Deep learning-based super-resolution and de-noising for XMM-newton images

Sweere, Sam F; Valtchanov, I.; Lieu, M.; Vojtekova, Antonia; Verdugo, E.; Santos-Lleo, M.; Pacaud, F.; Briassouli, Alexia; Pérez, Daniel Cámpora

doi:10.1093/mnras/stac2437

Cited by 16 publications

(5 citation statements)

References 47 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Our future work will involve finding additional, complementary methods to discriminate AGN presence within clusters, exploiting both the spectral and image properties of these objects, e.g. machine learning methods to denoise and increase the spatial resolution of XMM-Newton images (Sweere et al 2022) may lead to better classification and identification of AC systems. One limitation of our study is that unlike cool cores, contaminating AGN are not necessarily located in the centre of the X-ray cluster emission, hence the epn model developed so far is limited in identifying only missed clusters where AGN are sufficiently close to X-ray centre.…”

Section: Discussionmentioning

confidence: 99%

The XXL Survey L. AGN contamination in galaxy clusters: detection and cosmological impact

Bhargava¹,

Garrel²,

Koulouridis³

et al. 2023

Preprint

View full text Add to dashboard Cite

Context. X-ray observations of galaxy clusters are impacted by the presence of active galactic nuclei (AGN) in a manner that is challenging to quantify, leading to biases in the detection and measurement of cluster properties for both astrophysics and cosmological applications. Aims. We detect and characterise clusters contaminated by a central AGN within the XXL survey footprint and provide a systematic assessment of the cosmological impact of such systems in X-ray cluster samples. Methods. We introduce a new automated class for AGN-contaminated (AC) clusters in the XXL source detection pipeline. The majority of these systems are otherwise missed by current X-ray cluster detection methods. The AC selection is also effective at distinguishing AGN and cool core presence using supplementary optical and infrared information. Results. We present 33 AC objects, consisting of 25 clusters in the redshift range, 0.14 ≤ z ≤ 1.03, and 8 other sources with significantly peaked central profiles based on X-ray observations. Six of these are new confirmed clusters. We compute the missed fraction of the XXL survey, defined as the fraction of genuine clusters that are undetected due to their centrally peaked X-ray profiles. We report seven undetected AC clusters above z > 0.6, in the range where X-ray cluster detection efficiency drops significantly. The missed fraction is estimated to be at the level of 5% for the 50 square degree XXL area. The impact on cosmological estimates from missed clusters is negligible for XXL, but produces a ∼ 3σ tension with the fiducial cosmology when considering larger survey areas. Conclusions. This work demonstrates the first systematic attempt to quantify the percentage of missed clusters in X-ray surveys as a result of central AGN contamination. Looking towards surveys such as eROSITA and Athena, larger areas and increased sensitivity will significantly enhance cluster detection, therefore robust methods for characterising AGN contamination will be crucial for precise cluster cosmology, particularly in the redshift z > 1 regime.

show abstract

Section: Discussionmentioning

confidence: 99%

The XXL Survey L. AGN contamination in galaxy clusters: detection and cosmological impact

Bhargava¹,

Garrel²,

Koulouridis³

et al. 2023

Preprint

View full text Add to dashboard Cite

show abstract

“…For instance, Gheller & Vazza (2022) employed CNNs to remove noise and artifacts of radio interferometric images, while Bartlett et al (2023) employed CNNs to diminish the impact of noise on various observation targets and preserve the morphology of galaxies. Meanwhile, generative adversarial networks (GANs) have been applied to a variety of tasks (Hemmati et al 2022;Sweere et al 2022). Sweere et al (2022) applied GANs to generate superresolution and denoised images from the XMM-Newton telescope.…”

Section: Introductionmentioning

confidence: 99%

“…Meanwhile, generative adversarial networks (GANs) have been applied to a variety of tasks (Hemmati et al 2022;Sweere et al 2022). Sweere et al (2022) applied GANs to generate superresolution and denoised images from the XMM-Newton telescope. Hemmati et al (2022) utilized GANs to effectively deblend galaxies from Hubble Space Telescope observations.…”

