Parallel faceted imaging in radio interferometry via proximal splitting (Faceted HyperSARA): I. Algorithm and simulations

Thouvenin, Pierre-Antoine; Abdulaziz, Abdullah; Dabbech, Arwa; Repetti, Audrey; Wiaux, Yves

doi:10.48550/arxiv.2003.07358

Cited by 3 publications

(24 citation statements)

References 46 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The fact that these artefacts are left out of Faceted HyperSARA model image confirms the robustness of its underpinning wideband image model to channel-based calibration errors. Finally, the standard deviations of the residual images obtained by the two algorithms, reported in the captions of Figures 2-3, are highly comparable, though nearly five times above the noise estimate proposed in Thouvenin et al (2021), at both frequency channels, due to calibration errors.…”

Section: Imaging Qualitymentioning

confidence: 62%

“…The regularization parameter of SARA is set to 3 times the proposed heuristic. This deviation from the heuristics proposed in Thouvenin et al (2021) can be explained by the need to compensate for residual calibration errors affecting the accuracy of the measurement operator model. For both imaging algorithms, a maximum of 5 reweighting steps is considered.…”

Section: Experimental Settingmentioning

confidence: 98%

“…Several parameters, internal to the proposed imaging approach, are set to reliable default values in the auxiliary file default_parameters.json. These include the kernel type and size taken for the NUFFT, and all the algorithmic parameters necessary to configure the reweighting procedure and inner PDFB algorithm (see (Thouvenin et al 2021, Algorithms 1 and 3)). A complete list of these parameters with the associated description is provided in the associated documentation (Thouvenin et al 2022a).…”

Section: Default Internal Parametersmentioning

confidence: 99%

“…The resulting number of spatio-spectral facets is 𝑄 ×𝐶 = 960. Given the estimates of the noise level, the regularization parameters of both algorithms are set from the heuristics proposed in Thouvenin et al (2021) as follows. In Faceted HyperSARA, the low-rankness regularization parameter is fixed to the associated proposed heuristic, and the joint sparsity regularization parameter is set to 3 times the associated heuristic.…”

Section: Experimental Settingmentioning

confidence: 99%

See 3 more Smart Citations

Parallel faceted imaging in radio interferometry via proximal splitting (Faceted HyperSARA): II. Code and real data proof of concept

Thouvenin¹,

Dabbech²,

Jiang³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

In a companion paper, a faceted wideband imaging technique for radio interferometry, dubbed Faceted HyperSARA, has been introduced and validated on synthetic data. Building on the recent HyperSARA approach, Faceted HyperSARA leverages the splitting functionality inherent to the underlying primal-dual forward-backward algorithm to decompose the image reconstruction over multiple spatio-spectral facets. The approach allows complex regularization to be injected into the imaging process while providing additional parallelization flexibility compared to HyperSARA. The present paper introduces new algorithm functionalities to address real datasets, implemented as part of a fully fledged MATLAB imaging library made available on G. A large scale proof-of-concept is proposed to validate Faceted HyperSARA in a new data and parameter scale regime, compared to the state-of-the-art. The reconstruction of a 15 GB wideband image of Cyg A from 7.4 GB of VLA data is considered, utilizing 1440 CPU cores on a HPC system for about 9 hours. The conducted experiments illustrate the reconstruction performance of the proposed approach on real data, exploiting new functionalities to set, both an accurate model of the measurement operator accounting for known direction-dependent effects (DDEs), and an effective noise level accounting for imperfect calibration. They also demonstrate that, when combined with a further dimensionality reduction functionality, Faceted HyperSARA enables the recovery of a 3.6 GB image of Cyg A from the same data using only 91 CPU cores for 39 hours. In this setting, the proposed approach is shown to provide a superior reconstruction quality compared to the state-of-the-art wideband CLEAN-based algorithm of the WSClean software.

show abstract

Section: Imaging Qualitymentioning

confidence: 62%

Section: Experimental Settingmentioning

confidence: 98%

Section: Default Internal Parametersmentioning

confidence: 99%

Section: Experimental Settingmentioning

confidence: 99%

See 2 more Smart Citations

Parallel faceted imaging in radio interferometry via proximal splitting (Faceted HyperSARA): II. Code and real data proof of concept

