WAsp: The Wideband (W) Adaptive-Scale Pixel (Asp) Deconvolution Algorithm for Interferometric Imaging

M. Hsieh; S. Bhatnagar; U.Rau

arxiv: 2604.22691 · v1 · submitted 2026-04-24 · 🌌 astro-ph.IM

WAsp: The Wideband (W) Adaptive-Scale Pixel (Asp) Deconvolution Algorithm for Interferometric Imaging

M. Hsieh , S. Bhatnagar , U.Rau This is my paper

Pith reviewed 2026-05-08 09:49 UTC · model grok-4.3

classification 🌌 astro-ph.IM

keywords wide-band imaginginterferometric deconvolutionscale-sensitive reconstructionAsp algorithmMS-MFSspectral index mapsradio astronomy imagingadaptive scale pixel

0 comments

The pith

The WAsp algorithm reduces wide-scale residuals and spectral index errors in wide-band radio imaging

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

The paper introduces the Wideband Asp-Clean algorithm, called WAsp, for scale-sensitive image reconstruction in wide-band interferometric data. It replaces the MS-Clean algorithm in the MS-MFS framework with an improved Asp algorithm to fix problems like leftover large-scale features in images and mistakes in spectral index calculations. This is useful for modern telescopes that need accurate models of how sources look at different frequencies to match the noise in the data. The new method keeps the computations efficient and works for narrow-band, wide-band, spectral cubes, and combined single-dish plus interferometer imaging. Tests on simulations and real observations show it performs better than earlier approaches.

Core claim

WAsp is a novel scale-sensitive image reconstruction method for wide-band applications that replaces MS-Clean with the improved Asp algorithm in the MS-MFS algorithm. This change addresses deficiencies that cause significant wide-scale residuals in Stokes-I and relative errors in spectral index maps. The algorithm maintains computational efficiency and can be configured for various imaging modes.

What carries the argument

The Adaptive-Scale Pixel (Asp) deconvolution algorithm, enhanced for imaging and runtime, used in place of MS-Clean within the Multi-Scale Multi-Frequency Synthesis (MS-MFS) wide-band framework for joint spatio-frequency modeling.

If this is right

Residuals align with the noise model across the frequency band.
Relative errors in spectral index maps are reduced.
Computational efficiency is preserved compared to prior methods.
Support for narrow-band, wide-band, spectral-cube, and joint single-dish imaging is provided.
Improved performance is shown in stress-test simulations and real wide-band data.

Where Pith is reading between the lines

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

This replacement strategy could be applied to other deconvolution methods to improve wide-band performance further.
The reduced errors in spectral index maps would enable more precise studies of source properties in radio astronomy.
Adoption in standard pipelines could benefit processing of data from sensitive arrays without major computational overhead.

Load-bearing premise

That the documented improvements to the Asp algorithm will translate directly into reduced wide-scale residuals and lower relative errors in spectral index maps when used in the MS-MFS framework, without introducing new artifacts or biases.

What would settle it

A test on a new wide-band dataset showing that residuals after WAsp imaging are larger than the thermal noise level or that spectral index map errors are higher than with the original MS-MFS algorithm.

Figures

Figures reproduced from arXiv: 2604.22691 by M. Hsieh, S. Bhatnagar, U.Rau.

**Figure 1.** Figure 1: Top panel: Spectral index comparison between the truth (left), MS-MFS with gain=0.1 (middle), and WAsp with gain=0.4 (right) on the jet dataset, illustrating the wide-band imaging test of scale-sensitive modeling and spectral index reconstruction accuracy across mixed compact and extended structures. Bottom panel: Residual image comparison of the first three Taylor terms between the MS-MFS (top) and WAsp (… view at source ↗

**Figure 2.** Figure 2: Top: Restored image comparison between MS-Clean (left) and WAsp (right) on the papersky dataset, representing a narrow-band imaging test with incomplete large-scale uv coverage to assess scale selection behavior. Bottom: Residual image comparison between MS-Clean (left) and WAsp (right) on the papersky dataset view at source ↗

**Figure 3.** Figure 3: Top: Restored image comparison between Hogbom CLEAN (left), MS-Clean (middle) and ¨ WAsp (right) on the Cyg A dataset, representing a real-data narrow-band imaging test of scale modeling. Bottom: Residual image comparison between Hogbom CLEAN (left), ¨ MS-Clean (middle) and WAsp (right) on the Cyg A dataset, highlighting the improved convergence behavior and more noise-like residual structure obtained with… view at source ↗

