A Fourier neural operator trained on Boussinesq-compressible simulation pairs corrects Boussinesq predictions for natural convection, achieving SSIM near unity and MSE reductions of one to three orders of magnitude.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
physics.comp-ph 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
A Neural Surrogate Approach for Simulating Natural Convection Problems
A Fourier neural operator trained on Boussinesq-compressible simulation pairs corrects Boussinesq predictions for natural convection, achieving SSIM near unity and MSE reductions of one to three orders of magnitude.