FG-MoE: Benchmark Data for Turbulence Model Validation

FG-MoE is a Factorized-Gating Mixture of Experts framework designed to capture complex turbulent behaviors across different flow regimes. This dataset, created by HAOCHEN LIU and last updated in April 2026, is used to run benchmark test cases and reproduce results from the associated research paper. The data likely contains simulation outputs or parameters for validating the regime-specific expert models.

Use Cases

• Validate the FG-MoE turbulence model's performance based on benchmark test cases.
• Reproduce research findings from the associated paper based on the provided data.
• Compare regime-specific expert predictions based on spatially varying gating probabilities.
• Analyze cross-regime generalization capabilities based on the factorized gating mechanism.

Strengths

• Dataset is associated with a published research framework (FG-MoE) designed for interpretability and generalization.
• License is explicitly stated as Apache-2.0, facilitating open use and modification.
• Last updated date is provided (2026-04-27), indicating recent maintenance.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count and file size are unknown, which may limit suitability assessment.
• Data scope is limited to benchmark test cases for the specific FG-MoE model.

Provenance

Source

HAOCHEN LIU via figshare.

Collection Method

Likely generated from computational fluid dynamics simulations or model runs.

Freshness

Last updated 2026-04-27 08:01:20.