**Figure 4.** Figure 4: Top: Restored image comparison between the MS-MFS (left) and WAsp (right) on the G055.7+3.4 dataset, representing a real-data wide-band imaging test of scale modeling and spectral index recovery. Middle: Residual image comparison between the MS-MFS (left) and WAsp (right) on the G055.7+3.4 dataset. Bottom: Spectral index comparison between the MS-MFS (left) and WAsp (right) on the G055.7+3.4 dataset. All i… view at source ↗

**Figure 5.** Figure 5: Comparison of the convergence progress between the WAsp and MS-MFS algorithms with the jet dataset, illustrating their relative rates of approaching the final solution. Markers along each curve indicate where the calculations for the UpdateDir (“major cycle”) step were triggered. rather than the runtime of an individual iteration. A full characterization with ngVLA-scale simulations is ongoing, but the pr… view at source ↗

read the original abstract

This paper introduces the Wide-band Asp-Clean (\texttt{WAsp}) algorithm, a novel scale-sensitive image reconstruction method tailored for wide-band imaging applications. This algorithm is particularly beneficial for thermal noise-limited imaging with aperture synthesis telescopes, where joint spatio-frequency modeling of the sky brightness distribution is critical. The \texttt{WAsp} algorithm replaces the use of the MS-Clean algorithm in the MS-MFS algorithm with the {\tt Asp} algorithm \citep{Asp_Clean}, which itself has been improved for both imaging and runtime performance. With the high sensitivity of current and next-generation telescopes, spatio-frequency modeling in a scale-sensitive basis becomes crucial for ensuring that residuals align with the noise model across the frequency band. Although existing wide-band scale-sensitive algorithms have demonstrated superior performance over scale-insensitive counterparts, they often suffer from well-documented deficiencies, leading to significant wide-scale residuals in Stokes-I at low levels and consequently significant relative errors in spectral index maps. The \texttt{WAsp} algorithm addresses these limitations while maintaining computational efficiency. The implementation can be configured to support narrow-band and wide-band scale-sensitive imaging, spectral-cube imaging applications and joint single-dish and interferometer imaging. To demonstrate improved imaging performance, we show comparison with existing algorithms via carefully developed simulations for stress-testing the algorithms. We also present results from its application to real-world wide-band data, underscoring its effectiveness in practical imaging scenarios.

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.

Desk Editor's Note private letter to a colleague

WAsp is a straightforward swap of an improved Asp-Clean into the MS-MFS wide-band framework, but the simulations do not directly confirm that scale-frequency coupling is handled without new residuals or biases.

read the letter

The paper's main contribution is replacing MS-Clean with an updated Asp algorithm inside the existing MS-MFS multi-frequency synthesis loop. This produces WAsp, which the authors position as fixing low-level wide-scale Stokes-I residuals and the resulting spectral-index errors that plague other scale-sensitive wide-band methods. The implementation is set up to handle narrow-band, wide-band, spectral-cube, and single-dish plus interferometer cases, which is a practical plus for pipeline users. They also include real-data examples, which helps show the method is not purely theoretical. That combination of an existing framework with a refined scale-sensitive minor cycle is the actual new element; nothing here reinvents the Taylor-term expansion or the basic Asp scale selection logic. The comparisons are presented via simulations and real observations, and the authors claim better residual behavior and runtime. The soft spot is exactly where the stress-test note flags it: the abstract and description do not show explicit checks that Asp's adaptive scales, when driven by the MS-MFS minor cycle, avoid introducing scale-frequency cross terms or masking the original deficiencies. Without quantitative metrics, error bars, or side-by-side residual maps that isolate the coupling, the central performance claim rests on unexamined assumptions. The math and citation pattern look standard for this subfield, with no obvious circularity or invented entities. This paper is aimed at radio interferometry practitioners who already use MS-MFS and want a drop-in scale-sensitive alternative. A reader building or testing imaging pipelines could get value from trying the code, but the work is incremental rather than foundational. I would send it to peer review so the simulation details and any hidden interactions can be examined properly.

