CATO: Charted Attention for Neural PDE Operators

ArXi:2605.09016v1 Announce Type: new Neural operators have emerged as powerful data-driven solvers for PDEs, offering substantial acceleration over classical numerical methods. However, existing transformer-based operators still face critical challenges when modeling PDEs on complex geometries: directly processing over massive mesh points is computationally expensive, while operating in raw discretization coordinates may obscure the intrinsic geometry where physical interactions are naturally expressed. To address these limitations, we