CIS-Transolver: A Cross-Layer Incremental-State Neural Operator for Fine-Scale Automotive Aerodynamic Flow Prediction

Key points

• However, spectral bias and other factors limit accurate reconstruction of fine-scale flow variations near complex geometries.1
• We propose CIS-Transolver, a cross-layer incremental-state neural operator that extends fine-scale modeling from local information extraction to its use across layers.1
• We systematically evaluate CIS-Transolver on three public automotive aerodynamics benchmarks, ShapeNet-Car, AhmedML, and DrivAerNet++, covering surface pressure fields and volumetric pressure and velocity fields.1
• These results provide a new approach to fine-scale flow prediction for complex geometries.1

Sources (1)

• [1]CIS-Transolver: A Cross-Layer Incremental-State Neural Operator for Fine-Scale Automotive Aerodynamic Flow Prediction
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 19, 01:08 PM

  Neural operators offer an efficient data-driven approach to rapid prediction of complex automotive aerodynamic flows.

  However, spectral bias and other factors limit accurate reconstruction of fine-scale flow variations near complex geometries.

Extractive summary: sentences quoted from the sources.