Referee Report

2 major / 2 minor

Summary. The manuscript introduces the WAsp algorithm, which replaces MS-Clean with an improved Asp-Clean deconvolution method inside the MS-MFS wide-band imaging framework. It claims that this substitution eliminates documented low-level wide-scale Stokes-I residuals and consequent relative errors in spectral index maps for thermal noise-limited wide-band observations, while preserving computational efficiency. The algorithm supports narrow-band, wide-band, spectral-cube, and joint single-dish/interferometer configurations. Performance is asserted via comparisons with existing algorithms on stress-test simulations and real-world wide-band data.

Significance. If the central performance claims are substantiated, WAsp would offer a practical advance for high-sensitivity wide-band imaging with current and next-generation aperture-synthesis telescopes. By ensuring scale-sensitive modeling produces residuals consistent with the noise model across frequencies, it could reduce biases in derived spectral indices without sacrificing runtime, addressing a documented limitation of prior wide-band scale-sensitive methods.

major comments (2)

[Abstract] Abstract: The assertion of superior performance and reduced wide-scale residuals/spectral-index errors is not supported by any quantitative metrics, error bars, or description of the comparison methodology (e.g., how residuals were measured or how scale-frequency coupling was tested), rendering the simulation results unverifiable against the central claim.
[Simulations] The load-bearing step—that Asp's adaptive scale selection, when driven by the MS-MFS minor cycle, inherits imaging gains without introducing new scale-frequency cross terms—is stated but not demonstrated with direct validation; the manuscript supplies no evidence that the Taylor-term expansion interacts with Asp updates in a manner that preserves the claimed residual properties.

minor comments (2)

The title and abstract introduce the WAsp acronym without immediately spelling out its components, which may reduce immediate clarity for readers unfamiliar with the Asp lineage.
The description of configurability for narrow-band, wide-band, spectral-cube, and single-dish modes is useful but would benefit from a brief statement of which parameters control each mode.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the careful and constructive review of our manuscript. We address each major comment below and have revised the paper to strengthen the substantiation of our performance claims.

read point-by-point responses

Referee: [Abstract] Abstract: The assertion of superior performance and reduced wide-scale residuals/spectral-index errors is not supported by any quantitative metrics, error bars, or description of the comparison methodology (e.g., how residuals were measured or how scale-frequency coupling was tested), rendering the simulation results unverifiable against the central claim.

Authors: We agree that the abstract would benefit from greater specificity. In the revised manuscript we will incorporate quantitative metrics (e.g., measured residual RMS levels across scales and relative spectral-index errors) together with a concise description of the comparison methodology, including how residuals were quantified and how scale-frequency coupling was assessed in the stress-test simulations. revision: yes
Referee: [Simulations] The load-bearing step—that Asp's adaptive scale selection, when driven by the MS-MFS minor cycle, inherits imaging gains without introducing new scale-frequency cross terms—is stated but not demonstrated with direct validation; the manuscript supplies no evidence that the Taylor-term expansion interacts with Asp updates in a manner that preserves the claimed residual properties.

Authors: The stress-test simulations already compare WAsp and MS-MFS residuals and spectral-index maps across the band to illustrate the absence of new artifacts. Nevertheless, we accept that a more explicit demonstration of the Taylor-term / Asp-update interaction is warranted. We will add a short dedicated subsection (with supporting diagnostic plots) that directly examines the scale-frequency coupling within the minor cycle and confirms that no additional cross terms are introduced. revision: partial

Circularity Check

0 steps flagged

No circularity: algorithmic replacement validated by external simulations and data

full rationale

The paper describes WAsp as an algorithmic substitution of an improved Asp-Clean into the existing MS-MFS framework, with performance claims resting on comparative simulations and real-data applications rather than any closed mathematical derivation. No equations, fitted parameters, or predictions are presented that reduce by construction to the inputs; the load-bearing step is empirical testing of the combined method, which is independent of self-citation chains or ansatz smuggling. Self-citation to prior Asp work is normal background and does not carry the central claim.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract alone supplies no information on free parameters, background axioms, or newly postulated entities; the method is presented as a reconfiguration of existing components.

pith-pipeline@v0.9.0 · 5565 in / 1058 out tokens · 47537 ms · 2026-05-08T09:49:24.195503+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

15 extracted references · 15 canonical work pages

[1]

Bhatnagar, S., & Cornwell, T. J. 2004, A&A, 426, 747

work page 2004
[2]

