Factorization method for near-field inverse scattering problems in elastodynamics

Bo Zhang; Chun Liu; Guanghui Hu; Tao Yin

REVIEW 5 major objections 4 minor 26 references

Factorization method for near-field inverse scattering problems in elastodynamics

T0 review · 5 major / 4 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read This paper proves a factorization of the elastic near-field operator that turns point-source measurements on a sphere into a direct positivity test for membership in a rigid obstacle.

desk verdict A worthwhile elastic-wave extension of the near-field factorization, but the printed outgoing-to-incoming operator is the negative identity, so the central proof doesn't hold as written. read the letter →

arxiv 2505.24288 v1 pith:YW746HYO submitted 2025-05-30 math.AP cs.NAmath.NA

Chun Liu , Guanghui Hu , Tao Yin , Bo Zhang This is my paper · ORCID

classification math.APcs.NAmath.NA MSC 35P2535R3074J25

keywords factorizationmethodnear-fieldinversescatteringelastodynamicsNavierequationrigidobstacleoutgoing-to-incomingoperatorpointsourcesindicatorfunction

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper establishes a direct reconstruction method for the shape and location of a rigid obstacle in a homogeneous isotropic elastic medium, using only scattered near-field data measured on a sphere at one frequency. The central result is a factorization of the outgoing-to-incoming transformed near-field operator as $TN = -\mathcal{G} S^* \mathcal{G}^*$, together with a range criterion: a point $z$ lies inside the obstacle exactly when the spectral indicators $W^a(z)$ and $W(z)$, defined in (3.36)-(3.37), are positive. This turns the inverse problem into a spectral test on measured data, avoiding iterative forward solves and prior geometric information. Numerical experiments in two dimensions reconstruct kite-shaped, star-shaped, and two-component obstacles from data with up to 10% noise, showing the method is computationally usable as well as theoretically justified.

What carries the argument

The load-bearing object is the outgoing-to-incoming operator $T$, defined on the sphere $S_R$ by conjugating the radial multipole coefficients in the vector spherical harmonic expansion of an outgoing elastic wave. Its decisive property is Lemma 3.5: $T(\Pi(\cdot,z)a\big|_{S_R}) = \Pi(\cdot,z)a\big|_{S_R}$ for every source point $z$ inside the ball, so point-source signatures are fixed points of $T$ on the measurement surface and can serve as sampling probes. Multiplying the near-field operator by $T$ produces the symmetric factorization $TN = -\mathcal{G} S^* \mathcal{G}^*$ with $\mathcal{G}=TG$, and the abstract range identity in Lemma 3.11 converts this factorization into the positivity criterion for $W(z)$.

What would settle it

Evaluate $\operatorname{Im}\langle\psi,S\psi\rangle$ for the elastic single-layer operator on smooth closed surfaces with nonzero densities $\psi$ at frequencies where $\omega^2$ is not a Dirichlet eigenvalue; any case with nonnegative imaginary part would break Lemma 3.10 and invalidate the range identity. A second check is to compute $W(z)$ from noiseless synthetic near-field data around a known obstacle and search for any point outside the obstacle where $W(z)>0$.

Watch

Extended reading notes

Core claim

The paper's central claim is that the near-field operator $N$ for time-harmonic elastic scattering by a rigid (Dirichlet) obstacle admits the symmetric factorization $TN = -\mathcal{G} S^* \mathcal{G}^*$, where $T$ is an outgoing-to-incoming operator defined on the spherical measurement surface $S_R$, $S$ is the elastic single-layer operator on the obstacle boundary, and $\mathcal{G}=TG$ is a compact operator with dense range. From this factorization and an abstract range identity, the paper proves that a point $z$ belongs to the obstacle $D$ if and only if the test function $\varphi_z^a = \Pi(\cdot,z)a\big|_{S_R}$ lies in the range of $(TN)_\#^{1/2}$, which is equivalent to positivity of the indicator $W^a(z)$ defined in (3.36). Combining three independent polarization vectors yields $W(z)>0$ exactly for $z\in D$. The paper therefore provides a non-iterative inversion algorithm that recovers the obstacle from near-field measurements at a fixed frequency.

Load-bearing premise

The reconstruction criterion rests on a sign property of the elastic single-layer operator (a boundary integral operator that maps densities on the obstacle surface to elastic potentials), asserted in Lemma 3.10 and cited from earlier work: its imaginary part must be strictly negative on nonzero densities, and it must differ from a coercive self-adjoint operator by a compact term. If that sign or coercivity fails for the Navier operator, the range identity cannot be applied and positivity of $W(z)$ no longer characterizes the obstacle.

Editorial extensions

If this is right

With point sources and receivers on one sphere at a single frequency, the obstacle's location and boundary are recovered by evaluating the scalar indicators $W^a(z)$ or $W(z)$, with no iterative forward solves and no initial guess.
The range criterion is computable: it diagonalizes the finite matrix $(TN)_\# = |\operatorname{Re}(TN)|+|\operatorname{Im}(TN)|$ and applies the standard spectral range test, so the algorithm is direct and data-driven.
The theory is proved for three-dimensional rigid obstacles, and the two-dimensional experiments show the same indicator peaking on the boundary for kite, star, and disconnected two-component obstacles.
Combining several polarization vectors in $W(z)$ recovers boundary portions that a single polarization misses, which matters for disconnected obstacles.
Reconstructions remain recognizable at 5% and 10% relative noise in the reported experiments, indicating tolerance to measurement error.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

A natural next step, not taken in the paper, is to apply the same outgoing-to-incoming-plus-range-identity template to penetrable or fluid-solid elastic obstacles, where the middle operator must be checked for the same sign and coercivity properties.
Because $T$ is constructed from spherical multipoles, the method as written requires a spherical measurement surface; a non-spherical surface would need a new operator $T$ that still fixes the restricted Green tensor in the sampling test.
The numerical evidence that single polarizations can miss boundary parts suggests practical surveys should always record at least two independent polarizations, even though the theorem allows any polarization vector $a\in S^2$.
If the asserted identity $TT^*=T^*T=I$ were weakened, the compactness and dense-range arguments for $\mathcal{G}=TG$ would need to be reworked; a numerical check of the singular values of the truncated OtI matrix at high truncation order would show how much the identity matters in practice.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

The new thing here is real: this is the first attempt to carry the near-field factorization method with an outgoing-to-incoming operator over to the full Navier system, and the numerical reconstructions of kite, star, and two-component obstacles are genuinely encouraging. The authors also correctly credit the acoustic and fluid-solid precedents [12, 25], and the elastic single-layer properties are cited from standard sources rather than invented. So the project is worth taking seriously.

Unfortunately, the central construction as printed collapses. The OtI operator in (3.15) uses the same outgoing matrix An that appears in the field expansion (3.12). Substituting (3.12) into (3.15) gives tilde-u = -u on SR, so T = -I on its domain. That is not an outgoing-to-incoming map; it is just a sign flip. The claim after (3.15) is false, and Lemma 3.5 is also false: the point-source trace is mapped to its negative, not to itself. The proof's observation that E and Bn are real does not help, because An contains the complex Hankel functions. The error is not cosmetic; the missing conjugation is what would actually turn a radiating field into an incoming one.

The problems propagate. Theorem 3.7 asserts H* = TGS; with T = -I that becomes H* = -GS, which already conflicts with the direct single-layer-potential expression for H*. The factorization in Theorem 3.8 and the range-identity application in Theorem 3.12 inherit the sign trouble. At minimum, Theorem 3.12 applies Lemma 3.11 with T = S* while the factorization contains -S*, so the stated justification for the indicator criterion does not go through. The stress-test note gets the main point right.

One small disagreement with the reader's report: the claim TT* = T*T = I in Lemma 3.6 is actually true if T = -I, but it is true for the wrong reason. The real objection is not non-unitarity but that T is the wrong operator, so the unitarity statement is vacuous and the proof structure that relies on T being a genuine incoming map is invalid.

The fix is likely achievable by redefining T with complex conjugation or incoming Hankel functions, and the numerics suggest the underlying method may work. But the repair is not a one-line typo: Lemmas 3.5 through 3.12 and the factorization itself would need to be reworked. I would not cite the paper in its present form. I would, however, send it to a knowledgeable referee rather than desk reject, because the extension is meaningful and the flaw looks repairable. Recommendation: major revision, then reconsider.

Referee Report

5 major / 4 minor

Summary. The manuscript develops a factorization method for the inverse elastic obstacle scattering problem in three dimensions, using near-field data generated by point sources and receivers on a spherical surface. It introduces an outgoing-to-incoming operator T on the sphere, derives a factorization of the near-field operator TN, and proposes an indicator W(z) obtained from a range identity for the factorized operator. Numerical experiments in two dimensions illustrate the reconstruction for kite-shaped, star-shaped, and disconnected obstacles.

Significance. If the central claim were established, the paper would provide a useful extension of the near-field factorization method from acoustics to elastodynamics, with a non-iterative reconstruction from point-source data. The paper is clearly organized and includes concrete numerical experiments with noise stability tests. However, as printed, the mathematical core contains several sign and conjugation errors that invalidate key lemmas, including the definition of the OtI operator, Lemma 3.5, Theorem 3.7, and the application of the range identity. The central theorem is therefore not established in the present form.

major comments (5)

[§3.2, Definition 3.3 and Eq. (3.15)] The operator T defined in (3.14)-(3.15) is the negative identity on Range(G). Since (3.15) uses the same outgoing matrix A_n as the expansion (3.12), substitution gives \tilde u = -u for |x| ≥ R, hence Tφ = -φ for φ ∈ Range(G). In particular, (3.16) reduces to -φ. Consequently the statement after (3.15) that \tilde u is an incoming wave is false; \tilde u is still a Kupradze-radiating solution, and the interpretation of T as an outgoing-to-incoming operator is not valid.
[§3.2, Lemma 3.5] The proof of Lemma 3.5 contains a sign error that is independent of the reality of E and B_n. Inserting (3.28) into (3.26) gives (3.30) with prefactor -i, while the expansion (3.25) has prefactor +i; the computation therefore yields T(Π(·,z)a|SR) = -Π(·,z)a|SR, not +Π(·,z)a|SR. The claimed identity is false as stated, and this affects the proof of Lemma 3.9.
[§3.3, Theorem 3.7] The identity H* = TGS is inconsistent with the definitions. Directly, for x ∈ SR, (H*ψ)(x) = ∫∂D Π(x,y)ψ(y) ds(y) = (Vψ)(x) = (GSψ)(x), because Vψ is the radiating single-layer potential whose boundary trace is Sψ. Thus H* = GS, not TGS. The formula (3.34) in the proof appears to have an incorrect sign and also drops the complex conjugation of A_n(R) when forming the adjoint of (3.33). This error propagates into the subsequent factorization.
[§3.3, Theorems 3.8 and 3.12] The signs in the factorization and in the range-identity application are not consistent. If H* = GS and T = -I, then N = -G S* G* and TN = G S* G*, so the correct factorization is TN = \tilde G S* \tilde G* with \tilde G = TG, i.e., without the minus sign in (3.35). If instead one keeps the paper's H* = -GS, then TN = -\tilde G S* \tilde G*, and Theorem 3.12 should apply the range identity to the middle operator -S*, not to S* as written. In either reading, at least one of (3.35) and the proof of Theorem 3.12 has the wrong sign.
[§3.2, Lemma 3.6 and §4.1] Lemma 3.6 does not support the construction. As printed, TT* = T*T = I holds only because T = -I, which is not an incoming map. If T is replaced by a genuine incoming map involving \overline{A_n}, then TT* = I becomes a nontrivial block identity for \overline{A_n} A_n^{-1} that is neither proved nor evidently true for the non-normal matrices A_n. Moreover, the 2D implementation in §4.1 defines B_n(R) = A_n(R) in (4.42), so the numerical T is again -I; the experiments therefore do not validate the factorization claimed in Theorem 3.8.

minor comments (4)

[§3.1, Theorem 3.2] The compactness of G is only sketched via a trace embedding; please give the standard interior-regularity argument or a precise citation so that the hypotheses of the range identity are fully supported.
[References, [3]] Reference [3] appears to have the wrong volume number; the correct citation is Inverse Problems 17 (2001) 1445-1464, not volume 34.
[§2, notation] The line 'S2 := {x ∈ R3 : |x| = 1} = S1' is a typo: the unit sphere in R3 should be denoted S2, and the use of S1 conflicts with the usual notation for the unit circle in Section 4.
[§3.3, Eq. (3.37)] In (3.37), the summation index j is reused from the eigenfunction expansion in (3.36); please use a different index for the three polarization vectors to avoid confusion.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the obstacle reconstruction criterion is derived from a stated factorization and external operator-theoretic results, not from fitting or self-referential inputs.

full rationale

The derivation chain is not circular. The near-field operator N is defined from the scattering data by (3.6), and the factorization TN = -GS*G* in Theorem 3.8 is obtained by algebraic manipulation of the definitions of G, H, T and the relation N = -GH. The range identity (Lemma 3.11) is invoked as a standard external theorem from Kirsch and Grinberg [17], and its hypotheses are supposed to be verified through the cited single-layer estimates in Lemma 3.10, which the paper attributes to Alves and Kress [1] and Colton and Kress [6]. Those cited results are external mathematical facts, not restatements of the paper's conclusion, and the paper does not fit any parameter to a subset of the data and then rename that fit as a prediction. The final indicator W(z) in (3.36) is a direct spectral criterion applied to the factorized operator; its positivity is claimed to characterize membership in D by Picard's range criterion. The prior near-field factorization papers [12,25] are cited as motivation and as sources for the OtI-operator idea, but the elastodynamic OtI operator and factorization are constructed in this paper rather than imported as the conclusion. Whether the proofs are correct is a separate issue: the skeptic's concerns about the sign of T, Lemma 3.5, and the assertion TT*=T*T=I are mathematical correctness objections, not instances of circular reasoning. No step in the paper reduces by construction to its own input, and no load-bearing premise is justified solely by a self-citation that itself assumes the target result.

Assumptions & free parameters 0 free parameters · 3 assumptions · 0 invented entities

No invented physical entities. The free parameters are numerical truncation and physical constants not fitted to data. The burden sits on the imported properties of the elastic single-layer operator and on the unproved boundedness of the OtI operator.

assumptions (3)

domain assumption The single-layer operator S has sign-definite imaginary part and S - S_i compact, with S_i coercive
Lemma 3.10, cited from [1] and [6]; this is what makes the range identity applicable to the middle operator -S*.
domain assumption ω^2 is not a Dirichlet eigenvalue of -Δ* in D
Lemma 3.10 assumes this; if it fails, S is not an isomorphism and the factorization method may break down.
ad hoc to paper The OtI operator T is bounded and its blocks \overline{A_n} A_n^{-1} are bounded uniformly in n
Asserted in Section 3.2 without proof; the stronger claim T T* = T* T = I appears false, but only boundedness with dense range is needed for the factorization.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Factorization method for near-field inverse scattering problems in elastodynamics." pith.science (2026). https://pith.science/paper/YW746HYO

@misc{pith2026250524288,
  author       = {Pith},
  title        = {Pith review of: Factorization method for near-field inverse scattering problems in elastodynamics},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/YW746HYO}},
  note         = {Machine review of arXiv:2505.24288}
}

read the original abstract

Consider a time-harmonic elastic point source incident on a bounded obstacle which is embedded in an open space filled with a homogeneous and isotropic elastic medium. This paper is concerned with the inverse problem of recovering the location and shape of the obstacle from near-field data generated by infinitely many incident point source waves at a fixed energy. The incident point sources and the receivers for recording scattered signals are both located on a spherical closed surface, on which an outgoing-to-incoming operator is defined for facilitating the factorization of the near-field operator. Numerical examples in 2D are presented to show the validity and accuracy of the inversion algorithm.

Figures

Figures reproduced from arXiv: 2505.24288 by the authors.

**Figure 2.** Reconstruction of the kite-shaped obstacle with d [PITH_FULL_IMAGE:figures/full_fig_p016_2.png] view at source ↗

**Figure 3.** Reconstruction of the star-shaped obstacle with [PITH_FULL_IMAGE:figures/full_fig_p016_3.png] view at source ↗

Figures from the paper (2 more)

**Figure 4.** Figure 4: Reconstruction of the star-shaped obstacle with d [PITH_FULL_IMAGE:figures/full_fig_p017_4.png]

**Figure 5.** Figure 5: Reconstruction of two sound-soft obstacles with d [PITH_FULL_IMAGE:figures/full_fig_p018_5.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

26 extracted references · 26 canonical work pages

[1]

C. J. Alves and R. Kress, On the far-ﬁeld operator in elast ic obstacle scattering, IMA J. Appl. Math. 67 (2002), 1–21

work page 2002
[2]

Ammari, P

H. Ammari, P. Calmon and E. Iakovleva, Elastic wave propa gation in a semi-inﬁnite solid medium, SIAM J. Imaging Sci. 1 (2008), 169–187

work page 2008
[3]

Arens, Linear sampling methods for 2D inverse elastic wave scattering, Inverse Problems 34 (2001), 1445

T. Arens, Linear sampling methods for 2D inverse elastic wave scattering, Inverse Problems 34 (2001), 1445

work page 2001
[4]

G. Bao, G. Hu and J. Sun and T. Yin, Direct and inverse elast ic scattering from anisotropic media, J. Math. Pures Appl. 117 (2018), 263-301

work page 2018
[5]

Cheng and H

Z. Cheng and H. Dong, Uniqueness and reconstruction meth od for inverse elastic wave scattering with phaseless data, Inverse Problems and Imaging 18 (2024), 406-433

work page 2024
[6]

Colton and R

D. Colton and R. Kress, Inverse acoustic and electromagn etic scattering theory(Third Ed), Springer, 2013

work page 2013
[7]

Elschner, G

J. Elschner, G. Hu, Uniqueness and factorization method for inverse elastic scattering with a single incoming wave, Inverse Problems 35 (2019), 094002. 17 (a) α = 0 (b) α = 1 2 π (c) α = 2 3 π (d) Multiple α Figure 5: Reconstruction of two sound-soft obstacles with d iﬀerent α

work page 2019
[8]

Fata and B

N. Fata and B. Guzina, A linear sampling method for near-ﬁ eld inverse problems in elas- todynamics, Inverse Problems 20 (2004), 713–736

work page 2004

Show all 26 references

[9]

J. T. Fokkema and P. M. Berg, Elastodynamic diﬀraction by a periodic rough surface (stress-free boundary), J. Acoust. Soc. Am. 62 (1977), 1095–1101

1977
[10]

H¨ ahner and G

P. H¨ ahner and G. C. Hsiao, Uniqueness theorems in inver se obstacle scattering of elastic waves, Inverse Problems 9 (1993), 525-534

1993
[11]

G. Hu, A. Kirsch, and M. Sini, Some inverse problems aris ing from elastic scattering by rigid obstacles, Inverse Problems 29 (2012), 015009

2012
[12]

G. Hu, J. Yang, B. Zhang and H. Zhang, Near-ﬁeld imaging o f scattering obstacles with the factorization method, Inverse Problems. 30 (2014), 095005

2014
[13]

K. Ito, B. Jin, and J. Zou, A direct sampling method to an i nverse medium scattering problem, Inverse Problems. 28 (2012), 025003

2012
[14]

X. Ji, X. Liu and Y. Xi, Direct sampling methods for inver se elastic scattering problems, Inverse Problems 17 (2018), 035008

2018
[15]

Kirsch, Characterization of the shape of a scatterin g obstacle using the spectral data of the far ﬁeld operator, Inverse Problems

A. Kirsch, Characterization of the shape of a scatterin g obstacle using the spectral data of the far ﬁeld operator, Inverse Problems. 14 (1998), 1489–1512

1998
[16]

Kirsch, The factorization method for Maxwell’s equa tions, Inverse Problems

A. Kirsch, The factorization method for Maxwell’s equa tions, Inverse Problems. 20 (2004), S117–S134. 18

2004
[17]

Kirsch and N

A. Kirsch and N. Grinberg, The Factorization Method for Inverse Problems, Oxford Lecture Series in Mathematics and its Applications. vol 36 (2008), (Oxford: Oxford University Press)

2008
[18]

V. D. Kupradze, Three-dimensional problems of the math ematical theory of elasticity and thermoelasticity, North-Holland, Amsterdam, 1979

1979
[19]

Li and X

P. Li and X. Yuan, Inverse obstacle scattering for elast ic waves in three dimensions, Inverse Probl. Imaging. 13 (2019), 545–573

2019
[20]

P. Li, Y. Wang, Z. Wang and Y. Zhao, Inverse obstacle scat tering for elastic waves, Inverse Problems 32 (2016), 115018

2016
[21]

X. Liu, S. Meng and B. Zhang, Modiﬁed Sampling Method wit h Near Field Measurements, SIAM J. Appl. Math. 82 (2022), 244–266

2022
[22]

Rose and N´ ed´ elec, Elastic wave inverse scattering in nondestructive evaluation, PA- GEOPH 131 (1989), 715–739

J.H. Rose and N´ ed´ elec, Elastic wave inverse scattering in nondestructive evaluation, PA- GEOPH 131 (1989), 715–739

1989
[23]

J. W. C. Sherwood, Direct Elastic Imaging of a Small Incl usion, Proc. Phys. Soc. 131 (1958), 207–219

1958
[24]

Xiang and G

J. Xiang and G. Yan, The uniqueness of the inverse elasti c wave scattering problem based on the mixed reciprocity relation, Inverse Probl. Imaging. 15 (2021), 539–554

2021
[25]

T. Yin, G. Hu, L. Xu and B. Zhang, Near-ﬁeld imaging of obs tacles with the factorization method: ﬂuid–solid interaction, Inverse Problems. 32 (2016), 015003

2016
[26]

J. Yue, M. Li, P. Li and X. Yuan, Numerical solution of an i nverse obstacle scattering problem for elastic waves via the Helmholtz decomposition, Commun.Comput.Phys. 26 (2019), 809-837. 19

2019

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.

[1] [1]

C. J. Alves and R. Kress, On the far-ﬁeld operator in elast ic obstacle scattering, IMA J. Appl. Math. 67 (2002), 1–21

work page 2002

[2] [2]

Ammari, P

H. Ammari, P. Calmon and E. Iakovleva, Elastic wave propa gation in a semi-inﬁnite solid medium, SIAM J. Imaging Sci. 1 (2008), 169–187

work page 2008

[3] [3]

Arens, Linear sampling methods for 2D inverse elastic wave scattering, Inverse Problems 34 (2001), 1445

T. Arens, Linear sampling methods for 2D inverse elastic wave scattering, Inverse Problems 34 (2001), 1445

work page 2001

[4] [4]

G. Bao, G. Hu and J. Sun and T. Yin, Direct and inverse elast ic scattering from anisotropic media, J. Math. Pures Appl. 117 (2018), 263-301

work page 2018

[5] [5]

Cheng and H

Z. Cheng and H. Dong, Uniqueness and reconstruction meth od for inverse elastic wave scattering with phaseless data, Inverse Problems and Imaging 18 (2024), 406-433

work page 2024

[6] [6]

Colton and R

D. Colton and R. Kress, Inverse acoustic and electromagn etic scattering theory(Third Ed), Springer, 2013

work page 2013

[7] [7]

Elschner, G

J. Elschner, G. Hu, Uniqueness and factorization method for inverse elastic scattering with a single incoming wave, Inverse Problems 35 (2019), 094002. 17 (a) α = 0 (b) α = 1 2 π (c) α = 2 3 π (d) Multiple α Figure 5: Reconstruction of two sound-soft obstacles with d iﬀerent α

work page 2019

[8] [8]

Fata and B

N. Fata and B. Guzina, A linear sampling method for near-ﬁ eld inverse problems in elas- todynamics, Inverse Problems 20 (2004), 713–736

work page 2004