pith. sign in

arxiv: 1907.08660 · v1 · pith:IT7MYX7Enew · submitted 2019-07-19 · 🧮 math.NA · cs.NA

A parametric Bayesian level set approach for acoustic source identification using multiple frequency information

Zhiliang Deng , Xiaomei Yang , Jiangfeng Huang This is my paper

Pith reviewed 2026-05-24 19:01 UTC · model grok-4.3

classification 🧮 math.NA cs.NA
keywords acoustic source identificationBayesian level setradial basis expansionMetropolis-Hastingsmultiple frequency datainverse problemwell-posedness
0
0 comments X

The pith

A Bayesian level set method with radial basis parameterization reconstructs acoustic source supports from noisy multi-frequency measurements.

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

The paper establishes a parametric Bayesian level set approach for recovering the support of an acoustic source assumed to be piecewise constant. The level set function is expanded in radial basis functions to reduce the problem to finite dimensions, the well-posedness of the posterior is proven, and samples are drawn with the Metropolis-Hastings algorithm whose mean approximates the unknown source. A reader would care because the construction turns an infinite-dimensional inverse problem into a tractable sampling task while preserving geometric flexibility, and the reported tests indicate performance comparable to Matérn random field priors on several example shapes.

Core claim

The unknown acoustic source is formalized as a spatial-dependent piecewise constant function whose support is represented by a level set function. This function is parameterized by radial basis expansion, the well-posedness of the resulting posterior distribution is proven, posterior samples are generated by the Metropolis-Hastings algorithm, and the sample mean is used to approximate the source. Numerical experiments on several shapes show the algorithm is feasible and competitive with the Matérn random field for the acoustic source problem.

What carries the argument

Radial basis expansion of the level set function, which converts the infinite-dimensional Bayesian inverse problem into a finite-dimensional sampling task while encoding the support of the piecewise constant source.

If this is right

  • The posterior distribution exists and is well-defined once the level set is parameterized by radial basis functions.
  • Metropolis-Hastings sampling produces reconstructions of source support from noisy data collected on a remote closed surface at multiple frequencies.
  • The method yields results competitive with those obtained from a Matérn random field prior in the tested geometries.
  • Multiple-frequency information is directly incorporated into the likelihood to aid identification.

Where Pith is reading between the lines

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

  • The same parameterization and sampling strategy could be applied to other inverse problems that seek to recover interfaces or supports from wave data.
  • If the source were allowed to vary continuously inside its support, the level-set representation would have to be replaced by a different forward model.
  • Alternative posterior summaries beyond the sample mean might be examined to improve reconstruction quality.

Load-bearing premise

The unknown source is a spatial-dependent piecewise constant function.

What would settle it

A controlled test in which the true source support is known in advance but the sample-mean reconstruction from the Metropolis-Hastings chain deviates substantially from that support would falsify the claim of effectiveness.

Figures

Figures reproduced from arXiv: 1907.08660 by Jiangfeng Huang, Xiaomei Yang, Zhiliang Deng.

Figure 1
Figure 1. Figure 1: The problem geometry. We display some numerical reconstructions for the case of the source with single medium in [PITH_FULL_IMAGE:figures/full_fig_p013_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: The centers of the RBF function, left: 2161 center points; [PITH_FULL_IMAGE:figures/full_fig_p014_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Lines 1, 2: Whittle Mat´ern random field prior samples [PITH_FULL_IMAGE:figures/full_fig_p014_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: The numerical reconstructions for δ = 0.01. Left: the true domain; Middle: the reconstructions using Mat´ern field prior; Right: the reconstructions using the RBF expansion. References [1] A. Aghasi, M. Kilmer, and E. Miller, Parametric level set methods for inverse problems, SIAM J. Imaging Sciences, 4(2), (2011), 618-650. [2] A. Alzaalig, G. Hu, X. Liu and J. Sun, Fast acoustic source imaging using multi… view at source ↗
Figure 5
Figure 5. Figure 5: The numerical reconstructions with δ = 0.01 for two kinds of mediums. Left: the true domain; Middle: the reconstructions using Mat´ern field prior; Right: the reconstructions using the RBF expansion. (i) Lshape (ii) Single disk (iii) Two disks with the same medium value [PITH_FULL_IMAGE:figures/full_fig_p016_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: The numerical reconstructions with δ = 0.02 using 25 centers in the RBF expansion for single medium case. 16 [PITH_FULL_IMAGE:figures/full_fig_p016_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: The corresponding model data misfit functions [PITH_FULL_IMAGE:figures/full_fig_p017_7.png] view at source ↗
read the original abstract