Thouvenin¹,

Dabbech²,

Jiang³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…Interestingly, the approach succeeds in to detect background sources in the vicinity of Cyg A at S band, illustrating the level of accuracy of the estimated mapping operator and Cyg A radio map when imaging the highly sensitive JVLA data. Future developments consist in extending the current work to the context of wide-field wideband polarimetric imaging for real RI data by leveraging Faceted HyperSARA (Thouvenin et al 2020), a recently proposed wideband imaging algorithm involving sophisticated facet-specific prior models to efficiently handle the sheer volume of wide-band RI image cubes and the theoretical developments of Birdi et al (2019) proposing a joint polarimetric calibration and imaging framework, dubbed Polca SARA.…”

Section: Discussionmentioning

confidence: 99%

Cygnus A super-resolved via convex optimization from VLA data

Dabbech

Onose

Abdulaziz

et al. 2018

Monthly Notices of the Royal Astronomical Society

View full text Add to dashboard Cite

We leverage the Sparsity Averaging Reweighted Analysis (SARA) approach for interferometric imaging, that is based on convex optimisation, for the super-resolution of Cyg A from observations at the frequencies 8.422GHz and 6.678GHz with the Karl G. Jansky Very Large Array (VLA). The associated average sparsity and positivity priors enable image reconstruction beyond instrumental resolution. An adaptive Preconditioned Primal-Dual algorithmic structure is developed for imaging in the presence of unknown noise levels and calibration errors. We demonstrate the superior performance of the algorithm with respect to the conventional clean-based methods, reflected in super-resolved images with high fidelity. The high resolution features of the recovered images are validated by referring to maps of Cyg A at higher frequencies, more precisely 17.324GHz and 14.252GHz. We also confirm the recent discovery of a radio transient in Cyg A, revealed in the recovered images of the investigated data sets. Our matlab code is available online on GitHub.

show abstract

Parallel faceted imaging in radio interferometry via proximal splitting (Faceted HyperSARA) – II. Code and real data proof of concept

Thouvenin

Dabbech

Jiang

et al. 2022

Monthly Notices of the Royal Astronomical Society

View full text Add to dashboard Cite

In a companion paper, a faceted wideband imaging technique for radio interferometry, dubbed Faceted HyperSARA, has been introduced and validated on synthetic data. Building on the recent HyperSARA approach, Faceted HyperSARA leverages the splitting functionality inherent to the underlying primal-dual forward-backward algorithm to decompose the image reconstruction over multiple spatio-spectral facets. The approach allows complex regularization to be injected into the imaging process while providing additional parallelization flexibility compared to HyperSARA. The present paper introduces new algorithm functionalities to address real datasets, implemented as part of a fully fledged MATLAB imaging library made available on https://basp-group.github.io/Puri-Psi/Github. A large scale proof-of-concept is proposed to validate Faceted HyperSARA in a new data and parameter scale regime, compared to the state-of-the-art. The reconstruction of a 15 GB wideband image of Cyg A from 7.4 GB of VLA data is considered, utilizing 1440 CPU cores on a HPC system for about 9 hours. The conducted experiments illustrate the reconstruction performance of the proposed approach on real data, exploiting new functionalities to leverage known direction-dependent effects (DDEs), for an accurate model of the measurement operator, and an effective noise level accounting for imperfect calibration. They also demonstrate that, when combined with a further dimensionality reduction functionality, Faceted HyperSARA enables the recovery of a 3.6 GB image of Cyg A from the same data using only 91 CPU cores for 39 hours. In this setting, the proposed approach is shown to provide a superior reconstruction quality compared to the state-of-the-art wideband CLEAN-based algorithm of the WSClean software.

show abstract

Parallel faceted imaging in radio interferometry via proximal splitting (Faceted HyperSARA): I. Algorithm and simulations

Cited by 3 publications

References 46 publications

Parallel faceted imaging in radio interferometry via proximal splitting (Faceted HyperSARA): II. Code and real data proof of concept

Parallel faceted imaging in radio interferometry via proximal splitting (Faceted HyperSARA): II. Code and real data proof of concept

Cygnus A super-resolved via convex optimization from VLA data

Parallel faceted imaging in radio interferometry via proximal splitting (Faceted HyperSARA) – II. Code and real data proof of concept

Contact Info

Product

Resources

About