AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

CoEvol-NO: State and Coordinate Co-Evolution with an Error-Driven Predictor-Corrector Paradigm for Neural Operator Transformer

Jianqiao Zeng, Ruocheng Wang, Yanzhi Liu, Hao Xiong, Junchi Yan

ICML 2026 · Architectures and efficiency › Attention and transformers · approach: State space and recurrent sequence models

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
update the mesh sequence jointly with the latent state (co-evolution) instead of evolving state only (ablation)CoEvol-NO · Elasticityrelative L2 error: 0.3278 -> 0.0038Improved
keep a persistent latent state instead of recomputing state from mesh each layer (ablation)CoEvol-NO · Navier-Stokesrelative L2 error: 0.1200 -> 0.0731Improved
use exact gradient correction instead of first-order approximation (ablation)CoEvol-NO-dp · Navier-Stokesrelative L2 error: 0.0830 -> 0.0731Improved
add PC correction to Transolver attention and FFNTransolver · Darcyrelative L2 error: 0.0057 -> 0.0043Improved
add PC correction to LNO self-attentionLNO · Airfoilrelative L2 error: 0.0196 -> 0.0082Improved
add PC correction to LNO self-attentionLNO · Piperelative L2 error: 0.0131 -> 0.0141Hurt
add layer-wise momentum on the latent state (beta_S=0.9) (ablation)CoEvol-NO · Piperelative L2 error: 0.0044 -> 0.0032Improved
add momentum on the mesh sequence X (beta_X) (ablation)CoEvol-NO · Pipe, Airfoil, Darcyrelative L2 error: Pipe 0.0044 -> 0.0048 at beta_X=0.9No clear effect
use learnable LayerScale step sizes instead of fixed scale (ablation)CoEvol-NO · Piperelative L2 error: 0.0045 -> 0.0032Improved
add LayerScale to TransolverTransolver · Airfoilrelative L2 error: 0.0057 -> 0.011Hurt
apply PC to all modules with exact gradient (ablation)CoEvol-NO · Piperelative L2 error: 0.0033 -> 0.0049Hurt
remove PC from latent-only CoEvol-NO (ablation)CoEvol-NO-Latent · Airfoilrelative L2 error: 0.0223 -> 0.0069Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy