Skip to content

Optimization, inverse problems and uncertainty

English | Japanese

« Home

Adjoint gradients, parameter identification and uncertainty propagation. Verified where the true value is known.

4 cases.

coldplate-moo
Three-level simultaneous multi-objective optimization of a 700 W-class GPU liquid cold plate on 1 Miyabi-G GH200 node (1D 2,000 → Grace CPU 64-core FVM 116 → Hopper H100 3.07M-cell unsteady thermal LBM 15).
adjoint-poiseuille2d
Discrete-adjoint LBM sensitivity on body-force-driven plane Poiseuille flow, validated cell-for-cell against central finite differences.
param-id-poiseuille2d
Parameter identification (inverse problem) on body-force-driven plane Poiseuille flow: recover the unknown kinematic viscosity.
uq-poiseuille2d
Forward uncertainty quantification (UQ) around the plane-Poiseuille LBM forward map.