FlowReconBench
A benchmark for evaluating reconstruction of complete flow fields from sparse sensor data, covering three flow regimes: laminar cylinder wake, two-dimensional Kolmogorov turbulence, and three-dimensional isotropic turbulence. It compares learned methods and classical estimators in terms of accuracy, calibration, and cost.
Timeline (last 90 days)
FlowReconBench evaluated ten learning methods and five classical methods across three flow regimes (laminar cylinder wake, two-dimensional Kolmogorov turbulence, and large-scale three-dimensional isotropic turbulence).
The core conclusion of FlowReconBench is that there is no single optimal solution for flow field reconstruction, and method selection must jointly consider error, calibration, spectral fidelity and computational cost.
All Facts (2)
FlowReconBench evaluated ten learning methods and five classical methods across three flow regimes (laminar cylinder wake, two-dimensional Kolmogorov turbulence, and large-scale three-dimensional isotropic turbulence).
50%UnverifiedThe core conclusion of FlowReconBench is that there is no single optimal solution for flow field reconstruction, and method selection must jointly consider error, calibration, spectral fidelity and computational cost.
50%