Search arXiv⌕ Search
Home/Archive/arXiv

arXiv · 2610.33711

A Differentiable Framework for Inverse Identification of Heterogeneous Cohesive Laws

Sida Hao·Jinkyo Han·Bahador Bahmani
View original source ↗Read source document ↗Export metadata ↓

Abstract

This work develops a differentiable finite element framework based on an intrinsic cohesive-zone formulation for identifying traction-separation laws from global force-displacement histories and spatially sparse strain measurements. The framework employs physics-constrained neural networks to represent flexible, thermodynamically structured cohesive laws, with sensitivities computed through automatic differentiation and a discrete adjoint that accounts for the history-dependent evolution of interfacial damage. To enable identification of spatially heterogeneous cohesive behavior, we extend learned constitutive priors to irreversible interfacial damage by constructing a shared, low-dimensional family of cohesive laws from representative traction-separation histories. This representation exploits common constitutive structure and parameter sharing to reduce the number of unknowns and regularize the inverse problem, while allowing local cohesive behavior to vary spatially through latent coordinates. Numerical investigations involving double-cantilever beams and fiber-matrix composites demonstrate that accurate reproduction of global force-displacement responses does not necessarily imply accurate recovery of the underlying cohesive laws. Incorporating spatially resolved strain measurements substantially improves constitutive identification, even when the admissible responses are constrained by a learned prior. Comparisons with conventional bilinear cohesive models and Bayesian optimization further illustrate the benefits of the learned constitutive representation and adjoint-based identification for high-dimensional inverse problems.

Explore related subjects

physics.comp-phcs.CE

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Sida Hao, Jinkyo Han, Bahador Bahmani. 2026-10-24. A Differentiable Framework for Inverse Identification of Heterogeneous Cohesive Laws. https://arxiv.org/abs/2610.33711

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

arXiv2026

Physics-Informed Neural Networks for coupled stiff transport systems

Physics-Informed Neural Networks (PINN) struggle with stiff, regime-changing transport equations due to instability, loss imbalance, and violations of physical consistency. This paper investigates these failures through the Marshak wave equations, a canonical benchmark from radiative transport, where initial and boundary conditions differ by up to 12 orders of magnitude, and proposes modifications to the standard PINN framework to overcome them. Three modifications are introduced: a ScaledSigmoid final activation enforcing physical bounds and positivity of the unknowns; a logarithmic MSE loss replacing the standard quadratic loss for initial and boundary conditions, enabling training across extreme scale disparities; and explicit enforcement of global conservation laws as an additional physics loss term. Monte Carlo sampling with exponential time weighting is used throughout. The proposed framework recovers the Marshak wave dynamics - including the hot, cold, and wave-front regions - in good agreement with a reference Implicit Monte Carlo solution (sub-millimeter absolute error on the front position, although the relative error remains substantial at early times), with run times under 5 minutes. Ablation studies confirm that each ingredient is essential: linear activation, absence of the logarithmic loss, or removal of the PDE term each independently cause the method to fail qualitatively. This work identifies and resolves three concrete failure modes of standard PINN on stiff hyperbolic systems with nonlinear coupling. The combination of bounded activations, scale-aware loss functions, and conservation law enforcement constitutes a framework validated on the present one-dimensional Marshak wave problem, with potential extension to other coupled stiff transport systems in engineering.

physics.comp-ph↗
arXiv2026

Towards Exact Multidimensional Fluxes: A Structure-Preserving Data-Informed Framework for Riemann Solvers

Finite-volume and discontinuous Galerkin methods update cells using the space-time flux integral over their faces, which typically depends on the solutions to local Riemann problems. In the multidimensional Riemann problem setting, this integral is difficult to evaluate exactly because of complex wave interactions. In this work, we show that the exact space-time flux integral is a polynomial in the time step and face dimensions. This motivates a learning-based approach, as only the mapping from the initial states to the polynomial coefficients needs to be learned. We propose a structure-preserving, data-informed Riemann solver framework that uses a neural network trained on high-resolution simulation data to learn this mapping. Key structural properties, including consistency and symmetry, are enforced by construction. Provided that the high-resolution simulations converge to the exact flux integrals and the network accurately predicts the coefficients, this framework yields near-exact flux integrals for use in numerical simulations. We demonstrate the framework using classical implosion and explosion problems for the two-dimensional Euler equations. In an unsplit MUSCL-Hancock scheme, the learned Riemann solver remains stable at a CFL number of 0.9, whereas the same scheme using the exact one-dimensional Riemann solver fails. Compared with the latter scheme at CFL=0.45, the learned scheme is slightly more accurate and requires approximately 30% less computational time.

physics.comp-ph↗
arXiv2026

Machine learning kinetics from molecular dynamics data

Most molecular transitions occur on time scales far beyond direct molecular dynamics simulations. The committor, the probability that a configuration reaches a product state before a reactant state, is a central kinetic statistic, providing a mechanism-independent reaction coordinate and a foundation for transition path theory and the calculation of rates. This review surveys modern approaches for estimating the committor and related kinetic statistics from molecular simulations, with an emphasis on self-supervised methods that learn solutions of their defining dynamical equations rather than relying on labeled shooting data. We develop a common operator viewpoint connecting generator-based partial differential equations, variational principles, Markov state models, dynamical Galerkin approximation, and neural networks. Empirical and theoretical evidence points to the efficiency of these methods. We provide theoretical and practical guidance for realizing their full potential in applications, including strategies for treating non-Markovian effects and for sampling. We conclude by identifying opportunities for further research, including connections to reinforcement learning and generative modeling.

physics.comp-ph↗