AI Paper Atlas

Research map / Architectures and efficiency / RNNs and normalization

Smooth Dynamic Cutoffs for Machine Learning Interatomic Potentials

Kevin Han, Haolin Cong, Bowen Deng, Amir Barati Farimani

ICML 2026 · Architectures and efficiency › RNNs and normalization · approach: Scalable approximation and equivariant compute · 0 citations (OpenAlex)

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
use dynamic cutoff instead of fixed cutoff (mu=40)MACE, Nequip, Orbv3, TensorNet · MatPES-r2scanenergy and force MAE: MACE force 173 -> 190; Nequip force 167 -> 168Mixed
use dynamic cutoff instead of fixed cutoff (mu=20)MACE, Nequip, Orbv3, TensorNet · MD22energy and force MAE: DWNT force MACE 35.60 -> 41.22Hurt
use dynamic cutoff to induce graph sparsityfour MLIPs · MOF and protein benchmarksmemory and inference time: up to 2.26x less memory, 2.04x fasterImproved
use dynamic cutoff to induce graph sparsityMLIP edge count · MOF material systemnumber of edges: 677000 -> 297000Improved
use dynamic cutoff to induce graph sparsityMLIP edge count · protein systemnumber of edges: 805040 -> 588141Improved
replace hard max-neighbor limit with smooth dynamic cutoffTensorNet · NVE MD stabilityenergy drift / PES smoothnessImproved
use dynamic cutoff in 100 ps NVE simulations (DWNT, LiFePO4 at 3000K)four MLIPs · NVE MDtotal energy driftImproved
increase target neighbors mu from 5 to 40 (ablation)TensorNet 456k · MD22 buckyball catcher and DWNTforce MAEImproved
use dynamic cutoff instead of reduced fixed cutoff with equal average neighbors (ablation)TensorNet · MatPES-r2scanenergy and force MAE: energy 53 -> 43; force 171 -> 160Improved
use dynamic cutoff instead of reduced fixed cutoff (ablation)copper melting 4500K · 864-atom Cu meltaverage neighbors per atomImproved
dynamic cutoff on elements with large atomic radiiTensorNet · MatPES per-element force errorrelative force MAE increaseHurt

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