New ML model embeds physics for accurate fracture prediction

Researchers have developed a new machine learning approach called Thermodynamically Consistent Latent Dynamics Identification (tLaSDI) for predicting fracture patterns and evolution in heterogeneous materials. This method embeds thermodynamic principles, such as energy conservation and entropy generation, directly into the neural network architecture using parametric GENERIC Formalism-Informed Neural Networks (pGFINNs). The tLaSDI-pGFINNs model, enhanced with a Geometry-Conditioned Autoencoder, demonstrates superior accuracy in capturing fracture behavior compared to purely data-driven models like Autoencoder Long Short-Term Memory (AE-LSTM). The approach ensures thermodynamic constraints are maintained even on unseen data. AI

IMPACT This approach could lead to more reliable and efficient simulations in materials science and engineering, reducing the need for extensive physical testing.

RANK_REASON The cluster contains an academic paper detailing a new machine learning method for a scientific application. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• AE-LSTM
• arXiv
• Autoencoder Long Short-Term Memory
• GENERIC Formalism-Informed Neural Networks
• Geometry-Conditioned Autoencoder
• pGFINNs
• Thermodynamically Consistent Latent Dynamics Identification
• tLaSDI

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →