Pith. sign in

REVIEW 4 major objections 5 minor 34 references

Multi-frequency far-field data enrichment for electromagnetic source reconstruction

T0 review · 4 major / 5 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read The paper claims that missing multi-frequency far-field data for compact, geometrically sparse electromagnetic sources can be completed by low-rank Hankel matrix completion before Fourier inversion, eliminating under-sampling artifacts…

desk verdict Joint 3D ALOHA for EM source reconstruction is a legitimate new application, but the headline claim about resolving non-uniqueness is not supported by the current experiments because the DC mode is taken from ground truth. read the letter →

arxiv 2608.04829 v1 pith:ITL6MXFH submitted 2026-08-05 math-ph cs.ITmath.ITmath.MPmath.OC

Atyab Khalifa Al-Shaqsi , Heba Mohammed Al-Subhi , Xianchao Wang , Shujaat Khan , Abdul Wahab This is my paper · ORCID
classification math-phcs.ITmath.ITmath.MPmath.OC MSC 35R3078A46
keywords electromagneticinversesourceproblemfiniterateofinnovationsHankelmatrixcompletionALOHAmulti-frequencyfar-fielddatalow-rankFourierinversion
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

The paper tries to establish that the non-uniqueness caused by under-sampled multi-frequency far-field measurements can be tamed by first completing the missing Fourier spectrum, rather than by regularizing the inversion directly. It claims that compactly supported, geometrically sparse current sources have a finite rate of innovations, which makes a wrap-around block Hankel matrix built from the measured spectral coefficients low-rank. Recovering the missing coefficients by low-rank matrix completion and then applying Fourier inversion yields artifact-free reconstructions at 30% to 50% sampling and 10 dB noise, outperforming a DCT-based $\ell^1$-CS baseline by 8 to 18 dB PSNR in paired Monte Carlo trials. A sympathetic reader would care because it offers a practical route to sparse-sensor electromagnetic imaging in biomedical, non-destructive testing, and antenna synthesis applications.

What carries the argument

The load-bearing object is the block-diagonal, wrap-around Hankel matrix $\mathbf{H}_p(\mathbf{s})$ built from the concatenated transverse electric and magnetic Fourier coefficients $f_\ell$ and $g_\ell$, with $\mathbf{s}=[f^T,g^T]^T$. Its low rank follows from the finite-rate-of-innovations property: a compactly supported sparse source admits an annihilating filter, and the rank-deficiency of the wrap-around Hankel matrix is what turns missing-data recovery into a nuclear-norm minimization problem. The minimization is solved by ALOHA, the annihilating-filter low-rank Hankel matrix completion algorithm, via ADMM on a factorized representation, and the completed spectrum is fed into the Fourier inversion formula to reconstruct the current density.

What would settle it

Apply the same Joint 3D ALOHA pipeline to a source deliberately containing more than 30 well-separated geometric innovations (so the true FRI rank exceeds the fixed rank 30) at 30% sampling in noise-free conditions; if the PSNR advantage over $\ell_1$-CS shrinks toward zero or reverses, the claim that low-rank completion resolves non-uniqueness is limited to sources whose rank is known and small.

Watch

Extended reading notes

Core claim

The central claim is that the inverse problem of recovering a current source density from sparse multi-frequency far-field data becomes tractable if the missing spectrum is treated as a finite-rate-of-innovations signal and completed by low-rank Hankel matrix completion before Fourier inversion. The paper derives Fourier formulas linking electric and magnetic far-field data to the transverse electric and magnetic scalar components of the source, constructs a block-diagonal wrap-around Hankel matrix from those spectral components, and argues its low rank follows from the finite rate of innovations. Joint 3D ALOHA completes the missing coefficients, and the enriched Nyquist-sampled spectrum is then fed into the Fourier inversion formula. Numerical experiments on two 3D Maxwell source models with 30% to 50% conjugate-paired sampling and 10 dB AWGN show paired PSNR gains of 8.02 to 18.18 dB and SSIM gains of 0.1842 to 0.3451 over the $\ell^1$-CS baseline, with every trial won by the proposed method.

Load-bearing premise

The load-bearing premise is that the source is compactly supported and geometrically sparse enough that the wrap-around Hankel matrix has a known low rank; the paper fixes the ranks at 22 and 30 manually, so if the true FRI rank is unknown or larger, completion degrades.

Editorial extensions

If this is right

  • At 30% to 50% sampling, the completed spectrum yields stable reconstructions where direct zero filling and DCT-based $\ell_1$-CS leave heavy ringing and speckle.
  • The paired gains over $\ell_1$-CS range from 8.02 to 18.18 dB PSNR and 0.1842 to 0.3451 SSIM across all twelve tested source-noise-sampling conditions.
  • The reconstruction preserves the full three-dimensional source topology, not just central slices, as shown by consistent isosurface thresholds.
  • The method handles 10 dB complex Gaussian noise by relaxing the data-consistency weight, and it runs about 1.74 to 2.51 times faster than the implemented $\ell_1$-CS solver.
  • Accurate recovery of the zero-frequency coefficient, the mode most distorted by subsampling, is a direct corollary of completing the spectrum before inversion.

Reading between the lines

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

  • Editorial inference: replacing the manually fixed ranks (22 for J1, 30 for J2) with an automatic rank estimator, such as singular-value thresholding or cross-validation on the observed coefficients, should extend the method's applicability to sources whose FRI order is not known in advance.
  • Editorial inference: because the enrichment is a preprocessing stage rather than a new inversion operator, it should compose with other reconstruction schemes; a natural test is to run the same completed spectrum through TV-regularized or iterative refinement solvers and check whether the 8-18 dB gain persists.
  • Editorial inference: the same wrap-around Hankel structure should transfer to other inverse problems in which compact sparse sources are observed through Fourier-type data, such as acoustic source reconstruction or MEG/EEG source localization; the paper itself only validates the Fourier-Maxwell forward model, so this remains an open test.
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.

Referee Report

4 major / 5 minor

Summary. The paper proposes a two-stage strategy for reconstructing a compactly supported electromagnetic current source from sparse multi-frequency magnetic far-field data. In stage one, the missing Fourier coefficients of the transverse electric and magnetic components are completed by solving a low-rank Hankel matrix completion problem using the ALOHA algorithm; stage two applies a Fourier inversion formula to synthesize the current density. The method is tested on two synthetic 3D source models (J1 and J2) at 30%, 40%, and 50% sampling rates, in noise-free and 10 dB AWGN conditions, and compared with zero-filled reconstruction and a complex DCT-based l1-CS baseline. The reported results show large paired PSNR and SSIM gains for Joint 3D ALOHA, with all paired comparisons significant after Holm correction. The paper also includes runtime measurements and a limitations discussion.

Significance. If the claims were fully supported, the paper would provide a useful computational tool for inverse source problems with sparse far-field data, extending low-rank Hankel completion to vector electromagnetic sources. The manuscript has genuine strengths: the statistical protocol is unusually careful (30 Monte Carlo trials, nested masks, paired Wilcoxon tests with Holm correction, bootstrap confidence intervals, and effect sizes), the runtime benchmark is clearly delimited, the limitations section is candid, and the proposed joint completion of the three vector-current components is a sensible use of shared support. However, the experimental validation has a load-bearing oracle: the zero-frequency Fourier coefficient f0 is taken from the exact source (Section 4.1), and the measurement procedure never requires the completion method to recover it. Since f0 is precisely the type of global non-radiating null-space component the abstract claims to resolve, the headline 'resolves non-uniqueness challenges' is not tested by Tables 1-3. The same-model data generation and per-source rank selection further weaken the external validity of the reported gains.

major comments (4)
  1. [Section 4.1 and Section 4.2] The zero-frequency coefficient f0 is taken from the exact source, not reconstructed from the sparse measurements. Section 4.1 states that 'f0 is recovered from the mean projection of the exact current onto the polarization direction,' and Section 4.2 says 'the origin is retained in every reconstruction' so the random mask never removes f0. Consequently, the completion algorithm is never asked to recover the most global non-radiating component, and the abstract's claim that the framework 'resolves non-uniqueness challenges' is not supported by Tables 1-3. Please rerun the experiments with f0 estimated from the observed far-field data (e.g., via Eq. (9) or a data-driven estimate) or with f0 set to zero, and report both variants; alternatively, explicitly weaken the non-uniqueness claim to 'non-zero-frequency missing modes'.
  2. [Sections 4.1 and 5.6] The same Fourier-Maxwell forward model is used both to generate the synthetic far-field data and to define the reconstruction formula, which is an inverse-crime setup. The paper acknowledges this in Section 5.6 ('both use the same Fourier-Maxwell forward model employed to construct the reconstruction data'), but the quantitative PSNR/SSIM gains are still likely to overstate performance under model mismatch. Please add at least one validation with a different forward solver (e.g., a boundary-element or finite-difference Maxwell solve) or with perturbed model parameters (wavenumber, polarization, domain size) to demonstrate that the enrichment is not an artifact of using the same spectral model for both data generation and inversion.
  3. [Section 4.3.3] The ALOHA rank is fixed manually and source-specifically: ranks 22 for J1 and 30 for J2. Although the authors correctly note that the ranks are not changed across sampling rates, noise conditions, or trials, choosing a different rank for each source encodes prior knowledge of that source's complexity. Since the FRI rank is not estimated from the data, the claim that the method automatically exploits a finite rate of innovations is overstated. Please add a rank-sensitivity study (e.g., PSNR/SSIM versus rank for both sources) or an automatic rank-selection procedure, and discuss how the results would change if the rank were misspecified.
  4. [Section 5.3 and Tables 2-3] The statistical comparison is internally sound, but the reference for PSNR/SSIM is the fully sampled Fourier reconstruction, which is generated with the same forward model used for data synthesis. This is appropriate for isolating completion error, but it does not measure fidelity to an independent ground truth. Please clarify in the text that the reported absolute PSNR/SSIM values are relative to a self-consistent numerical reference and cannot be directly compared with experimental imaging benchmarks.
minor comments (5)
  1. [Section 3.3, Eq. (12)] The notation in Eq. (12) is inconsistent: the vector es is declared as belonging to real 2M-dimensional space, but the Fourier coefficients f_m and g_m are complex-valued; the correct space is complex 2M-dimensional space.
  2. [Section 3.3, Eq. (13) area] There is a missing parenthesis in the sentence about the rank of H_p(es), and the phrase 'if p(< M) is chosen larger' is awkward; please rewrite as 'if p < M is chosen larger than the minimum filter size'.
  3. [Section 4.1] The phrase 'The magnetic far-field data comprise of 9260 Fourier modes' should read 'comprise 9260 Fourier modes'.
  4. [Section 5.6] The limitations paragraph would be strengthened by an explicit statement that no experimental or independently simulated far-field data were used, and by a note on whether the code and data will be made available for reproducibility.
  5. [Throughout] Some figure captions contain hyphenated phrases such as 'electromagnetic-sources' and 'Sub-sampled' in tables; please standardize the hyphenation and use 'subsampled' consistently.

Circularity Check

1 steps flagged · score 6.0 of 10

Oracle f0 is inserted from the exact source into Eq. (7), so the headline 'resolves non-uniqueness' is untested for the DC null-space mode.

  1. fitted input called prediction [Section 4.1 (Source Models and Nyquist-Sampled Reference); used in Eq. (7) and evaluated in Tables 1–3]
    "The magnetic far-field data comprise of 9260 Fourier modes other than the zero-frequency coefficient f0 which is recovered from the mean projection of the exact current onto the polarization direction."

    In formula (7), F(x) includes the \hat p f_0 term, so the reconstruction contains the true zero-frequency coefficient of the transverse-electric component. Section 4.2 defines the sparse masks over the 9260 nonzero modes and retains the origin, so ALOHA never completes f0 from the sparse far-field data. Section 3.2 identifies f0 as a major source of artifacts and non-uniqueness ('inaccuracies in estimating the zero-frequency mode f0 significantly reduce reconstruction fidelity and add severe visual artifacts'). Supplying f0 from the exact current forces the output to contain the true DC component, so the reported PSNR/SSIM gains and the abstract's 'resolves non-uniqueness challenges' are not a test of recovering that null-space mode.

full rationale

The core completion algorithm is not circular in itself: ALOHA solves the nuclear-norm/Hankel completion problem (14)/(15) for the missing nonzero coefficients, and the low-rank Hankel property is imported from independent published theorems [28,30], not from the present authors. The central circularity is experimental and load-bearing for the headline claim. Section 4.1 states that f0 is 'recovered from the mean projection of the exact current onto the polarization direction', and this exact f0 enters the reconstruction formula (7). Since Section 3.2 identifies inaccurate f0 as a primary source of artifacts and non-uniqueness, supplying the true f0 means the paper never tests recovery of the DC null-space mode from sparse far-field data. The 30–50% masks are defined over the 9260 nonzero modes (Section 4.2), so ALOHA is never asked to complete f0; the evaluation measures ALOHA plus an oracle. For J1 the leak is benign because the longitudinal component vanishes, but the larger-gain J2 results depend on it. Section 5.6 candidly lists same-model data generation and manual per-source ranks as limitations, but it does not list the f0 oracle. The source-specific ranks (22 for J1, 30 for J2, Section 4.3.3) are also a model-order oracle, compounding the issue, though they are not a separate construction-identity. Overall the derivation of the completion algorithm is self-contained, but the flagship numerical claim of resolving non-uniqueness is partially forced by injecting the true DC component of the unknown source.

Assumptions & free parameters 5 free parameters · 5 assumptions · 0 invented entities

The completion relies on the FRI/low-rank assumption and on the rank being known a priori; the zero-frequency leak is a data artifact, not a free parameter. No invented entities are introduced.

free parameters (5)
  • rank_J1 = 22
    Assumed rank for the J1 curl-type source Hankel matrix, set per source and fixed across trials (Section 4.3.3).
  • rank_J2 = 30
    Assumed rank for the J2 mixed-source Hankel matrix, set per source and fixed across trials (Section 4.3.3).
  • ADMM_penalty_mu0 = 10
    ADMM penalty parameter used for the factorized ALOHA model (Section 4.3.3).
  • ADMM_iterations = 20
    Number of ADMM iterations used for ALOHA (Section 4.3.3).
  • Hankel_patch_size = 3x3x3
    Size of the 3D block-Hankel lifting patch, fixed for both sources (Section 4.3.3).
assumptions (5)
  • domain assumption Compactly supported, geometrically sparse sources exhibit a finite rate of innovations (FRI) property.
    Invoked in Section 3.2 to justify the low-rank structure of the Hankel matrix; the central premise of the data enrichment strategy.
  • standard math An annihilating filter exists for FRI signals, guaranteeing the wrap-around Hankel matrix is low-rank.
    Cited to Vetterli et al. [25] and Ye et al. [28, Theorem II.1] in Section 3.3.
  • standard math Helmholtz-Hodge decomposition expresses the current density as F = p-hat f + p-hat x grad g.
    Used in Section 2.3 to split the source into transverse electric and magnetic components.
  • domain assumption The Fourier-inversion formulas (8)-(11) relate far-field measurements to coefficients f_l and g_l.
    Taken from Wang et al. [16] and used in Sections 3.1-3.2; assumes the forward model is exactly the Fourier-Maxwell model.
  • domain assumption The sparse measurement set is a random subset of the Nyquist-sampled grid with conjugate-pair symmetry.
    Stated in Section 2.4 and used in the experimental design (Section 4.2).

how reviews work

0 comments
Cite this review

Pith. "Pith review of Multi-frequency far-field data enrichment for electromagnetic source reconstruction." pith.science (2026). https://pith.science/paper/ITL6MXFH

@misc{pith2026260804829,
  author       = {Pith},
  title        = {Pith review of: Multi-frequency far-field data enrichment for electromagnetic source reconstruction},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ITL6MXFH}},
  note         = {Machine review of arXiv:2608.04829}
}
abstract

Reconstructing unknown electromagnetic sources from far-field radiation patterns is a fundamental inverse problem with broad applications in biomedical imaging, non-destructive testing, and telecommunications. In practical settings, however, collecting dense multi-frequency far-field measurements at the Nyquist sampling rate is often infeasible. Under-sampled or sparse data introduce non-radiating source components that sever the uniqueness of the solution, creating severe artifacts when standard inversion techniques are applied. To overcome this limitation, we present a two-stage reconstruction strategy exploiting the physical property that compactly supported, geometrically sparse sources exhibit a finite rate of innovations (FRI). In the first stage, we construct an associated wrap-around structured Hankel matrix. By leveraging the low-rank property of the matrix due to FRI of the unknown sources, we enrich the sub-sampled data. To that end, we convert missing multi-frequency far-field data recovery into a constrained matrix completion task solved via Annihilating Filter-based Low-rank Hankel Matrix Completion Approach (ALOHA). In the second stage, a Fourier inversion scheme reconstructs the current source density from the enriched dataset. Extensive numerical evaluations on electromagnetic source models show that our enrichment framework effectively eliminates under-sampling artifacts and resolves non-uniqueness challenges. The method delivers accurate and stable reconstructions under high sub-sampling rates (e.g., with $30$\% to $50$\% available samples) and strong noise conditions ($10$ dB SNR), outperforming standard $\ell_1$-compressed sensing baselines.

Figures

Figures reproduced from arXiv: 2608.04829 by the authors.

Figure 1
Figure 1. Illustration of the geometry. We refer to [26, 27] for more details on this decomposition. Throughout this article, the polarization direction pˆ is assumed to be known and is taken from the set of admissible polarization directions P := n pˆ ∈ S 2 : pˆ × ℓ ̸= 0, ℓ ∈ Z 3 \ {0} o , where Z is the set of integers. 2.4 Inverse Problem To present the ISP at hand, we first give some notation. Definition 2.1 (Admissible d… view at source ↗
Figure 2
Figure 2. Central-plane representations of electromagnetic-sources. [PITH_FULL_IMAGE:figures/full_fig_p010_2.png] view at source ↗
Figure 3
Figure 3. Central-plane electromagnetic-source reconstruction from 30% of the Fourier mea [PITH_FULL_IMAGE:figures/full_fig_p014_3.png] view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: PSNR and SSIM versus measurement rate for J1 (top row) and J2 (bottom row). [PITH_FULL_IMAGE:figures/full_fig_p015_4.png]
Figure 5
Figure 5. Figure 5: Paired Monte Carlo improvement of Joint 3D ALOHA over [PITH_FULL_IMAGE:figures/full_fig_p016_5.png]
Figure 6
Figure 6. Figure 6: Representative full-volume reconstruction at 30% measurements without noise. Rows [PITH_FULL_IMAGE:figures/full_fig_p017_6.png]
Figure 7
Figure 7. Figure 7: Representative full-volume reconstruction at 30% measurements with 10 dB AWGN. [PITH_FULL_IMAGE:figures/full_fig_p018_7.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

34 extracted references · 34 canonical work pages

  1. [1]

    Isakov, Inverse source problems, ser

    V. Isakov, Inverse source problems, ser. Mathematical surveys and monographs. Provi- dence, R.I.: American Mathematical Society, 1990, vol. 34. 19

    work page 1990

  2. [2]

    Ideal current dipoles are appro- priate source representations for simulating neurons for intracranial recordings,

    B. J. Thio, A. S. Aberra, G. E. Dessert, and W. M. Grill, “Ideal current dipoles are appro- priate source representations for simulating neurons for intracranial recordings,” Clinical Neurophysiology, vol. 145, pp. 26–35, 2023

    work page 2023

  3. [3]

    Optimal design of on- scalp electromagnetic sensor arrays for brain source localisation,

    L. Beltrachini, N. von Ellenrieder, R. Eichardt, and J. Haueisen, “Optimal design of on- scalp electromagnetic sensor arrays for brain source localisation,” Human Brain Mapping, vol. 42, no. 15, pp. 4869–4879, 2021

    work page 2021

  4. [4]

    Localization of realistic cortical activity in MEG using current multipoles,

    K. Jerbi, S. Baillet, J. Mosher, G. Nolte, L. Garnero, and R. Leahy, “Localization of realistic cortical activity in MEG using current multipoles,” NeuroImage, vol. 22, no. 2, pp. 779–793, 2004

    work page 2004

  5. [5]

    An inverse source problem for Maxwell’s equations in magnetoencephalography,

    H. Ammari, G. Bao, and J. L. Fleming, “An inverse source problem for Maxwell’s equations in magnetoencephalography,” SIAM Journal on Applied Mathematics, vol. 62, no. 4, pp. 1369–1382, 2002

    work page 2002

  6. [6]

    Multipolar acoustic source reconstruction from sparse far-field data using ALOHA,

    Y. Guo, S. Khan, A. Wahab, and X. Wang, “Multipolar acoustic source reconstruction from sparse far-field data using ALOHA,” IEEE Signal Processing Letters, vol. 30, pp. 1627–1631, 2023

    work page 2023

  7. [7]

    Ammari, J

    H. Ammari, J. Garnier, H. Kang, L. H. Nguyen, and L. Seppecher, Multi-Wave Medical Imaging. WORLD SCIENTIFIC (EUROPE), 2017. [Online]. Available: https://www.worldscientific.com/doi/abs/10.1142/q0067

    work page doi:10.1142/q0067 2017

  8. [8]

    Microwave subsurface imag- ing method by incorporating radar and tomographic approaches,

    S. Takahashi, K. Suzuki, T. Hanabusa, and S. Kidera, “Microwave subsurface imag- ing method by incorporating radar and tomographic approaches,” IEEE Transactions on Antennas and Propagation, vol. 70, no. 11, pp. 11 009–11 023, 2022

    work page 2022

Show all 34 references
  1. [9]

    Noise source localization in an at- tenuating medium,

    H. Ammari, E. Bretin, J. Garnier, and A. Wahab, “Noise source localization in an at- tenuating medium,” SIAM Journal on Applied Mathematics, vol. 72, no. 1, pp. 317–336, 2012

    2012

  2. [10]

    Application of inverse source reconstruction to conformal antennas synthesis,

    G. Leone, M. A. Maisto, and R. Pierri, “Application of inverse source reconstruction to conformal antennas synthesis,” IEEE Transactions on Antennas and Propagation, vol. 66, no. 3, pp. 1436–1445, 2018

    2018

  3. [11]

    Inverse source problem in nonhomogeneous background media,

    A. J. Devaney, E. A. Marengo, and M. Li, “Inverse source problem in nonhomogeneous background media,” SIAM Journal on Applied Mathematics, vol. 67, no. 5, pp. 1353–1378, 2007

    2007

  4. [12]

    Fast inverse equivalent source solutions with directive sources,

    T. F. Eibert, D. Vojvodi´ c, and T. B. Hansen, “Fast inverse equivalent source solutions with directive sources,” IEEE Transactions on Antennas and Propagation, vol. 64, no. 11, pp. 4713–4724, 2016

    2016

  5. [13]

    A multifrequency MUSIC algorithm for locating small inhomogeneities in inverse scattering,

    R. Griesmaier and C. Schmiedecke, “A multifrequency MUSIC algorithm for locating small inhomogeneities in inverse scattering,” Inverse Problems, vol. 33, no. 3, p. 035015, 2017

    2017

  6. [14]

    Electromagnetic time reversal algorithms and source localization in lossy dielectric media,

    A. Wahab, A. Rasheed, T. Hayat, and R. Nawaz, “Electromagnetic time reversal algorithms and source localization in lossy dielectric media,” Communications in Theoretical Physics, vol. 62, no. 6, p. 779, 2014

    2014

  7. [15]

    An inverse random source problem for the Helmholtz equation,

    G. Bao, S.-N. Chow, P. Li, and H. Zhou, “An inverse random source problem for the Helmholtz equation,” Mathematics of Computation, vol. 83, pp. 215–233, 2014. 20

    2014

  8. [16]

    Fourier method for identifying electromag- netic sources with multi-frequency far-field data,

    X. Wang, M. Song, Y. Guo, H. Li, and H. Liu, “Fourier method for identifying electromag- netic sources with multi-frequency far-field data,” Journal of Computational and Applied Mathematics, vol. 358, pp. 279–292, 2019

    2019

  9. [17]

    Nonuniqueness in the inverse source problem in acoustics and electromagnetics,

    N. Bleistein and J. K. Cohen, “Nonuniqueness in the inverse source problem in acoustics and electromagnetics,” Journal of Mathematical Physics, vol. 18, no. 2, pp. 194–201, 1977

    1977

  10. [18]

    Inverse source problems for the Helmholtz equation and the windowed Fourier transform,

    R. Griesmaier, M. Hanke, and T. Raasch, “Inverse source problems for the Helmholtz equation and the windowed Fourier transform,” SIAM Journal on Scientific Computing, vol. 34, no. 3, pp. A1544–A1562, 2012

    2012

  11. [19]

    Inverse source problems for the Helmholtz equation and the windowed Fourier transform II,

    R. Griesmaier, M. Hanke, and T. Raasch, “Inverse source problems for the Helmholtz equation and the windowed Fourier transform II,” SIAM Journal on Scientific Computing, vol. 35, no. 5, pp. A2188–A2206, 2013

    2013

  12. [20]

    Localization of extended current source with finite frequencies,

    A. Wahab, A. Rasheed, R. Nawaz, and S. Anjum, “Localization of extended current source with finite frequencies,” Comptes Rendus Mathematique, vol. 352, no. 11, pp. 917–921, 2014

    2014

  13. [21]

    Acoustic source identification using multiple frequency in- formation,

    M. Eller and N. P. Valdivia, “Acoustic source identification using multiple frequency in- formation,” Inverse Problems, vol. 25, no. 11, p. 115005, 2009

    2009

  14. [22]

    Inverse scattering problems with multi-frequencies,

    G. Bao, P. Li, J. Lin, and F. Triki, “Inverse scattering problems with multi-frequencies,” Inverse Problems, vol. 31, no. 9, p. 093001, 2015

    2015

  15. [23]

    A recursive algorithm for multifrequency acoustic inverse source problems,

    G. Bao, S. Lu, W. Rundell, and B. Xu, “A recursive algorithm for multifrequency acoustic inverse source problems,” SIAM Journal on Numerical Analysis, vol. 53, no. 3, pp. 1608– 1628, 2015

    2015

  16. [24]

    A factorization method for multifrequency inverse source problems with sparse far field measurements,

    R. Griesmaier and C. Schmiedecke, “A factorization method for multifrequency inverse source problems with sparse far field measurements,” SIAM Journal on Imaging Sciences, vol. 10, no. 4, pp. 2119–2139, 2017

    2017

  17. [25]

    Sampling signals with finite rate of innovation,

    M. Vetterli, P. Marziliano, and T. Blu, “Sampling signals with finite rate of innovation,” IEEE Transactions on Signal Processing, vol. 50, no. 6, pp. 1417–1428, 2002

    2002

  18. [26]

    The Helmholtz-Hodge decompo- sition—a survey,

    H. Bhatia, G. Norgard, V. Pascucci, and P.-T. Bremer, “The Helmholtz-Hodge decompo- sition—a survey,” IEEE Transactions on Visualization and Computer Graphics, vol. 19, no. 8, pp. 1386–1404, 2013

    2013

  19. [27]

    TE/TM decomposition of electromagnetic sources,

    I. Lindell, “TE/TM decomposition of electromagnetic sources,” IEEE Transactions on Antennas and Propagation, vol. 36, no. 10, pp. 1382–1388, 1988

    1988

  20. [28]

    Compressive sampling using annihilating filter-based low-rank interpolation,

    J. C. Ye, J. M. Kim, K. H. Jin, and K. Lee, “Compressive sampling using annihilating filter-based low-rank interpolation,” IEEE Transactions on Information Theory, vol. 63, no. 2, pp. 777–801, 2017

    2017

  21. [29]

    Annihilating filter-based low-rank Hankel matrix approach for image inpainting,

    K. H. Jin and J. C. Ye, “Annihilating filter-based low-rank Hankel matrix approach for image inpainting,” IEEE Transactions on Image Processing, vol. 24, no. 11, pp. 3498–3511, 2015

    2015

  22. [30]

    A general framework for compressed sensing and par- allel MRI using annihilating filter based low-rank Hankel matrix,

    K. H. Jin, D. Lee, and J. C. Ye, “A general framework for compressed sensing and par- allel MRI using annihilating filter based low-rank Hankel matrix,” IEEE Transactions on Computational Imaging, vol. 2, no. 4, pp. 480–495, 2016. 21

    2016

  23. [31]

    J. C. Ne´ ed´ elec,Acoustic and Electromagnetic Equations: Integral Representations for Harmonic Problems, ser. Applied Mathematical Sciences. Springer, 2001, vol. 144

    2001

  24. [32]

    Monk, Finite Element Methods for Maxwell’s Equations

    P. Monk, Finite Element Methods for Maxwell’s Equations. Oxford University Press, 2003

    2003

  25. [33]

    Colton and R

    D. Colton and R. Kress, Inverse Acoustic and Electromagnetic Scattering Theory, 4th ed., ser. Applied Mathematical Sciences. Springer-Cham, 2019, vol. 93

    2019

  26. [34]

    Compressed remote sensing of sparse objects,

    A. C. Fannjiang, T. Strohmer, and P. Yan, “Compressed remote sensing of sparse objects,” SIAM Journal on Imaging Sciences, vol. 3, no. 3, pp. 595–618, 2010. 22

    2010

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.