Section: Introductionmentioning

confidence: 99%

Denoising Diffusion Probabilistic Models to Predict the Density of Molecular Clouds

Tan

Hsu

et al. 2023

ApJ

View full text Add to dashboard Cite

We introduce the state-of-the-art deep-learning denoising diffusion probabilistic model as a method to infer the volume or number density of giant molecular clouds (GMCs) from projected mass surface density maps. We adopt magnetohydrodynamic simulations with different global magnetic field strengths and large-scale dynamics, i.e., noncolliding and colliding GMCs. We train a diffusion model on both mass surface density maps and their corresponding mass-weighted number density maps from different viewing angles for all the simulations. We compare the diffusion model performance with a more traditional empirical two-component and three-component power-law fitting method and with a more traditional neural network machine-learning approach. We conclude that the diffusion model achieves an order-of-magnitude improvement on the accuracy of predicting number density compared to that by other methods. We apply the diffusion method to some example astronomical column density maps of Taurus and the infrared dark clouds G28.37+0.07 and G35.39-0.33 to produce maps of their mean volume densities.

show abstract

“…In the last two decades, astronomical datasets underwent a rapid growth in size and complexity thus pushing Astronomy in the big data regime (Longo et al 2019;Baron 2019;Pesenson et al 2010;Zelinka et al 2021). Traditional approaches such as interactive data reduction and analysis are laborious due to the size and the complexity of the data hence the ability of machine learning methodologies, both supervised and unsupervised to cope with very complex data, has been extensively exploited by the community to solve a wide variety of problems spanning all aspects of the astronomical data life, from instrument monitoring to data acquisition and ingestion, to data analysis and interpretation (Becker et al 2020;Kovačević et al 2020;Huertas-Company et al 2018;Margalef-Bentabol et al 2020;Lanusse et al 2021;Morningstar et al 2019;Sweere et al 2022;Zhao et al 2022;Lin et al 2022;Duarte et al 2022;Cheng et al 2020). This has also led to the implementations of many deep learning-based pipelines in the field of Astrophysics.…”

Section: Introductionmentioning

confidence: 99%

3D detection and characterization of ALMA sources through deep learning

Veneri

Tychoniec²,

Guglielmetti³

et al. 2022

Monthly Notices of the Royal Astronomical Society

View full text Add to dashboard Cite

We present a Deep-Learning (DL) pipeline developed for the detection and characterization of astronomical sources within simulated Atacama Large Millimeter/submillimeter Array (ALMA) data cubes. The pipeline is composed of six DL models: a Convolutional Autoencoder for source detection within the spatial domain of the integrated data cubes, a Recurrent Neural Network (RNN) for denoising and peak detection within the frequency domain, and four Residual Neural Networks (ResNets) for source characterization. The combination of spatial and frequency information improves completeness while decreasing spurious signal detection. To train and test the pipeline, we developed a simulation algorithm able to generate realistic ALMA observations, i.e. both sky model and dirty cubes. The algorithm simulates always a central source surrounded by fainter ones scattered within the cube. Some sources were spatially superimposed in order to test the pipeline deblending capabilities. The detection performances of the pipeline were compared to those of other methods and significant improvements in performances were achieved. Source morphologies are detected with subpixel accuracies obtaining mean residual errors of 10−3 pixel (0.1 mas) and 10−1 mJy/beam on positions and flux estimations, respectively. Projection angles and flux densities are also recovered within $10\%$ of the true values for $80\%$ and $73\%$ of all sources in the test set, respectively. While our pipeline is fine-tuned for ALMA data, the technique is applicable to other interferometric observatories, as SKA, LOFAR, VLBI, and VLTI.

show abstract

Deep learning-based super-resolution and de-noising for XMM-newton images

Cited by 16 publications

References 47 publications

The XXL Survey L. AGN contamination in galaxy clusters: detection and cosmological impact

The XXL Survey L. AGN contamination in galaxy clusters: detection and cosmological impact

Denoising Diffusion Probabilistic Models to Predict the Density of Molecular Clouds

3D detection and characterization of ALMA sources through deep learning

Contact Info

Product

Resources

About