aiqbench analysis
This paper applies physics-informed machine learning to predict fracture evolution in heterogeneous materials, using neural networks that hard-enforce energy conservation and entropy generation in a learned latent space. The authors claim better accuracy than a purely data-driven AE-LSTM baseline and preserved thermodynamic constraints on unseen samples, but these are results from a single arXiv preprint on specific simulation setups. Readers should watch whether the approach generalizes beyond the tested geometries and whether it reduces the training data needed relative to standard surrogates.
Thermodynamically Consistent Machine Learning for Fracture Prediction in Heterogeneous Materials
arXiv:2610.20614v1 Announce Type: cross Abstract: Understanding fracture behavior is critical for assessing the structural integrity and longevity. Highly time-consuming physics-based fracture simulations motivate the development of machine-learning-based fracture evolution prediction with surrogate models. However, most existing machine-learning models for fracture prediction operate as black boxes and require large training datasets to generalize reliably. In this work, Thermodynamically Consistent Latent Dynamics Identification (tLaSDI) for parametric systems is introduced to predict fracture patterns and evolution in heterogeneous materials. The model learns the thermodynamic state variable in a lower-dimensional latent space using parametric GENERIC Formalism-Informed Neural Networks (pGFINNs) that enforce key thermodynamic principles, such as total energy conservation and entropy generation. This model embeds physics into the neural network architecture using a hard-constraint method that inherently complies with the underlying thermodynamic principles. The proposed tLaSDI-pGFINNs is further enhanced by a Geometry-Conditioned Autoencoder in learning the geometry configuration for effective geometry-guided learning. The energy and entropy evolution histories demonstrate that the model preserves the thermodynamic constraints on unseen testing samples. The results further show that tLaSDI-pGFINNs can more accurately capture both fracture patterns and the subsequent fracture evolution than the purely data-driven Autoencoder Long Short-Term Memory (AE-LSTM) model.
Abstract:Understanding fracture behavior is critical for assessing the structural integrity and longevity. Highly time-consuming physics-based fracture simulations motivate the development of machine-learning-based fracture evolution prediction with surrogate models. However, most existing machine-learning models for fracture prediction operate as black boxes and require large training datasets to generalize reliably. In this work, Thermodynamically Consistent Latent Dynamics Identification (tLaSDI) for parametric systems is introduced to predict fracture patterns and evolution in heterogeneous materials. The model learns the thermodynamic state variable in a lower-dimensional latent space using parametric GENERIC Formalism-Informed Neural Networks (pGFINNs) that enforce key thermodynamic principles, such as total energy conservation and entropy generation. This model embeds physics into the neural network architecture using a hard-constraint method that inherently complies with the underlying thermodynamic principles. The proposed tLaSDI-pGFINNs is further enhanced by a Geometry-Conditioned Autoencoder in learning the geometry configuration for effective geometry-guided learning. The energy and entropy evolution histories demonstrate that the model preserves the thermodynamic constraints on unseen testing samples. The results further show that tLaSDI-pGFINNs can more accurately capture both fracture patterns and the subsequent fracture evolution than the purely data-driven Autoencoder Long Short-Term Memory (AE-LSTM) model.
Computational Physics (physics.comp-ph); Materials Science (cond-mat.mtrl-sci); Computational Engineering, Finance, and Science (cs.CE); Machine Learning (cs.LG)
arXiv:2610.20614 [physics.comp-ph]
(or arXiv:2610.20614v1 [physics.comp-ph] for this version)
Full text under the publisher's open licence · 212 words
Related coverage
Synthesis Superintelligence: from Semiconductors to Superconductors — Periodic Labs’ Liam Fedus and Ekin Dogus Cubuk
Latent Space · Oct 8, 2026
Reactivity-Informed Machine Learning for Performance Prediction and Design Space Exploration of Alkali-Activated Slag
arXiv · Jun 8, 2026
News
Quantum Computing Report · Jul 25, 2026
SPLIT-PINN: Separable Probability Learning Technique via Physics-Informed Neural Networks for High-Dimensional Probabilistic Modeling
arXiv · Jun 5, 2026