Bayesian neural networks correct RANS turbulence models via kinetic energy source terms and anisotropy tensors, improving velocity predictions on training flows but showing reduced accuracy and under-coverage on unseen separated flows.
Title resolution pending
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
years
2026 2verdicts
UNVERDICTED 2representative citing papers
Bayesian PINNs with Hamiltonian Monte Carlo sampling deliver the most consistent uncertainty estimates for turbulent flow inverse problems, while repulsive deep ensembles provide a faster but slightly less calibrated alternative.
citing papers explorer
-
Bayesian neural network correction of RANS turbulence models with uncertainty quantification in separated flows
Bayesian neural networks correct RANS turbulence models via kinetic energy source terms and anisotropy tensors, improving velocity predictions on training flows but showing reduced accuracy and under-coverage on unseen separated flows.
-
Uncertainty Quantification in PINNs for Turbulent Flows: Bayesian Inference and Repulsive Ensembles
Bayesian PINNs with Hamiltonian Monte Carlo sampling deliver the most consistent uncertainty estimates for turbulent flow inverse problems, while repulsive deep ensembles provide a faster but slightly less calibrated alternative.