pith. sign in

arxiv: 2507.20846 · v2 · submitted 2025-07-28 · 🌌 astro-ph.IM · eess.SP· stat.AP

Precision spectral estimation at sub-Hz frequencies: closed-form posteriors and Bayesian noise projection

Lorenzo Sala , Stefano Vitale This is my paper

Pith reviewed 2026-05-19 03:33 UTC · model grok-4.3

classification 🌌 astro-ph.IM eess.SPstat.AP
keywords spectral estimationBayesian inferenceWishart distributioncross-spectral densitynoise projectionLISA Pathfindergravitational experimentscoherence estimation
0
0 comments X

The pith

Bayesian analysis of Wishart periodograms yields closed-form posteriors for cross-spectral quantities at any sub-Hz frequency.

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

The paper develops a Bayesian method for spectral estimation in long-duration multivariate time series where limited averaging prevents Gaussian approximations. Modeling the periodogram as Wishart-distributed under Gaussian assumptions allows exact integration against priors to obtain closed-form marginal posteriors for power spectral densities, pairwise coherence, multiple coherence, and the joint distribution of the full cross-spectral density matrix. The approach further supplies closed-form posteriors for susceptibilities and background noise power in linear noise-projection models. Application to LISA Pathfinder data demonstrates effective removal of temperature-induced acceleration noise and reliable recovery of its coupling strength.

Core claim

For multivariate Gaussian time series the periodogram at a fixed frequency follows a Wishart distribution; exact integration of this likelihood against suitable priors on the cross-spectral density matrix produces closed-form expressions for the marginal posteriors of individual power spectral densities, pairwise and multiple coherences, and the joint posterior of the full matrix, plus analogous closed forms for susceptibilities and residual background PSD in a linear noise-projection setting.

What carries the argument

Exact analytic marginalization of the Wishart likelihood for the periodogram against conjugate or chosen priors on the cross-spectral density matrix.

If this is right

  • Spectral quantities and their uncertainties can be computed exactly even when only a single long stretch of data is available.
  • Coherence and noise-coupling estimates carry rigorous Bayesian credible intervals without large-sample approximations.
  • Noise-projection filters and residual backgrounds are inferred jointly with the full spectral matrix.
  • The same closed-form expressions apply at every frequency bin independently.

Where Pith is reading between the lines

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

  • The closed forms could support sequential updating as new data segments arrive in real-time monitoring.
  • Extensions to mildly non-stationary series might be obtained by allowing the prior parameters to vary slowly with time.
  • Similar exact posteriors could be derived for other matrix-variate likelihoods that admit conjugate integration.

Load-bearing premise

The underlying time series are multivariate Gaussian.

What would settle it

A direct comparison showing that the observed histogram of periodogram matrices from the data deviates systematically from the Wishart shape expected under the Gaussian model.

Figures

Figures reproduced from arXiv: 2507.20846 by Lorenzo Sala, Stefano Vitale.

Figure 1
Figure 1. Figure 1: Cumulative density function cdf of the posterior [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Plot of p(S|Π) for S|Π ∼ invΓ(M, MΠ), for a few different values of M. For the sake of clarity, the PDF is for the ratio S/Π. method that accounts for the fact that, in Gaussian statistics, t = (St − Π)/(sΠ/ √ M) follows a Student’s t￾distribution with M − 1 degrees of freedom. Thus, in the calculation of the equal-tail credible intervals, it would be more accurate to replace −k by the ℓ 2 -quantile of the… view at source ↗
Figure 4
Figure 4. Figure 4: the ℓ(2)(≃ 0.95) likelihood, equal tail credible interval, and the median predicted by the posterior in Eq. (26). We do that as a function of both |ρˆ| 2 and M. The figure shows that one reaches a reasonable confidence that some correlation exists between the two processes, only when both M and |ρˆ| 2 are large enough. For instance, this confidence is never reached for M = 2, only if |ρˆ| 2 ≳ 0.6 for M = 5… view at source ↗
Figure 5
Figure 5. Figure 5: Contour plot of the probability density function  [PITH_FULL_IMAGE:figures/full_fig_p006_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: The equal tail, ℓ(2)(≃ 0.95.5) likelihood credible intervals for the multiple coherence R2 (error bars) as a function of the observed value of Rˆ2 and of the number of averaged periodograms M. The calculation is for p = 5- variate stochastic process. The central dots are the values of the median. The dashed line R2 = Rˆ2 is given for reference. For the sake of clarity, for different values of M we plot R2 … view at source ↗
Figure 7
Figure 7. Figure 7: ASD of time series used in the simulation. The [PITH_FULL_IMAGE:figures/full_fig_p010_7.png] view at source ↗
Figure 9
Figure 9. Figure 9: Black data points: susceptibility α1(f) of x(t) to y1(t) for the example data of [PITH_FULL_IMAGE:figures/full_fig_p011_9.png] view at source ↗
Figure 11
Figure 11. Figure 11: The joint posterior for the three susceptibilities [PITH_FULL_IMAGE:figures/full_fig_p012_11.png] view at source ↗
read the original abstract

We consider the problem of estimating cross-spectral quantities in the low-frequency regime, where long observation times limit averaging over large ensembles of periodograms, thereby preventing the use of approximate Gaussian statistics. This case is relevant for precision low-frequency gravitational experiments such as LISA and LISA Pathfinder. We present a Bayesian method for estimating spectral quantities in multivariate Gaussian time series. The approach, based on periodograms and Wishart statistics, yields closed-form expressions at any given frequency for the marginal posterior distributions of the individual power spectral densities, the pairwise coherence, and the multiple coherence, as well as for the joint posterior distribution of the full cross-spectral density matrix. In the context of noise projection -- where one series is modeled as a linear combination of filtered versions of the others, plus a background component -- the method also provides closed-form posteriors for both the susceptibilities, i.e., the filter transfer functions, and the power spectral density of the background. We apply the method to data from the LISA Pathfinder mission, showing effective decorrelation of temperature-induced acceleration noise and reliable estimation of its coupling coefficient.

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

1 major / 2 minor

Summary. The manuscript proposes a Bayesian method for estimating cross-spectral quantities (PSDs, coherences, and the full cross-spectral density matrix) in multivariate Gaussian time series at sub-Hz frequencies. It exploits the Wishart distribution of single-frequency periodograms together with conjugate inverse-Wishart priors to obtain closed-form marginal posteriors; the framework is extended to a linear noise-projection model that yields closed-form posteriors for susceptibilities and residual background PSD. The method is demonstrated on LISA Pathfinder data, where it is used to decorrelate temperature-induced acceleration noise.

Significance. If the conjugacy arguments hold exactly, the work supplies a computationally efficient, approximation-free route to Bayesian spectral inference precisely where long observation times preclude ensemble averaging. This is directly relevant to LISA and similar precision gravitational experiments. The explicit closed-form marginals for coherence and the matrix-variate posterior constitute a clear technical advance over MCMC-based alternatives, and the LISA Pathfinder application provides a concrete, falsifiable demonstration of noise-projection performance.

major comments (1)
  1. [§3.2] §3.2 (noise-projection extension): the claim that the joint posterior for susceptibilities and residual PSD remains inverse-Wishart after marginalization over the linear filter coefficients is load-bearing for the closed-form assertion. The manuscript should display the explicit integration step (or the resulting matrix-variate marginal) to confirm that frequency-dependent filtering does not break conjugacy or introduce hidden approximations.
minor comments (2)
  1. [Abstract and §2.1] The abstract and §2.1 should state the precise form of the inverse-Wishart scale matrix prior (e.g., whether it is data-independent or uses a small number of pseudo-observations) so that readers can immediately reproduce the marginal expressions.
  2. [Figure 3] Figure 3 (LISA Pathfinder posterior plots): the frequency axis and the number of independent periodogram segments used should be stated in the caption to allow direct comparison with the theoretical Wishart degrees of freedom.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for their careful reading of the manuscript and for the constructive comment on the noise-projection extension. We address the point below and will revise the manuscript to incorporate the requested clarification.

read point-by-point responses

  1. Referee: [§3.2] §3.2 (noise-projection extension): the claim that the joint posterior for susceptibilities and residual PSD remains inverse-Wishart after marginalization over the linear filter coefficients is load-bearing for the closed-form assertion. The manuscript should display the explicit integration step (or the resulting matrix-variate marginal) to confirm that frequency-dependent filtering does not break conjugacy or introduce hidden approximations.

    Authors: We agree that the explicit marginalization step should be shown to confirm the closed-form result. In the revised manuscript we will add the derivation in §3.2. The noise-projection model is linear in the frequency-domain susceptibilities at each frequency bin. With the Wishart likelihood for the periodogram and the conjugate inverse-Wishart prior on the cross-spectral density matrix, the joint posterior for the susceptibilities and residual background PSD follows from standard multivariate conjugate updating. Marginalizing the susceptibilities (which enter as regression coefficients) produces an inverse-Wishart marginal for the residual PSD whose scale matrix is the residual sum of squares after projection onto the orthogonal complement of the filter space. Because the analysis is performed independently at each frequency, frequency-dependent filtering introduces no cross-frequency terms and preserves exact conjugacy under the Gaussian assumption; no hidden approximations are required. revision: yes

Circularity Check

0 steps flagged

No significant circularity; derivation follows directly from Wishart conjugacy

full rationale

The paper's central results are closed-form marginal posteriors for PSDs, coherences, and the cross-spectral matrix obtained by integrating the Wishart likelihood (arising from multivariate Gaussian time series) against an inverse-Wishart prior. This is a standard conjugate update whose marginals (inverse-gamma, beta, matrix-variate) are known independently of the present work. The noise-projection extension treats susceptibilities as linear regression coefficients and the residual as a separate Wishart term, again preserving exact conjugacy and closed-form marginals. No parameter is fitted to a data subset and then relabeled as a prediction, no self-citation supplies a load-bearing uniqueness theorem, and no ansatz is smuggled in. The derivation is therefore self-contained against external mathematical benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central claim rests on the assumption that the time series are multivariate Gaussian so that the periodogram matrix follows a complex Wishart distribution; no free parameters or new entities are introduced in the abstract.

axioms (1)
  • domain assumption The observed time series are jointly multivariate Gaussian.
    Invoked to justify the Wishart distribution of the periodogram matrix at each frequency.

pith-pipeline@v0.9.0 · 5726 in / 1363 out tokens · 27795 ms · 2026-05-19T03:33:11.859932+00:00 · methodology

discussion (0)

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

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

21 extracted references · 21 canonical work pages

  1. [1]

    The use of fast Fourier transform for the esti- mation of power spectra: a method based on time averaging over short, modified periodograms,

    P. D. Welch, “The use of fast Fourier transform for the esti- mation of power spectra: a method based on time averaging over short, modified periodograms,” IEEE Trans. Audio and Electroacoustics, vol. 15, no. 2, pp. 70–73, 1967

    work page 1967

  2. [2]

    Papoulis and S

    A. Papoulis and S. Pillai, Probability, Random Variables, and Stochastic Processes, ser. McGraw-Hill series in electrical engi- neering: Communications and signal processing. Tata McGraw- Hill, 2002

    work page 2002

  3. [3]

    In-depth analysis of LISA Pathfinder perfor- mance results: Time evolution, noise projection, physical models, and implications for LISA,

    M. Armano et al. , “In-depth analysis of LISA Pathfinder perfor- mance results: Time evolution, noise projection, physical models, and implications for LISA,” Phys. Rev. D , vol. 110, p. 042004, Aug 2024

    work page 2024

  4. [4]

    Beyond the Required LISA Free-Fall Performance: New LISA Pathfinder Results down to 20 µHz,

    ——, “Beyond the Required LISA Free-Fall Performance: New LISA Pathfinder Results down to 20 µHz,” Phys. Rev. Lett. , vol. 120, p. 061101, 2 2018

    work page 2018

  5. [5]

    Sub-Femto- g Free Fall for Space-Based Gravitational Wave Observatories: LISA Pathfinder Results,

    ——, “Sub-Femto- g Free Fall for Space-Based Gravitational Wave Observatories: LISA Pathfinder Results,” Phys. Rev. Lett. , vol. 116, p. 231101, 2016

    work page 2016

  6. [6]

    Nano-Newton electrostatic force actuators for femto- Newton-sensitive measurements: System performance test in the LISA Pathfinder mission,

    M. Armano, H. Audley, J. Baird, M. Bassan, P. Binetruy, M. Born, D. Bortoluzzi, E. Castelli, A. Cavalleri, A. Cesarini et al. , “Nano-Newton electrostatic force actuators for femto- Newton-sensitive measurements: System performance test in the LISA Pathfinder mission,” Phys. Rev. D , vol. 109, p. 102009, May 2024

    work page 2024

  7. [7]

    Improved spectrum estimation from digitized time series on a logarithmic frequency axis,