A., & Rupen, M

Bhatnagar, S., Rau, U., Green, D. A., & Rupen, M. P. 2011, The Astrophysical Journal Letters, 739, L20, doi: 10.1088/2041-8205/739/1/L20

work page doi:10.1088/2041-8205/739/1/l20 2011
[3]

2025, The Astronomical Journal, 170, 246, doi: 10.3847/1538-3881/adfe61

Bhatnagar, S., Rau, U., Hsieh, M., Kern, J., & Xue, R. 2025, The Astronomical Journal, 170, 246, doi: 10.3847/1538-3881/adfe61

work page doi:10.3847/1538-3881/adfe61 2025
[4]

1999, ALGLIB, http://www.alglib.net

Bochkanov, S. 1999, ALGLIB, http://www.alglib.net

work page 1999
[5]

Cornwell, T. J. 2008, IEEE Journal of Selected Topics in Signal Processing, 2, 793–801, doi: 10.1109/jstsp.2008.2006388 H¨ogbom, J. A. 1974, A&AS, 15, 417

work page doi:10.1109/jstsp.2008.2006388 2008
[6]

R., & Enßlin, T

Junklewitz, H., Bell, M. R., & Enßlin, T. 2015, A&A, 581, A59, doi: 10.1051/0004-6361/201423465 M. Galassi, M., Davies, J. T., Gough, B., et al. 2009, GNU Scientific Library Reference Manual - Third Edition (Network Theory Limited) M¨uller, H., & Bhatnagar, S. 2025, A&A, 698, A176, doi: 10.1051/0004-6361/202553990 M¨uller, H., Hsieh, M., & Bhatnagar, S. 2...

work page doi:10.1051/0004-6361/201423465 2015
[7]

1986, Annual Reviews of Astronomy and Astrophysics, 24, 127

Narayan, R., & Nityananda, R. 1986, Annual Reviews of Astronomy and Astrophysics, 24, 127

work page 1986
[8]

, keywords =

Offringa, A. R., & Smirnov, O. 2017, Monthly Notices of the Royal Astronomical Society, 471, 301, doi: 10.1093/mnras/stx1547

work page doi:10.1093/mnras/stx1547 2017
[9]

H., Teukolsky, S

Press, W. H., Teukolsky, S. A., Vetterling, W. T., & Flannery, B. P. 2007, Numerical Recipes 3rd Edition: The Art of Scientific Computing (Cambridge university press)

work page 2007
[10]

2015, LBFGS++

Qiu, Y . 2015, LBFGS++. https://lbfgspp.statr.me/

work page 2015
[11]

Rau, U., & Cornwell, T. J. 2011, A&A, 532, A71, doi: 10.1051/0004-6361/201117104 The CASA Team, et al. 2022, PASP, 134, 114501, doi: 10.1088/1538-3873/ac9642 The LibRA Team. 2025, LibRA: A library of algorithms for indirect imaging, paper in preparation, doi: https://ascl.net/2601.012

work page doi:10.1051/0004-6361/201117104 2011
[12]

2016, CppOptimizationLibrary, https://github.com/PatWie/CppNumericalSolvers

Wieschollek, P. 2016, CppOptimizationLibrary, https://github.com/PatWie/CppNumericalSolvers

work page 2016
[13]

2018, A&A, 618, A117, doi: 10.1051/0004-6361/201833090

Zhang, L. 2018, A&A, 618, A117, doi: 10.1051/0004-6361/201833090

work page doi:10.1051/0004-6361/201833090 2018
[14]

2020, Publications of the Astronomical Society of the Pacific, 132, 041001, doi: 10.1088/1538-3873/ab7345

Zhang, L., Xu, L., & Zhang, M. 2020, Publications of the Astronomical Society of the Pacific, 132, 041001, doi: 10.1088/1538-3873/ab7345

work page doi:10.1088/1538-3873/ab7345 2020
[15]

2021, Research in Astronomy and Astrophysics, 21, 063, doi: 10.1088/1674-4527/21/3/63

Zhang, L., Mi, L.-G., Xu, L., et al. 2021, Research in Astronomy and Astrophysics, 21, 063, doi: 10.1088/1674-4527/21/3/63

work page doi:10.1088/1674-4527/21/3/63 2021

[1] [1]

Bhatnagar, S., & Cornwell, T. J. 2004, A&A, 426, 747

work page 2004

[2] [2]

