The Multi-Scale Attention Transformer achieves state-of-the-art accuracy on PDEs with complex geometries, delivering 3.7 times lower error than FNO on a heat benchmark while running inference thousands of times faster than Mamba-NO.
Computer Methods in Applied Mechanics and Engineering , volume =
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
When Attention Beats Fourier: Multi-Scale Transformers for PDE Solving on Irregular Domains
The Multi-Scale Attention Transformer achieves state-of-the-art accuracy on PDEs with complex geometries, delivering 3.7 times lower error than FNO on a heat benchmark while running inference thousands of times faster than Mamba-NO.