Coupled LSTM-GNN Framework Accelerates Stress Field Reconstruction in Complex Materials
Researchers developed a machine learning model combining recurrent neural networks and graph neural networks to reconstruct stress fields in heterogeneous materials under non-linear loading conditions. The approach achieves 1,000-fold speedup over traditional finite element simulations while generalizing to loading sequences beyond its training data. This advancement could significantly reduce computational bottlenecks in multi-scale material simulations used in engineering and materials science.
A new computational framework couples Long Short-Term Memory (LSTM) networks with physics-informed Graph Neural Networks (GNNs) to reconstruct local stress fields in materials undergoing complex, history-dependent deformation. The LSTM component captures temporal, path-dependent material behavior from macroscopic stress-strain data, while the GNN reconstructs spatially-resolved stress at each time step. The researchers introduced a dynamic weighting strategy that balances data-driven accuracy with physics-based equilibrium constraints, solving a convergence problem in elasto-plastic regimes. Trained on 10,000 non-proportional loading scenarios on a plate-with-hole microstructure, the model achieves three orders of magnitude speedup compared to finite element analysis with only 1.9% cumulative error. A key advantage is mesh-agnosticism: the trained model transfers directly to different element types and mesh resolutions without retraining, while the LSTM hidden states reveal low-dimensional structure aligned with classical constitutive model variables.
Limitations & open questions
The study does not discuss limitations of the approach, such as applicability to other material types beyond von Mises elasto-plasticity, failure modes, or computational requirements for training. No comparison with other surrogate modeling approaches (e.g., convolutional neural networks or other physics-informed methods) is provided. The generalization to loading sequences twice the training length is demonstrated, but long-term stability and performance on fundamentally different loading regimes remain unexplored.
What different sources said
- arXiv cs.LGCenter
Non-linear mechanical field reconstruction coupling recurrent neural networks with physics-informed graph neural networks
Related
Researchers Identify D-Retro-Inverso Peptide Candidates for Treating Cardiac Amyloidosis
Researchers have designed and computationally evaluated four peptide candidates made from D-amino acids that may inhibit the formation of Serum Amyloid A (SAA) fibrils associated with cardiac complications following heart attacks. The study builds on mouse model evidence suggesting SAA aggregates contribute to long-term myocardial infarction complications and may operate similarly in humans. Two candidates, DRI-R5S and DRI-H6A, were identified as particularly promising for potential drug development.
Study Identifies D-Retro-Inverso Peptides as Potential Treatments for Cardiac Amyloidosis
Researchers designed four peptide candidates using D-amino acids to inhibit Serum Amyloid A (SAA) fibril formation, which may contribute to complications following heart attacks. The study, conducted using molecular dynamics simulations in a mouse model, identified two peptides—DRI-R5S and DRI-H6A—as promising drug candidates. This work could lead to new therapeutic approaches for cardiac amyloidosis, a serious post-infarction complication.
Scientists Discover Previously Unknown Branch of Tryptophan Metabolism in Humans
Researchers identified a new enzymatic step in human tryptophan catabolism, showing that the protein ASPDH acts as a 2-aminomuconate reductase to produce a previously unknown amino acid. This discovery fills a gap in the kynurenine pathway, one of the body's major metabolic routes for processing the amino acid tryptophan. The finding expands understanding of human metabolism and may have implications for understanding NAD cofactor production and related metabolic diseases.