{"id":"arxiv/2610.14887","url":"https://arxiv.org/abs/2610.14887","tags":["math.NA"],"score":0,"title":"A Domain-specific adaptation network for elliptic interface problems","domain":"math.NA","source":"arxiv","authors":["Jiaqi Luo","Hongmei Kang","Xingye Yue","Shixin Xu"],"reviews":0,"subject":"math.NA","summary":"Deep learning provides a flexible mesh-free framework for solving elliptic interface problems. However, domain-specific neural approaches often rely on separate networks for individual subdomains, causing the model complexity to grow rapidly as the number of subdomains increases. To address this limitation, we propose a Domain-Specific Adaptation Network (DSAN), which combines a shared backbone with lightweight subdomain-specific adapters to provide local represen- representation flexibility without replicating a complete network for each subdomain. We further establish a stability and error…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-10-10","reactions":{}}