Erratum for: A&A, 691, A360 (2024), https://doi.org/10.1051/0004-6361/202451429

A technical error occurred during the production process. The citations for some bibliographic references were inadvertently swapped in the text. These errors concern the references listed in the bibliography from “Garsden et al. (2015)” to “Shen et al. (2021)”, including the misspelled reference “Karl Pearson (1901)” which is corrected here.

The sentences should read:

– Introduction, 2nd paragraph: “... and system health inspection (Mesarcik et al. 2020) of modern radio telescopes. [...] as, for instance, in Garsden et al. (2015). [...] explored by Rustige et al. (2023)...”

– Introduction, 9th paragraph: “In a recent paper, Reddy et al. (2024) used DMs for implementing...”

– Introduction, 10th paragraph: “The generation of synthetic radio data that includes the dirty images (...) is done elsewhere (e.g. Geyer et al. 2023).”

– Sect. 2, 2nd paragraph: “... and further developed in Ho et al. (2020) and subsequent studies.” and “... and Karras et al. (2022) proceeded to show the benefits...”.

– Sect. 2.1, 5th paragraph: “In the scope of a more general description, Karras et al. (2022) give a different formulation...”.

– Sect. 2.2, 1st sentence: “Karras et al. (2022) proceed to show that the score function...”.

– Sect. 2.3, 1st paragraph: “In order to achieve high quality for conditional sampling, Ho & Salimans (2022) introduce the concept...”.

– Sect. 2.3, 2nd paragraph: “Ho & Salimans (2022) show that the choice...”

– Sect. 3, 1st paragraph: “The other is the FIRST dataset from Griese et al. (2023)...”.

– Sect. 3.1, 1st paragraph: “... the corresponding optical cross-match catalog described in Hardcastle et al. (2023) that lists known sources...”.

– Sect. 3.2, 1st paragraph: “This dataset was introduced in Griese et al. (2023) and contains 2158 images...” and “For a more detailed description, we refer to Griese et al. (2023).”.

– Sect. 4.1, 2nd paragraph: “To implement the network, we employ a U-Net (Ronneberger et al. 2015) architecture...”.

– Sect. 4.1, 3rd paragraph: “... Sigmoid Linear Unit (SiLU) activation (Hendrycks & Gimpel 2016)...”.

– Sect. 4.1, 5th paragraph: “... through feature-wise linear modulation (Perez et al. 2018).”.

– Sect. 4.1, 6th paragraph: “... the efficient self-attention mechanism introduced in Shen et al. (2021)...”.

– Sect. 4.1, 7th paragraph: “... followed by layer normalization (Lei Ba et al. 2016)...”.

– Sect. 4.1, 9th paragraph: “... we employ a scaling to the network as described in Karras et al. (2022)...”.

– Sect. 4.2, 1st paragraph: “As introduced in Karras et al. (2022), noise levels...”.

– Sect. 4.2, 2nd paragraph: “We train both models with the Adam optimizer (Kingma & Ba 2014).”.

– Sect. 4.3, 1st paragraph: “... discretization time steps are taken from Karras et al. (2022).”.

– Sect. 5.1, 2nd paragraph: “... we run the Python Blob Detector and Source Finder (PyBDSF) (Mohan & Rafferty 2015)...”.

– Sect. 5.3, 2nd paragraph: “The encoder is implemented using the resnet-18 architecture (He et al. 2016)...”.

– Sect. 5.3, 3rd paragraph: “... using the Adam optimizer (Kingma & Ba 2014)....”.

– Sect. 7, 3rd paragraph: “... by following the methods proposed in Karras et al. (2022).”; “... by implementing architectural improvements to the U-Net proposed in Karras et al. (2024).” and “... would be the use of latent diffusion (Rombach et al. 2022)...”

– Sect. 7, 4th paragraph: “... used for natural image sampling (Ho & Salimans 2022).”.

– Sect. 7, 7th paragraph: “Rustige et al. (2023) investigated the use of GAN-generated images...”

– Sect. 7, last paragraph: “... radio images similar to the work of Reddy et al. (2024)...”.

– Appendix A, 2nd paragraph: “... We subsequently run a principal component analysis (PCA) (Pearson 1901)...”.

The corrections have also been made in the originally published article.

References

© The Authors 2025