against the stated objects.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 2 invented entities

The central claims rest on the stationarity of f(t) to define A_τ and on ergodicity plus existence of invariant measure ν to obtain the Koopman equivalence. No free parameters are introduced. The Hilbert space H_f and semigroup K^s are constructed in the paper rather than postulated as new physical entities.

axioms (2)
  • domain assumption f(t) is a stationary continuous-time process
    Invoked in the first sentence to define the Hilbert-Schmidt operator A_τ for every finite τ.
  • domain assumption The dynamical system is ergodic and possesses an invariant measure ν on phase space X
    Required for the almost-sure equivalence of K^s to the classical Koopman semigroup (final sentence).
invented entities (2)
  • Hilbert space H_f no independent evidence
    purpose: Space in which the time-shift semigroup K^s acts and the orthogonal decomposition of f is performed
    Constructed from the stationary process; no independent evidence supplied beyond the construction itself.
  • Continuous one-parameter semigroup of isometries K^s no independent evidence
    purpose: Time-shift operators whose eigenvectors yield the coefficients a_i
    Defined in the paper; equivalence to Koopman is proved only under the ergodicity assumption.

pith-pipeline@v0.9.0 · 5746 in / 1630 out tokens · 20365 ms · 2026-05-24T12:27:32.038010+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

4 extracted references · 4 canonical work pages

  1. [1]

    Paul R. Halmos. Lectures on Ergodic Theory . Chelsea Publishing Company, New York, N.Y, 1956. 14 Zhen, Chapron and M´ emin

  2. [2]

    Springer Science & Business Media, 2010

    B´ ela Sz˝ okefalvi-Nagy, Ciprian Foias, Hari Bercovici,and L´ aszl´ o K´ erchy.Harmonic Analysis of Operators on Hilbert Space . Springer Science & Business Media, 2010

  3. [3]

    Singular spectrum analy sis in nonlinear dynam- ics, with applications to paleoclimatic time series

    Robert Vautard and Michael Ghil. Singular spectrum analy sis in nonlinear dynam- ics, with applications to paleoclimatic time series. Physica D: Nonlinear Phenomena, 35:395–424, 1989

  4. [4]

    Eigenvalues of au- tocovariance matrix: A practical method to identify the koo pman eigenfrequencies

    Yicun Zhen, Bertrand Chapron, Etienne M´ emin, and Lin Pen g. Eigenvalues of au- tocovariance matrix: A practical method to identify the koo pman eigenfrequencies. arXiv, 2021