    M. Tr¨ obs and G. Heinzel, “Improved spectrum estimation from digitized time series on a logarithmic frequency axis,” Measurement, vol. 39, pp. 120–129, 02 2006

    work page 2006

  8. [8]

    An optimised variant of the periodogram method for noise power spectral density estimation in LISA hardware testing,

    S. Vitale, L. Sala, and D. Vetrugno, “An optimised variant of the periodogram method for noise power spectral density estimation in LISA hardware testing,” University of Trento, Tech. Rep. LISA-UTN-INST-TN-0035, 2025

    work page 2025

  9. [9]

    Statistical Analysis Based on a Certain Multivariate Complex Gaussian Distribution (An Introduction),

    N. R. Goodman, “Statistical Analysis Based on a Certain Multivariate Complex Gaussian Distribution (An Introduction),” The Annals of Mathematical Statistics , vol. 34, no. 1, pp. 152–177, 03 1963

    work page 1963

  10. [10]

    Expectations of Functions of Complex Wishart Matrix,

    D. K. Nagar and A. K. Gupta, “Expectations of Functions of Complex Wishart Matrix,” Acta Applicandae Mathematicae, vol. 113, no. 3, pp. 265–288, 03 2011

    work page 2011

  11. [11]

    Estimation of the magnitude-squared coherence function via overlapped fast Fourier transform processing,

    G. Carter, C. Knapp, and A. Nuttall, “Estimation of the magnitude-squared coherence function via overlapped fast Fourier transform processing,” IEEE Transactions on Audio and Electroacoustics, vol. 21, no. 4, pp. 337–344, 08 1973

    work page 1973

  12. [12]

    Schur complements and statistics,

    D. V. Ouellette, “Schur complements and statistics,” Linear Algebra and its Applications , vol. 36, pp. 187–295, 03 1981

    work page 1981

  13. [13]

    C. R. Rao, Ed., Linear Statistical Inference and its Applications , ser. Wiley Series in Probability and Statistics. Hoboken, NJ, USA: John Wiley & Sons, Inc., 1973

    work page 1973

  14. [14]

    An invariant form for the prior probability in estimation problems,

    H. Jeffreys, “An invariant form for the prior probability in estimation problems,” Proceedings of the Royal Society of London. Series A. Mathematical and Physical Sciences , vol. 186, no. 1007, pp. 453–461, 1946

    work page 1946

  15. [15]

    The inverted complex Wishart distribution and its application to spectral estimation,

    P. Shaman, “The inverted complex Wishart distribution and its application to spectral estimation,” J. Multivar. Anal. , vol. 10, no. 1, pp. 51–59, 03 1980

    work page 1980

  16. [16]

    A. M. Mathai, S. B. Provost, and H. J. Haubold, Multivariate statistical analysis in the real and complex domains . Springer Nature, 2022

    work page 2022

  17. [17]

    On posterior distributions for signals in gaussian noise with unknown covariance matrix,

    L. Svensson and M. Lundberg, “On posterior distributions for signals in gaussian noise with unknown covariance matrix,”IEEE Transactions on Signal Processing , vol. 53, no. 9, pp. 3554–3571, 2005

    work page 2005

  18. [18]

    The Multivariate t-Distribution Associated with a Set of Normal Sample Deviates,

    E. Coornish, “The Multivariate t-Distribution Associated with a Set of Normal Sample Deviates,” Australian Journal of Physics , vol. 7, no. 4, p. 532, 1954

    work page 1954

  19. [19]

    Some windows with very good sidelobe behavior,

    A. H. Nuttall, “Some windows with very good sidelobe behavior,” IEEE Transactions on Acoustics, Speech, and Signal Processing , vol. 29, no. 1, pp. 84–91, 1981

    work page 1981

  20. [20]

    Complex singular Wishart matrices and applications,

    T. Ratnarajah and R. Vaillancourt, “Complex singular Wishart matrices and applications,” Computers & Mathematics with Applications, vol. 50, no. 3, pp. 399–411, 2005

    work page 2005

  21. [21]

    Is the rank of the sum of two positive semi-definite matrices larger than their individual ranks?

    S. M. S. Exchange), “Is the rank of the sum of two positive semi-definite matrices larger than their individual ranks?” Mathematics Stack Exchange, (version: 2017-02-20). [Online]. Available: https://math.stackexchange.com/q/2153772

    work page arXiv 2017