Pith. sign in

REVIEW 3 major objections 7 minor 56 references

Multiobjective optimization for scattering mitigation and scattering screen reconstruction in VLBI observations of the Galactic Center

T0 review · 3 major / 7 minor · reviewed 2026-08-16 · deepseek-v4-flash

Pith's one-line read A multiobjective evolutionary strategy jointly recovers the intrinsic Sgr A* image and the scattering screen in VLBI data, reconstructing the 86 GHz ring with nxcorr = 0.97 and screen velocities within 10 km/s.

desk verdict A self-consistent method demonstration that deserves refereeing but needs model-mismatch tests before the 86 GHz Sgr A* promise is taken at face value. read the letter →

arxiv 2504.16257 v2 pith:D53HMJ2K submitted 2025-04-22 astro-ph.IM astro-ph.GAmath.OC

classification astro-ph.IMastro-ph.GAmath.OC PACS 95.75.Kk95.75.Mn
keywords verylongbaselineinterferometryinterstellarscatteringmultiobjectiveoptimizationParetofrontevolutionaryalgorithmblackholeshadowSgrA*imagereconstruction
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 interstellar scattering toward the Galactic Center does not have to be removed by treating the screen as an unknown nuisance: the intrinsic source image and the phase screen can be recovered together by a multiobjective evolutionary optimizer. It embeds a stochastic-optics screen model as one objective among image regularizers and data-fidelity terms, then explores the Pareto front of solutions with two nature-inspired algorithms. In simulations, the joint recovery works at 230 GHz, where scattering is weak and barely constrained, and at 86 GHz, where scattering dominates and the image would otherwise be severely blurred. A sympathetic reader should care because a working descattering pipeline at 86 GHz would expose event-horizon-scale structure of Sgr A* at a frequency where the ring has not been seen, and would turn the scattering screen itself into a measurable astrophysical object.

What carries the argument

The load-bearing object is the scalarized multiobjective optimization problem whose decision vector $x = (\omega_I, \varphi_1, \varphi_2, \ldots)$ holds the wavelet coefficients of the intrinsic image and one phase screen per frame. A solution is Pareto-optimal when no objective can improve without worsening another; the set of all such solutions is the Pareto front. The objective vector has eight components: six standard imaging regularizers, the stochastic-optics screen regularizer $R_{\mathrm{SO}}$, which penalizes deviations of the reconstructed screen's power spectrum from the assumed $Q(q)$ of Eq. (13), and a data-fidelity term combining $\chi^2$ of visibilities, amplitudes, closure phases, and closure amplitudes evaluated on the forward-scattered model $\Phi([I_{\mathrm{src}},\varphi])$. The algorithm MOEA/D approximates the Pareto front by genetic evolution over a population, while the faster variant MO-PSO uses particle-swarm optimization to navigate the front and solves each scalarized subproblem by L-BFGS-B minimization. The Pareto-front clusters are interpreted as local modes of the multimodal imaging posterior, which is what makes the joint source-screen recovery possible.

What would settle it

Create synthetic VLBI observations with a two-screen scattering model, or with a screen spectral index different from $\alpha=1.383$, run the pipeline, and check whether the recovered intrinsic ring keeps $nxcorr \ge 0.97$ and the screen velocity stays within 10 km/s of the input.

Watch

Extended reading notes

Core claim

On its own terms, the paper establishes that scattering mitigation and screen reconstruction can be solved as one multiobjective inverse problem instead of separate deblurring and screen-modeling steps. In synthetic 230 GHz VLBI observations the method recovers the intrinsic structure and the moving phase screen, tracking screen evolution over about 100 minutes and reading off velocities of 51.4 km/s and 193.8 km/s for inputs of 50 and 200 km/s. In synthetic 86 GHz observations with the array configuration considered in the paper, where refractive scattering dominates the noise, the method recovers the intrinsic ring with normalized cross-correlation 0.97 and the screen with 0.81 (beam-blurred) and 0.88 (unblurred). The paper further shows that the optimization problem is genuinely multimodal: the starting geometry changes the probability of finding a ring from about 38 percent with a ring prior, to about 25 percent with a Gaussian prior, to zero with a disk prior, and that the screen regularizer is more tightly constrained at 86 GHz than at 230 GHz.

Load-bearing premise

The load-bearing premise is that the scattering toward Sgr A* is produced by a single thin screen with homogeneous Gaussian statistics and the fixed power spectrum used by the algorithm's screen regularizer; if the real screen is statistically different or has multiple layers, the reconstructed screen and the descattered image inherit that error.

Editorial extensions

If this is right

  • At 86 GHz, the multiobjective strategy recovers the intrinsic about-50 microarcsecond ring with a normalized cross-correlation of 0.97, so descattered imaging of the Sgr A* shadow at that frequency is within reach.
  • Screen velocities are recovered from the time evolution of the reconstructed phase screen: 51.4 km/s for a 50 km/s input, 193.8 km/s for a 200 km/s input, with errors below 10 km/s across tested source geometries.
  • Starting the search from a Gaussian prior yields ring-like solutions in about 25 percent of MOEA/D runs, a ring prior in about 38 percent, and a disk prior in zero percent, so the starting geometry is a real control on the outcome.
  • The stochastic-optics regularizer's marginal contribution is more tightly clustered at 86 GHz than at 230 GHz, reflecting that refractive noise more strongly constrains the screen at lower frequencies.
  • Because the screen and source are recovered together, the same framework can in principle separate screen kinematics from intrinsic source structure in future dynamic analyses.

Reading between the lines

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

  • Editorial inference: the same joint source-screen formulation should apply to any compact Galactic-plane source observed through a thin turbulent screen, provided a screen power spectrum is supplied from calibrator observations.
  • Editorial inference: because a Gaussian starting point finds ring solutions in only about 25 percent of runs while a disk starting point finds none, single-start local-optimization pipelines may systematically miss the true structure on real 86 GHz data even when that structure is present in the posterior.
  • Editorial inference: a measured screen velocity on real observations would separate screen kinematics from intrinsic source variability, a step the paper itself identifies as necessary before making movies of Sgr A* at horizon scales.
  • Editorial inference: the tight screen-regularizer constraint seen at 86 GHz suggests that lower-frequency or longer-baseline data could be used to estimate the screen power spectrum empirically rather than assuming the standard form.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 7 minor

Summary. This paper presents a multiobjective optimization approach to jointly reconstruct the intrinsic source image and the interstellar scattering screen in VLBI observations of Sgr A*. The authors incorporate the stochastic-optics screen model of Johnson (2016) as an additional regularizer (RSO, Eq. 11) within the MOEA/D and MO-PSO frameworks. They demonstrate on synthetic data that the method recovers (i) a static asymmetric ring at 230 GHz in the presence of gain corruptions, (ii) the velocity of a moving screen (51.4 vs 50 km/s and 193.8 vs 200 km/s), and (iii) a ring structure at 86 GHz with GMVA+ALMA coverage, reporting nxcorr=0.97 for the intrinsic source and 0.81/0.88 for the blurred/non-blurred screen. They also examine the influence of the starting geometry on ring recovery and the marginal contribution of the SO regularizer. The paper concludes that the method opens the possibility of descattered imaging and screen characterization for Sgr A* at 86-116 GHz.

Significance. If the claims held, the method would be an important new capability for low-frequency VLBI imaging of the Galactic Center, enabling tests of black hole shadow morphology at 86 GHz and measurements of screen properties. The main strengths are the principled integration of stochastic optics into a multiobjective framework, the explicit handling of multimodality through Pareto-front exploration, and the release of the pipeline in the MrBeam package. However, the demonstrated results are on synthetic data generated from the same screen model used in the regularizer, so the evidence supports the method as a self-consistent inversion tool, not yet as a validated astrophysical screen model. The overclaim in Sec. 8 that the approach is 'not restricted to a specific screen model' is unsupported.

major comments (3)
  1. [Sec. 4.1, Eq. (11), Sec. 4.2] The RSO regularizer in Eq. (11) is defined as sum |epsilon_{o,s}|^2, where the screen is parameterized as phi_{l,m} = (1/sqrt(A_phi)) sum sqrt(Q(o,s)) epsilon_{o,s} exp(...). This is exactly a Gaussian prior with covariance Q(q) from Eq. (13). Since all synthetic screens in Sec. 4.2 are generated with the StochasticOptics module using the same Q(q), alpha=1.383, and M=0.43, the reported nxcorr values (0.55-0.75 at 230 GHz, 0.81/0.88 at 86 GHz) establish that the algorithm can invert its own generative model. They do not establish robustness to a real screen with a different spectral slope, anisotropy, multiple screens, or temporal evolution. The Sec. 8 statement that the method is 'not restricted to a specific screen model' is contradicted by the implementation, and Sec. 4.4.2 concedes that a physically feasible screen prior is needed for convergence. I recommend either adding a validation test with a screen drawn from a different model or clearly restricting the claims to the Johnson (2016) single-screen power-law model.
  2. [Sec. 5.2, Table 1] The percentages in Table 1 (25.45%, 0%, 37.57%) are reported as 'probabilities' of ring recovery, but they are computed from a single MOEA/D run per starting geometry (340 individuals per run). A single population is a sample, not a probability distribution; no independent restarts, noise resamplings, or repeated self-calibration iterations are used to get uncertainties. Thus the statement in Sec. 5.3 that 'starting from a Gaussian prior yields a probability of approximately 25%' is an overinterpretation. Repeated runs with different seeds and noise realizations are needed before these frequencies can be interpreted as probabilities.
  3. [Sec. 4.4, Sec. 5.1] All demonstrations use a single synthetic realization per setup, and no error bars are provided for the key quantitative claims. In particular, the screen velocity estimates (51.4 km/s for a 50 km/s input and 193.8 km/s for 200 km/s, Sec. 4.4.3) are given as point values with no measure of uncertainty, and the screen nxcorr values at 230 GHz are modest (0.55-0.75). Additionally, Fig. 3 and Table B.1 show that the recovered screen depends on the prior screen (e.g., nxcorr 0.45 vs 0.57 for the two cases). The abstract and conclusions should therefore be tempered: the paper has demonstrated feasibility on model-consistent synthetic data, but the accuracy of screen recovery in real observations remains unquantified.
minor comments (7)
  1. [Sec. 2.1] In the paragraph after Eq. (2), 'exasperated' should be 'exacerbated'.
  2. [Sec. 4.1] The heading 'Discretizsation' is misspelled; it should be 'Discretization'.
  3. [Sec. 5.1] There is a typo 'a a low SNR' in Sec. 5.1; also 'grad' should be 'deg' in 'PA of 86 grad'.
  4. [Table 1] In Table 1, 'morpohologies' should be 'morphologies'.
  5. [Eq. (B.1)] In Eq. (B.1), define sigma_X and sigma_Y as the standard deviations of X and Y, and clarify whether the population or sample standard deviation is used.
  6. [Sec. 2.3] The citation 'Jonson 1999' appears in Sec. 2.3; please verify the spelling of this author name (the reference list also gives 'Jonson, B.').
  7. [Abstract] In the abstract, '86G Hz' should be '86 GHz'.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the paper's method is validated on simulations generated with the same screen model used by its regularizer, but this is a standard self-consistency test, not a derivation that reduces to its inputs.

full rationale

The paper's central contribution is algorithmic: it embeds Johnson's stochastic-optics screen regularizer (Eq. 11, RSO = sum |epsilon_{o,s}|^2) into the MOEA/D and MO-PSO multiobjective frameworks and tests the inversion on synthetic data. The screen parameterization phi_lm proportional to sum sqrt(Q(o,s)) epsilon_{o,s} exp(...) and RSO share the assumed power spectrum Q(q) of Eq. (13), and the synthetic screens in Sec. 4.2 are drawn from the same StochasticOptics model. This means the screen-recovery experiments are MAP-inversion tests under a correct generative model; they do not validate Q(q) against the real Sgr A* screen. However, this is a limitation on external validity, not circularity: the paper explicitly states 'There are no updates to the modeling of the screen itself in this paper,' the recovered screen phases and the measured screen velocities (51.4 vs 50 km/s, 193.8 vs 200 km/s) are not encoded in the prior, and the 86 GHz ring recovery uses generic image regularizers (l1, TV, TSV, l2, flux, entropy) rather than a ring prior. The paper itself flags the prior dependence in Sec. 4.4.2 ('a physically feasible screen prior is needed to recover the screen') and the suboptimality of the nxcorr metric in Appendix B. The Sec. 8 statement that the method provides a solution 'without being restricted to a specific screen model' overstates the implementation, which does assume Q(q), but this is an overclaim about scope, not a circular reduction. The self-citations to Mueller et al. (2023) and Mus et al. (2024a,b) introduce the MO framework but are supplemented by new demonstrations in this paper, so they are not load-bearing in a circular sense.

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

No new physical entities are introduced. The model relies on the standard Johnson (2016) stochastic optics screen with assumed spectral index, screen location, and diffractive scales taken from prior literature. The key unstated inputs are the specific prior screen used at 230 GHz and the MO hyperparameters, both of which affect the reported reconstructions.

free parameters (5)
  • phase spectral index alpha = 1.383
    Power-law index of the phase fluctuation spectrum; taken as the default value in StochasticOptics (Johnson 2016), not derived in this paper.
  • screen distance parameter M = 0.43
    Location of the scattering screen as a fraction of the distance to Sgr A*, from Bower et al. 2014; assumed input.
  • diffractive scale r0,x and r0,y
    Dimensional scales of the scattering kernel in Eq. (13); values are not stated in the text and are presumably inherited from eht-imaging defaults.
  • prior screen
    Algorithm 1 requires a prior screen; the paper says a uniform screen is not valid and a physically feasible prior is needed, but the exact prior used in the 230 GHz tests is not specified.
  • MO hyperparameters (populations, iterations, thresholds)
    Population sizes, iteration counts, cluster similarity threshold of 10%, and self-calibration iterations are chosen by the authors and are not optimized or justified in detail.
assumptions (5)
  • domain assumption A single thin scattering screen with homogeneous Gaussian phase statistics describes the line of sight to Sgr A*
    The entire forward model and the RSO regularizer rely on this; the number of screens in the line of sight is an open question (Dexter et al. 2017).
  • domain assumption The phase fluctuation power spectrum follows Q(q) proportional to |q|^{-(1+alpha/2)} with alpha = 1.383
    Used to parameterize the screen and define the RSO regularizer in Sec 4.1.
  • domain assumption The screen evolves as a frozen pattern advected at constant velocity over the observation
    Needed for the velocity recovery test in Sec 4.4.3; real Sgr A* source variability is acknowledged but not modeled.
  • ad hoc to paper Closure quantities and self-calibration preserve the ring structure at 86 GHz
    Algorithm 2 uses only closure quantities for MOEA/D and then self-calibrates; the paper states that relative position can be lost and recentering is needed, indicating the assumption is not guaranteed.
  • standard math Wavelet dictionary representation I = Psi omega_I
    Standard sparse representation for images, used in Sec 3.1.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Multiobjective optimization for scattering mitigation and scattering screen reconstruction in VLBI observations of the Galactic Center." pith.science (2026). https://pith.science/paper/D53HMJ2K

@misc{pith2026250416257,
  author       = {Pith},
  title        = {Pith review of: Multiobjective optimization for scattering mitigation and scattering screen reconstruction in VLBI observations of the Galactic Center},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/D53HMJ2K}},
  note         = {Machine review of arXiv:2504.16257}
}
read the original abstract

Imaging reconstruction of interferometric data is a hard ill-posed inverse problem. Its difficulty is increased when observing the Galactic Center, which is obscured by a scattering screen. This is because the scattering breaks the one-to-one correspondence between images and visibilities. Solving the scattering problem is one of the biggest challenges in radio imaging of the Galactic Center. In this work we present a novel strategy to mitigate its effect and constrain the screen itself using multiobjective optimization. We exploit the potential of evolutionary algorithms to describe the optimization landscape to recover the intrinsic source structure and the scattering screen affecting the data. We successfully recover both the screen and the source in a wide range of simulated cases, including the speed of a moving screen at 230 GHz. Particularly, we can recover a ring structure in scattered data at 86 GHz. Our analysis demonstrates the huge potential that recent advancements in imaging and optimization algorithms offer to recover image structures, even in weakly constrained and degenerated, possibly multi-modal settings. The successful reconstruction of the scattering screen opens the window to event horizon scale works on the Galactic Center at 86G Hz up to 116 GHz, and the study of the scattering screen itself.

Figures

Figures reproduced from arXiv: 2504.16257 by the authors.

Figure 1
Figure 1. Top left panel: uv coverage for Sgr A [PITH_FULL_IMAGE:figures/full_fig_p007_1.png] view at source ↗
Figure 2
Figure 2. Scattering reconstruction using MO-PSO of the static ring with an asymmetric brightness distribution with [PITH_FULL_IMAGE:figures/full_fig_p009_2.png] view at source ↗
Figure 3
Figure 3. Comparison of the scattering reconstructions of the ring with an asymmetric brightness with different screen [PITH_FULL_IMAGE:figures/full_fig_p010_3.png] view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: Recovery of the scattering screen velocity for a screen that affects a static crescent. Top row: True scattering [PITH_FULL_IMAGE:figures/full_fig_p010_4.png]
Figure 5
Figure 5. Figure 5: Dynamic scattering reconstruction of the intrinsic source and the evolving screen equivalent to that in Fig. 4. Top [PITH_FULL_IMAGE:figures/full_fig_p011_5.png]
Figure 6
Figure 6. Figure 6: uv-coverage for the GMVA + ALMA array observing Sgr A [PITH_FULL_IMAGE:figures/full_fig_p011_6.png]
Figure 7
Figure 7. Figure 7: Scattered ring observed at 86 GHz (first panel) with a gain corruption of 4%. The second panel shows the simulated [PITH_FULL_IMAGE:figures/full_fig_p013_7.png]
Figure 8
Figure 8. Figure 8: Marginal contribution of the hyperparameter asso [PITH_FULL_IMAGE:figures/full_fig_p013_8.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

56 extracted references · 28 canonical work pages

  1. [1]

    2023, MNRAS, 520, 1872

    Beniamini, P., Wadiasingh, Z., Hare, J., et al. 2023, MNRAS, 520, 1872

  2. [2]

    & Narayan, R

    Blandford, R. & Narayan, R. 1985, MNRAS, 213, 591

  3. [3]

    & Wolf, E

    Born, M. & Wolf, E. 1980, Principles of Optics Electromagnetic The- ory of Propagation, Interference and Diffraction of Light

  4. [4]

    L., Johnson, M

    Bouman, K. L., Johnson, M. D., Dalca, A. V., et al. 2017, arXiv e- prints, arXiv:1711.01357

  5. [5]

    C., Deller, A., Demorest, P., et al

    Bower, G. C., Deller, A., Demorest, P., et al. 2014, ApJ, 780, L2

  6. [6]

    C., Falcke, H., Herrnstein, R

    Bower, G. C., Falcke, H., Herrnstein, R. M., et al. 2004, Science, 304, 704

  7. [7]

    C., Markoff, S., Dexter, J., et al

    Bower, G. C., Markoff, S., Dexter, J., et al. 2015, ApJ, 802, 69

  8. [8]

    E., Gold, R., Karami, M., et al

    Broderick, A. E., Gold, R., Karami, M., et al. 2020, ApJ, 897, 139

Show all 56 references
  1. [9]

    2022, Nature Astronomy, 6, 828

    Caleb, M., Heywood, I., Rajwade, K., et al. 2022, Nature Astronomy, 6, 828

  2. [10]

    A., Johnson, M

    Chael, A. A., Johnson, M. D., Bouman, K. L., et al. 2018, ApJ, 857, 23

  3. [11]

    A., Johnson, M

    Chael, A. A., Johnson, M. D., Narayan, R., et al. 2016, ApJ, 829, 11

  4. [12]

    D., et al

    Chavez, E., Issaoun, S., Johnson, M. D., et al. 2024, ApJ, 974, 116

  5. [13]

    Clark, B. G. 1980, A&A, 89, 377

  6. [14]

    Cordes, J. M. & Lazio, T. J. W. 2002, arXiv e-prints, astro

  7. [15]

    Cornwell, T. J. 2008, IEEE Journal of Selected Topics in Signal Pro- cessing, 2, 793

  8. [16]

    C., et al

    Dexter, J., Deller, A., Bower, G. C., et al. 2017, MNRAS, 471, 3563

  9. [17]

    & Swamy, M

    Du, K.-L. & Swamy, M. N. S. 2016, Search and Optimization by Meta- heuristics: Techniques and Algorithms Inspired by Nature, 1st edn. (Birkhäuser Basel) Event Horizon Telescope Collaboration, Akiyama, K., Alberdi, A., et al. 2024a, ApJ, 964, L25 Event Horizon Telescope Collabo...

  10. [18]

    2022, The Astrophysical Journal Letters, 930, 1

    Farah, J., Galison, P., Akiyama, K., et al. 2022, The Astrophysical Journal Letters, 930, 1

  11. [19]

    L., Johnson, M

    Fish, V. L., Johnson, M. D., Lu, R.-S., et al. 2014, ApJ, 795, 134

  12. [20]

    2017, International Journal of Modern Physics D, 26, 1730001

    Goddi, C., Falcke, H., Kramer, M., et al. 2017, International Journal of Modern Physics D, 26, 1730001

  13. [21]

    & Narayan, R

    Goodman, J. & Narayan, R. 1989, MNRAS, 238, 995

  14. [22]

    R., Danen, R

    Gwinn, C. R., Danen, R. M., Middleditch, J., Ozernoy, L. M., & Tran, T. K. 1991, ApJ, 381, L43 Högbom, J. A. 1974, A&AS, 15, 417

  15. [23]

    D., Blackburn, L., et al

    Issaoun, S., Johnson, M. D., Blackburn, L., et al. 2019, ApJ, 871, 30

  16. [24]

    Johnson, M. D. 2016, ApJ, 833, 74

  17. [25]

    D., Bouman, K

    Johnson, M. D., Bouman, K. L., Blackburn, L., et al. 2017, ApJ, 850, 172

  18. [26]

    Johnson, M. D. & Gwinn, C. R. 2015, ApJ, 805, 180

  19. [27]

    Johnson, M. D. & Narayan, R. 2016, ApJ, 826, 170

  20. [28]

    D., Narayan, R., Psaltis, D., et al

    Johnson, M. D., Narayan, R., Psaltis, D., et al. 2018, ApJ, 865, 104

  21. [29]

    1999, in American Institute of Physics Conference Series, Vol

    Jonson, B. 1999, in American Institute of Physics Conference Series, Vol. 495, Experimental Nuclear Physics in europe: Facing the next millennium (AIP), 3–8

  22. [30]

    S., et al

    Kim, J.-S., Mueller, H., Nikonov, A. S., et al. 2024a, arXiv e-prints, arXiv:2409.00540

  23. [31]

    S., Roth, J., et al

    Kim, J.-S., Nikonov, A. S., Roth, J., et al. 2024b, arXiv e-prints, arXiv:2407.14873

  24. [32]

    Kouroshnia, A., Nguyen, K., Ni, C., SaraerToosi, A., & Broderick, A. E. 2025, arXiv preprint arXiv:2501.14055, submitted to ApJ

  25. [33]

    & Zhang, Q

    Li, H. & Zhang, Q. 2009, IEEE Transactions on Evolutionary Com- putation, 13, 284

  26. [34]

    Little, L. T. 1973, Astrophys. Lett., 13, 115

  27. [35]

    Litvak, M. M. 1971, ApJ, 170, 71

  28. [36]

    P., et al

    Lu, R.-S., Asada, K., Krichbaum, T. P., et al. 2023, Nature, 616, 686 Mościbrodzka, M., Falcke, H., & Shiokawa, H. 2016, A&A, 586, A38 Mościbrodzka, M., Falcke, H., Shiokawa, H., & Gammie, C. F. 2014, A&A, 570, A7 Müller, H. 2024, A&A, 689, A299 Müller, H. & Lobanov, A. P. 202...

  29. [37]

    & Martí-Vidal, I

    Mus, A. & Martí-Vidal, I. 2024, Monthly Notices of the Royal Astro- nomical Society, stae234

  30. [38]

    1992, Philosophical Transactions of the Royal Society of London Series A, 341, 151

    Narayan, R. 1992, Philosophical Transactions of the Royal Society of London Series A, 341, 151

  31. [39]

    & Goodman, J

    Narayan, R. & Goodman, J. 1989, MNRAS, 238, 963

  32. [40]

    M., Žilinskas, A., & Žilinskas, J

    Pardalos, P. M., Žilinskas, A., & Žilinskas, J. 2017, Non-Convex Multi- Objective Optimization

  33. [41]

    2022, ngehtsim Documentation, https://smithsonian

    Pesce, D. 2022, ngehtsim Documentation, https://smithsonian. github.io/ngehtsim/html/docs/source/index.html

  34. [42]

    2018, arXiv e-prints, arXiv:1805.01242

    Psaltis, D., Johnson, M., Narayan, R., et al. 2018, arXiv e-prints, arXiv:1805.01242

  35. [43]

    W., Doeleman, S

    Raymond, A. W., Doeleman, S. S., Asada, K., et al. 2024, AJ, 168, 130

  36. [44]

    regpy:Pythontoolsforregularizationmethods

    Raymond, A. W., Palumbo, D., Paine, S. N., et al. 2021, ApJS, 253, 5 Regpy.2019,"regpy:Pythontoolsforregularizationmethods", https: //github.com/regpy/regpy

  37. [45]

    J., Menten, K

    Reid, M. J., Menten, K. M., Brunthaler, A., et al. 2019, ApJ, 885, 131

  38. [46]

    A., Bender, R., et al

    Richstone, D., Ajhar, E. A., Bender, R., et al. 1998, Nature, 385, A14

  39. [47]

    Rickett, B. J. 1990, ARA&A, 28, 561

  40. [48]

    2023, Galaxies, 11, 12

    Roelofs, F., Blackburn, L., Lindahl, G., et al. 2023, Galaxies, 11, 12

  41. [49]

    Taylor, J. H. & Cordes, J. M. 1993, ApJ, 411, 674

  42. [50]

    R., Moran, J

    Thompson, A. R., Moran, J. M., & Swenson, George W., J. 2017, Interferometry and Synthesis in Radio Astronomy, 3rd Edition

  43. [51]

    2022, The Journal of Open Source Software, 7, 4457

    Tiede, P. 2022, The Journal of Open Source Software, 7, 4457

  44. [52]

    2022, ApJ, 930, L19

    Wielgus, M., Marchili, N., Martí-Vidal, I., et al. 2022, ApJ, 930, L19

  45. [53]

    1949, Extrapolation, Interpolation, and Smoothing of Sta- tionary Time Series: With Engineering Applications (The MIT Press)

    Wiener, N. 1949, Extrapolation, Interpolation, and Smoothing of Sta- tionary Time Series: With Engineering Applications (The MIT Press)

  46. [54]

    Zhang, Q. & Li, H. 2007, IEEE Transactions on Evolutionary Com- putation, 11, 712 Article number, page 16 of 19

  47. [55]

    ring Fig

    Mus, A.: Multiobjective optimization for scattering mitigation in very long baseline interferometry Appendix A: Additional figures 350 µas x0 for MOEA/D 0.5 1.0 1.5 2.0 2.5 3.0 3.5 u− v Distance ×109 10□10 10□8 10□6 10□4 10□2 100 Amplitude x0 for MOEA/D Gaussian Disk Ring Simu...

  48. [56]

    C.2: Pareto front for the intrinsic source at 86GHz

    Mus, A.: Multiobjective optimization for scattering mitigation in very long baseline interferometry Cluster 0; sol=38.18%; chi2=20485.31 Cluster 1; sol=16.97%; chi2=21494.65 Cluster 2; sol=6.36%; chi2=20503.95 Cluster 3; sol=2.12%; chi2=19761.02 Cluster 4; sol=16.97%; chi2=336...

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.