Vision transformer models segment proto-halo regions in initial cosmological density fields by final mass at z=0, outperforming both CNNs and the PINOCCHIO perturbation-theory code.
Lucie-Smith, H.V
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
fields
astro-ph.CO 2verdicts
UNVERDICTED 2representative citing papers
Transformer and GBDT models trained on AbacusSummit mocks for DESI LRGs/ELGs recover nonlinear velocity power spectra and cross-correlations better than linear theory across a wider range of scales, with applications to kSZ analyses.
citing papers explorer
-
Segmenting proto-halos with vision transformers
Vision transformer models segment proto-halo regions in initial cosmological density fields by final mass at z=0, outperforming both CNNs and the PINOCCHIO perturbation-theory code.
-
Full Nonlinear Velocity Reconstruction With Transformer and Ensemble Tree Machine Learning
Transformer and GBDT models trained on AbacusSummit mocks for DESI LRGs/ELGs recover nonlinear velocity power spectra and cross-correlations better than linear theory across a wider range of scales, with applications to kSZ analyses.