A., & Rupen, M

Bhatnagar, S., Rau, U., Green, D. A., & Rupen, M. P. 2011, The Astrophysical Journal Letters, 739, L20, doi: 10.1088/2041-8205/739/1/L20

work page doi:10.1088/2041-8205/739/1/l20 2011

[3] [3]

2025, The Astronomical Journal, 170, 246, doi: 10.3847/1538-3881/adfe61

Bhatnagar, S., Rau, U., Hsieh, M., Kern, J., & Xue, R. 2025, The Astronomical Journal, 170, 246, doi: 10.3847/1538-3881/adfe61

work page doi:10.3847/1538-3881/adfe61 2025

[4] [4]

1999, ALGLIB, http://www.alglib.net

Bochkanov, S. 1999, ALGLIB, http://www.alglib.net

work page 1999

[5] [5]

Cornwell, T. J. 2008, IEEE Journal of Selected Topics in Signal Processing, 2, 793–801, doi: 10.1109/jstsp.2008.2006388 H¨ogbom, J. A. 1974, A&AS, 15, 417

work page doi:10.1109/jstsp.2008.2006388 2008

[6] [6]

R., & Enßlin, T

Junklewitz, H., Bell, M. R., & Enßlin, T. 2015, A&A, 581, A59, doi: 10.1051/0004-6361/201423465 M. Galassi, M., Davies, J. T., Gough, B., et al. 2009, GNU Scientific Library Reference Manual - Third Edition (Network Theory Limited) M¨uller, H., & Bhatnagar, S. 2025, A&A, 698, A176, doi: 10.1051/0004-6361/202553990 M¨uller, H., Hsieh, M., & Bhatnagar, S. 2...

work page doi:10.1051/0004-6361/201423465 2015

[7] [7]

1986, Annual Reviews of Astronomy and Astrophysics, 24, 127

Narayan, R., & Nityananda, R. 1986, Annual Reviews of Astronomy and Astrophysics, 24, 127

work page 1986

[8] [8]

, keywords =

Offringa, A. R., & Smirnov, O. 2017, Monthly Notices of the Royal Astronomical Society, 471, 301, doi: 10.1093/mnras/stx1547

work page doi:10.1093/mnras/stx1547 2017

[9] [9]

H., Teukolsky, S

Press, W. H., Teukolsky, S. A., Vetterling, W. T., & Flannery, B. P. 2007, Numerical Recipes 3rd Edition: The Art of Scientific Computing (Cambridge university press)

work page 2007

[10] [10]

2015, LBFGS++

Qiu, Y . 2015, LBFGS++. https://lbfgspp.statr.me/

work page 2015

[11] [11]

Rau, U., & Cornwell, T. J. 2011, A&A, 532, A71, doi: 10.1051/0004-6361/201117104 The CASA Team, et al. 2022, PASP, 134, 114501, doi: 10.1088/1538-3873/ac9642 The LibRA Team. 2025, LibRA: A library of algorithms for indirect imaging, paper in preparation, doi: https://ascl.net/2601.012

work page doi:10.1051/0004-6361/201117104 2011

[12] [12]

2016, CppOptimizationLibrary, https://github.com/PatWie/CppNumericalSolvers

Wieschollek, P. 2016, CppOptimizationLibrary, https://github.com/PatWie/CppNumericalSolvers

work page 2016

[13] [13]

2018, A&A, 618, A117, doi: 10.1051/0004-6361/201833090

Zhang, L. 2018, A&A, 618, A117, doi: 10.1051/0004-6361/201833090

work page doi:10.1051/0004-6361/201833090 2018

[14] [14]

2020, Publications of the Astronomical Society of the Pacific, 132, 041001, doi: 10.1088/1538-3873/ab7345

Zhang, L., Xu, L., & Zhang, M. 2020, Publications of the Astronomical Society of the Pacific, 132, 041001, doi: 10.1088/1538-3873/ab7345

work page doi:10.1088/1538-3873/ab7345 2020

[15] [15]

2021, Research in Astronomy and Astrophysics, 21, 063, doi: 10.1088/1674-4527/21/3/63

Zhang, L., Mi, L.-G., Xu, L., et al. 2021, Research in Astronomy and Astrophysics, 21, 063, doi: 10.1088/1674-4527/21/3/63

work page doi:10.1088/1674-4527/21/3/63 2021