Filter-substrate refraction causes dominant lateral shifts yielding 0.3-0.4% PSF size and ellipticity residuals across most Roman bands that exceed weak lensing requirements by an order of magnitude, while longitudinal defocus shifts remain negligible.
Arjun Dey, David J
5 Pith papers cite this work. Polarity classification is still indexing.
5
Pith papers citing it
years
2026 5representative citing papers
SCREAM adapts the CATHODE method to treat stellar streams as feature-space over-densities, incorporates measurement uncertainties into neural network training, and achieves F1=0.745 on GD-1 while recovering faint members and a diffuse cocoon missed by prior methods.
PITA, a new semi-supervised deep learning algorithm, outperforms prior photo-z methods by using a triple-task loss on images, colors, and available redshifts to produce a smooth latent space.
LightCurveLynx is a flexible forward-modeling tool that produces supernova light-curve simulations matching ZTF observations with low KL divergence and consistent completeness limits.
ASTRID simulation with dust model calibrated to SDSS at z=0 produces validated luminosity functions and LSST-ready mock catalogs of 378 million galaxies with predicted number counts in ugrizy bands from z=0 to 2.
citing papers explorer
-
Modeling the impact of filter-substrate refraction in the Roman point spread function
Filter-substrate refraction causes dominant lateral shifts yielding 0.3-0.4% PSF size and ellipticity residuals across most Roman bands that exceed weak lensing requirements by an order of magnitude, while longitudinal defocus shifts remain negligible.
-
Characterizing Stellar Streams with Error-Aware Machine Learning
SCREAM adapts the CATHODE method to treat stellar streams as feature-space over-densities, incorporates measurement uncertainties into neural network training, and achieves F1=0.745 on GD-1 while recovering faint members and a diffuse cocoon missed by prior methods.
-
Optimizing Deep Learning Photometric Redshifts for the Roman Space Telescope with HST/CANDELS
PITA, a new semi-supervised deep learning algorithm, outperforms prior photo-z methods by using a triple-task loss on images, colors, and available redshifts to produce a smooth latent space.
-
LightCurveLynx: Forward Modeling of Time-Domain Surveys with Application to ZTF SN Ia DR2
LightCurveLynx is a flexible forward-modeling tool that produces supernova light-curve simulations matching ZTF observations with low KL divergence and consistent completeness limits.
-
The Galaxy Luminosity Functions in ASTRID: Predictions for LSST
ASTRID simulation with dust model calibrated to SDSS at z=0 produces validated luminosity functions and LSST-ready mock catalogs of 378 million galaxies with predicted number counts in ugrizy bands from z=0 to 2.