The reconstruction of the unknown acoustic source is studied using the noisy multiple frequency data on a remote closed surface. Assume that the unknown source is coded in a spatial dependent piecewise constant function, whose support set is the target to be determined. In this setting, the unknown source can be formalized by a level set function. The function is explored with Bayesian level set approach. To reduce the infinite dimensional problem to finite dimension, we parameterize the level set function by the radial basis expansion. The well-posedness of the posterior distribution is proven. The posterior samples are generated according to the Metropolis-Hastings algorithm and the sample mean is used to approximate the unknown. Several shapes are tested to verify the effectiveness of the proposed algorithm. These numerical results show that the proposed algorithm is feasible and competitive with the Mat\'ern random field for the acoustic source problem.

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 / 0 minor

Summary. The manuscript develops a parametric Bayesian level set method for acoustic source reconstruction from noisy multiple-frequency data measured on a remote closed surface. The source is modeled as a spatially dependent piecewise constant function whose support is recovered via a level-set representation; this function is parameterized by a radial-basis expansion to reduce the problem to finite dimensions. The well-posedness of the resulting posterior is proven, posterior samples are generated by Metropolis-Hastings, and the sample mean is used as the reconstruction. Numerical tests on several shapes are presented to demonstrate feasibility and competitiveness with a Matérn random-field approach.

Significance. If the well-posedness proof is rigorous and the numerical comparisons are reproducible, the work supplies a concrete parametric reduction for Bayesian level-set inversion in acoustics. The combination of radial-basis parameterization with Metropolis-Hastings sampling and a proven posterior offers a practical route to shape recovery when the source is known a priori to be piecewise constant.

major comments (1)
  1. [Abstract] Abstract and introduction: the well-posedness proof and the claimed competitiveness both rest on the premise that the unknown source is exactly piecewise constant. This modeling choice is load-bearing for the level-set formalization, the radial-basis parameterization, and the posterior construction; the manuscript should either (i) state the precise regularity assumptions under which the proof holds or (ii) include at least one numerical counter-example with a non-piecewise-constant source to delineate the method’s scope.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the constructive comment on clarifying the scope of the method. We address the major comment below.

read point-by-point responses

  1. Referee: [Abstract] Abstract and introduction: the well-posedness proof and the claimed competitiveness both rest on the premise that the unknown source is exactly piecewise constant. This modeling choice is load-bearing for the level-set formalization, the radial-basis parameterization, and the posterior construction; the manuscript should either (i) state the precise regularity assumptions under which the proof holds or (ii) include at least one numerical counter-example with a non-piecewise-constant source to delineate the method’s scope.

    Authors: The manuscript explicitly assumes the unknown source is a spatially dependent piecewise constant function (as stated in the abstract and Section 1), and the level-set representation, radial-basis parameterization, posterior construction, and well-posedness proof are all developed under this modeling premise. The competitiveness claims are likewise restricted to this class of sources. To address the referee's point, we will revise the abstract and introduction to state the precise regularity assumptions under which the proof holds (piecewise constant sources with finitely many discontinuities, represented exactly via the level-set function). This clarification delineates the method's scope without the need for counter-examples on non-piecewise-constant sources, which lie outside the intended application. revision: yes

Circularity Check

0 steps flagged

No circularity; derivation is self-contained from forward model and explicit modeling assumptions.

full rationale

The paper states its core modeling choice (unknown source as spatial-dependent piecewise constant function, represented via level-set) explicitly in the abstract and proceeds to define the Bayesian posterior from the forward acoustic model plus likelihood in the standard way. Well-posedness is claimed to be proven directly from this setup, and sampling uses the Metropolis-Hastings algorithm on the resulting finite-dimensional parameterization via radial basis functions. No step reduces a claimed prediction or theorem back to a fitted parameter or self-citation by construction; the piecewise-constant premise is an input assumption rather than a derived result. The numerical comparisons are presented as verification, not as the justification for the method itself. This is the normal non-circular pattern for a Bayesian inverse-problem paper.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

Review performed from abstract only; full text unavailable so ledger entries are limited to explicitly stated modeling choices.

axioms (1)
  • domain assumption The unknown source is a spatial-dependent piecewise constant function
    Stated in the abstract; enables level-set representation of the support.

pith-pipeline@v0.9.0 · 5677 in / 1279 out tokens · 43022 ms · 2026-05-24T19:01:09.020434+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

47 extracted references · 47 canonical work pages · 1 internal anchor

  1. [1]

    Aghasi, M

    A. Aghasi, M. Kilmer, and E. Miller, Parametric level set methods for inverse problems, SIAM J. Imaging Sciences, 4(2), (2011), 618-650

    work page 2011

  2. [2]

    Fast acoustic source imaging using multi-frequency sparse data

    A. Alzaalig, G. Hu, X. Liu and J. Sun, Fast acoustic source imaging using multi-frequency sparse data, arXiv:1712.02654

    work page internal anchor Pith review Pith/arXiv arXiv

  3. [3]

    G. Bao, S. Lu, W. Rundell and B. Xu, A recursive algorithm for multi-frequency acoustic inverse source problems, SIAM J. Numer. Anal., 53 (2015), 1608-1628

    work page 2015

  4. [4]

    Beskos, A

    A. Beskos, A. Jasra, E. Muzaffer, A. Stuart, Sequential Monte Carlo methods for Bayesian elliptic inverse problems, Stat Comput, 2015, DOI 10.1007/s11222-015-9556-7

    work page doi:10.1007/s11222-015-9556-7 2015

  5. [5]

    Burger, A level set method for inverse problems, Inverse problems, 17(5): 1327-1355, 2001

    M. Burger, A level set method for inverse problems, Inverse problems, 17(5): 1327-1355, 2001

    work page 2001

  6. [6]

    Bui-Thanh and O

    T. Bui-Thanh and O. Ghattas, An analysis of infinite dimen- sional Bayesian inverse shape acoustic scattering and its nu- 15 (i) Two disks with different medium values (ii) Two different domains with different medium values (iii) Embedded domain with different medium values Fig. 5. The numerical reconstructions with δ = 0.01 for two kinds of mediums. Left: the...

    work page 2014

  7. [7]

    Chada, M

    N. Chada, M. Iglesias, L. Roininen and A. Stuart, Parame- terizations for ensemble Kalman inversion, Inverse Problems 34 (2018) 055009 (31pp)

    work page 2018

  8. [8]

    Cotter, M

    S. Cotter, M. Dashti, J. Robinson and A.M. Stuart, Bayesian inverse problems for functions and applications to fluid me- chanics. Inverse Problems 25, 2009, 115008

    work page 2009

  9. [9]

    Deng and X

    Z. Deng and X. Yang, Convergence of spectral likelihood ap- proximation based on q-Hermite polynomial in Bayesian in- version, Proceedings of the American Mathematical Society, 2019

    work page 2019

  10. [10]

    Dorn and D

    O. Dorn and D. Lesselier, Level set methods for inverse scat- tering, Inverse Problems 22 (2006) R67R131

    work page 2006

  11. [11]

    Dunlop, M

    M. Dunlop, M. Iglesias and A. Stuart, Hierarchical Bayesian level set inversion, Statistics and Computing, November 2017, Volume 27, Issue 6, 1555-1584

    work page 2017

  12. [12]

    Dunlop and A

    M. Dunlop and A. Stuart, The Bayesian formulation of EIT: analysis and algorithms, Inverse Problems and Imaging, 10(4), 2016, 10071036

    work page 2016

  13. [13]

    Eller and N

    M. Eller and N. Valdivia, Acoustic source identification us- ing multiple frequency information, Inverse Problems 25, 115005(20pp), 2009

    work page 2009

  14. [14]

    Griesmaier, Multi-frequency orthogonality sampling for inverse obstacle scattering problems, Inverse Problems, 27 (2011), 085005

    R. Griesmaier, Multi-frequency orthogonality sampling for inverse obstacle scattering problems, Inverse Problems, 27 (2011), 085005

    work page 2011

  15. [15]

    Griesmaier and C

    R. Griesmaier and C. Schmiedecke. A factorization method for multi-frequency inverse source problems with spars far field measurements, SIAM J. Imag. Sci., (2017), to appear

    work page 2017

  16. [16]

    Iglesias, Y

    M. Iglesias, Y. Lu and A. Stuart, A Bayesian Level Set Method for Geometric Inverse Problems, Interfaces and Free Boundaries 18 (2016), 181-217

    work page 2016

  17. [17]

    Iglesias, K

    M. Iglesias, K. Law and A. Stuart, Ensemble Kalman meth- ods for inverse problems, Inverse Problems 29 (2013) 045001 (20pp). 18

    work page 2013

  18. [18]

    Iglesias, A regularizing iterative ensemble Kalman method for PDE-constrained inverse problems, Inverse Problems 32 (2016) 025002 (45pp)

    M. Iglesias, A regularizing iterative ensemble Kalman method for PDE-constrained inverse problems, Inverse Problems 32 (2016) 025002 (45pp)

    work page 2016

  19. [19]

    Iglesias, K

    M. Iglesias, K. Lin, A. Stuart, Well-posed Bayesian geometric inverse problems arising in subsurface flow, Inverse Problems, 30 2014, 114001

    work page 2014

  20. [20]

    K. Ito, K. Kunisch and Z. Li, Level-set function approach to an inverse interface problem, Inverse Problems 17 (2001) 12251242

    work page 2001

  21. [21]

    Jiang and N

    L. Jiang and N. Ou, Bayesian inference using intermediate distribution based on coarse multiscale model for time frac- tional diffusion equation, Multiscale Modeling and Simulation 16 ( 2018), 327-355

    work page 2018

  22. [22]

    Kaipio, E

    J. Kaipio, E. Somersalo, Statistical and Computational In- verse Problems, Springer Verlag, London, 2005

    work page 2005

  23. [23]

    Kaipio, V

    J. Kaipio, V. Kolehmainen, M. Vauhkonen and E. Somersalo, Inverse problems with structural prior information, Inverse Problems, 15(3), 1999, 713729

    work page 1999

  24. [24]

    Kaipio, V

    J. Kaipio, V. Kolehmainen, E. Somersalo and M. Vauhko- nen, Statistical inversion and Monte Carlo sampling methods in electrical impedance tomography, Inverse Problems 16(5), 2000, 1487-1522

    work page 2000

  25. [25]

    van der Kruk, C

    J. van der Kruk, C. Wapenaar, J. Fokkema and P. van den Berg, Improved three-dimensional image reconstruction tech- nique for multi-component ground penetrating radar data, Subsurface Sensing Technologies and Applications, 4(1), 61- 99, 2003

    work page 2003

  26. [26]

    Lindgren and H

    F. Lindgren and H. Rue and J. Lindstr¨ om, An explicit link between Gaussian fields and Gaussian Markov random fields: the stochastic partial differential equation approch, J. R. Statist. Soc. B, 73, (2011), 423-498

    work page 2011

  27. [27]

    D. Liu, A. Khambampati and J. Du, A parametric level set method for electrical impedance tomography, IEEE Transac- tions on medical imaging 37(2), (2018), 451-460

    work page 2018

  28. [28]

    Liu and J

    X. Liu and J. Sun, Reconstruction of Neumann eigenvalues and support of sound hard obstacles, Inverse Problems, 30 (2014), 065011. 19

    work page 2014

  29. [29]

    Lorentzen, K

    R. Lorentzen, K. Flornes and G. Nævdal, History matching channelized reservoirs using the ensemble Kalman filter. SPE Journal 17 (2012), 137151

    work page 2012

  30. [30]

    Lorentzen, G

    R. Lorentzen, G. Nævdal, and A. Shafieirad, Estimating fa- cies fields by use of the ensemble Kalman filter and distance functions applied to shallow-marine environments. SPE Jour- nal 3 (2012), 146158

    work page 2012

  31. [31]

    Osher and R

    S. Osher and R. Fedkiw, The Level Set Method and Dy- namic Implicit Surfaces. Springer, New York (2003). Zbl pre01873119

    work page 2003

  32. [32]

    Osher and A

    S. Osher and A. Sethian, Fronts propagating with curvature- dependent speed: algorithms based on HamiltonJacobi for- mulations. J. Comput. Phys. 79 (1988), 1249. Zbl 0659.65132 MR 89h:80012

    work page arXiv 1988

  33. [33]

    Ping and D

    J. Ping and D. Zhang, History matching of channelized reser- voirs with vector-based level-set. SPE Journal 19 (2014), 514529

    work page 2014

  34. [34]

    Pingen, M

    G. Pingen, M. Waidmann, A. Evgrafov and K. Maute, A parametric level-set approach for topology optimization of flow domains, Struct Multidisc Optim 41 (2010), 117131

    work page 2010

  35. [35]

    Potthast, A study on orthogonality sampling, Inverse Problems, 26 (2010), 074015

    R. Potthast, A study on orthogonality sampling, Inverse Problems, 26 (2010), 074015

    work page 2010

  36. [36]

    Rasmussen and C

    C. Rasmussen and C. Williams, ”Gaussian Processes for Ma- chine Learning”, Adaptive Computation and Machine Learn- ing, The MIT Press, 2006

    work page 2006

  37. [37]

    Roininen, J

    L. Roininen, J. Huttunen and S. Lasanen, Whittle-Mat´ ern priors for Bayesian statistical inversion with applications in electrical impedance tomography, Inverse Problems and Imaging, 8(2), 2014, 561586

    work page 2014

  38. [38]

    Santin and R

    G. Santin and R. Schaback, Approximation of eigenfunctions in kernel-based spaces, Adv Comput Math 42, (2016), 973993

    work page 2016

  39. [39]

    Sun, An eigenvalue method using multiple frequency data for inverse scattering problems, Inverse Problems, 28 (2012), 025012

    J. Sun, An eigenvalue method using multiple frequency data for inverse scattering problems, Inverse Problems, 28 (2012), 025012

    work page 2012

  40. [40]

    Stuart, Inverse problems: A Bayesian perspective, Acta Numerica (2010), 451559

    A. Stuart, Inverse problems: A Bayesian perspective, Acta Numerica (2010), 451559. 20

    work page 2010

  41. [41]

    Santosa, A level-set approach for inverse problems involv- ing obstacles, ESAIM: Control, Optimization and Calculus of Variations, 1: 17-33, 1996

    F. Santosa, A level-set approach for inverse problems involv- ing obstacles, ESAIM: Control, Optimization and Calculus of Variations, 1: 17-33, 1996

    work page 1996

  42. [42]

    Tai and T

    X. Tai and T. Chan, A survey on multiple level set methods with applications for identifying piecewise constant functions, International Journal of Numerical Analysis and Modeling 1 (2004), 2547

    work page 2004

  43. [43]

    Tarantola, Inverse Problem Theory and Methods for Model Parameter Estimation, SIAM, 2005

    A. Tarantola, Inverse Problem Theory and Methods for Model Parameter Estimation, SIAM, 2005

    work page 2005

  44. [44]

    Z. Wang, J. Bardsley, A. Solonen, T. Cui and Y. Marzouk, Bayesian inverse problems with l1 priors: a randomize-then- optimize approch, SIAM J. Sci. Comput., 39(5), 2017, S140- S166

    work page 2017

  45. [45]

    X. Wang, Y. Guo, D. Zhang and H. Liu, Fourier method for recovering acoustic sources from multi-frequency far-field data, Inverse Problems, 33 (2017), 035001

    work page 2017

  46. [46]

    Yang and Z

    X. Yang and Z. Deng, An ensemble Kalman filter approach based on operator splitting for solving nonlinear Hammerstein type ill-posed operator equations, Modern Physics Letters B, 32(28), (2018), 1850335

    work page 2018

  47. [47]

    Zhang and J

    R. Zhang and J. Sun, Efficient finite element method for grat- ing profile reconstruction, J. Comput. Phys., 302 (2015), 405- 419. 21

    work page 2015