Two related research papers, AI4Plasma I and II, explore data-efficient surrogate modeling techniques for simulating capacitively coupled plasma (CCP) fields. AI4Plasma I compares five architectures, finding Fourier Neural Operator (FNO) to be a data-efficient baseline for predicting plasma fields across various operating conditions. AI4Plasma II introduces geometry-aware models, including a prescribed-mapping Geo-FNO, to handle variations in geometric parameters like electrode gap, demonstrating significant reductions in required training data for accurate predictions. AI
IMPACT These studies advance the use of AI for scientific simulation, potentially accelerating research and development in fields requiring complex plasma modeling.
RANK_REASON The cluster contains two arXiv papers detailing new research on surrogate modeling techniques for plasma simulations.
- AI4Plasma I
- AI4Plasma II
- Capacitively coupled plasma
- Fourier Neural Operator
- Geo-FNO
- Hao Yang
- Transolver
- U-Net
- Zhenyu Zou
AI-generated summary · Google Gemini · from 2 sources. How we write summaries →