Papers & Surveys

Original Research Paper

Reframes Transolver's Physics-Attention as a special case of linear attention and proposes LinearNO (Linear Attention Neural Operator); achieves SOTA on six PDE benchmarks with 40.0% fewer parameters and 36.2% less compute, plus superior performance on AirfRANS and Shape-Net Car industrial datasets.

Materials

Presents FRPN, a Full-Resolution Polymer Network that retains both atom-level details and monomer-level connectivity for multi-scale polymer property prediction.

Materials

Investigating nonequilibrium phase-ordering kinetics in binary mixtures of uniaxial ellipsoidal particles using large-scale molecular dynamics simulations with anisotropic Gay-Berne potential, revealing two-stage ordering processes across different time scales after temperature quench.

ScientificML

LPINNs propose multiplying the first hidden activation of a dense PINN by input-dependent localization functions, endowing first-layer units with receptive fields to improve optimization on long domains and for high-order operators without domain partitioning.

Multibody

This paper proposes a data-driven policy construction method for dynamical systems using non-expert MPC demonstrations, introducing a memory-based nonparametric policy with Lipschitz-regularized upper envelope of the finite-horizon surrogate value function to approximate infinite-horizon optimal control.

ScientificML

This paper investigates using artificial neural networks as surrogate models for black-box optimization, enabling efficient optimization when analytical function descriptions are unavailable by obtaining objective values through numerical simulation.

ScientificML

Introduces a decision-epoch-based framework for dynamic electric autonomous dial-a-ride, jointly optimizing vehicle routing, scheduling, and charging decisions using combinatorial optimization and machine learning.

Quantum/Chemical

For quantum many-body physics and ab initio quantum chemistry, this work derives an unbiased direct gradient estimator and introduces the adaptive minimum-variance phase (AMVP) estimator to overcome training failures in weak-gradient regimes for neural-network variational optimization.

Geophysics

This paper presents a domain-decomposed random-feature evolutional deep neural network (DD RF-EDNN) for solving transient pressure diffusion in heterogeneous porous media with discontinuous and high-contrast permeability, combining spatial approximation with temporal evolution.

Computational Geometry

This paper presents smooth spherical splines combined with conformal factors to solve Poisson equations and eigenvalue problems for the Laplace-Beltrami operator on genus-zero Riemannian surfaces.

ScientificML

Evaluates six sleep staging foundation models under reduced PSG channel conditions, providing insights for lightweight sensing solutions in long-term home monitoring.

ScientificML

Proposes a unified scalable framework combining structured decomposition with SDP certificates and small-gain analysis for robust stability and performance analysis of feedback systems with high-dimensional NNs and unmodeled dynamics.

Materials

A method to predict how force correction changes energy exchange and structural statistics in molecular dynamics by measuring leading response coefficients on shared reference trajectories.

Materials

Hybrid quantum neural network extracts quantitative parameters like phase weight fractions directly from 1D powder diffraction patterns without iterative refinement, suitable for complex multiphase materials and noisy measurements.

Multibody

Proposes AffordanceWAM, an affordance-aware generative World Action Model that leverages both action-labeled robot videos and egocentric human videos for generalizable robot manipulation.

CFD

Diff-NekRS is an open-source framework that enables automatic differentiation through the spectral element CFD solver NekRS, enabling gradient backpropagation across multi-timestep rollouts to train neural correction networks for error mitigation.

Structural

A mesh-free KAN-PINN method for solving nonlinear strain-limiting partial differential equations, featuring exact hard boundary ansatz and pseudo-Huber residual loss for enhanced training stability.

CFD

This paper presents an integrated CFD-machine learning framework for optimizing pod geometry in supersonic evacuated tube transport, using composite cubic Bezier curves with seven design variables and Latin Hypercube Sampling to evaluate 200 configurations across subsonic and supersonic regimes.

Structural

DeepFEAv2 applies deep learning surrogate models to transient finite element analysis on unstructured meshes, enabling joint prediction of node-based and element-based outputs to reduce computational cost in structural mechanics.

CFD

FAST-ML is a hybrid framework combining data-driven efficiency with physical constraints for tropical cyclone rapid intensification forecasting.