arXiv subjects

Ghatu Subhash

Publications and source records attributed to Ghatu Subhash.

4 recordsLinked to original sources

arXiv2026

A matrix-free, differentiable PyTorch solver for phase-field fracture: Formulation, benchmarks, and inverse analysis

A matrix-free, open-source PyTorch solver is presented for phase-field fracture on central processing units (CPUs) and graphics processing units (GPUs) without custom compiled extensions. In the explicit dynamic pathway, finite-element operations are formulated as element-wise tensor contractions with scatter-based accumulation, removing global sparse mechanics-stiffness assembly from the core time-stepping loop. Both Ambrosio-Tortorelli regularisations (AT1 and AT2), multiple energy decompositions (spectral, volumetric-deviatoric, and star-convex), and plane strain or plane stress assumptions are supported. The explicit mechanics kernels are compatible with PyTorch's automatic differentiation engine (autograd), while the implicit, bound-constrained damage solve is wrapped in a custom backward rule. This rule implements implicit differentiation through the conjugate-gradient (CG) linear solve and keeps memory independent of the internal CG iteration count. The same implementation runs unmodified across macOS, Linux, and Windows, and has been run on meshes of order $10^6$ nodes on a single NVIDIA A100 GPU. The solver is compared against four dynamic fracture cases (straight crack propagation, shear-induced kinking, dynamic branching, and crack-hole interaction in perforated plates) and two quasi-static cases (single-edge notched tension and a notched-holed plate). As a differentiability demonstration, the scalar fracture energy $G_c$ is recovered from observed crack patterns using PyTorch gradients through the forward solve and limited-memory Broyden-Fletcher-Goldfarb-Shanno (L-BFGS) optimisation. Recovery of $G_c$ with relative error below $10^{-3}$ is achieved after three accepted L-BFGS states for glass and two for alumina. The implementation can be extended and combined with differentiable optimisation and machine-learning components.

cs.CE↗

arXiv2024

A Genetic Algorithm Trained Machine-Learned Interatomic Potential for the Silicon-Carbon System

A linear regression-based machine learned interatomic potential (MLIP) was developed for the silicon-carbon system. The MLIP was predominantly trained on structures discovered through a genetic algorithm, encompassing the entire silicon-carbon composition space, and uses as its foundation the Ultra-Fast Force Fields (UF3) formulation. To improve MLIP performance, the learning algorithm was modified to include higher spline interpolation resolution in regions with large potential energy surface curvature. The developed MLIP demonstrates exceptional predictive performance, accurately estimating energies and forces for structures across the silicon-carbon composition and configuration space. The MLIP predicts mechanical properties of SiC with high precision and captures fundamental volume-pressure and volume-temperature relationships. Uniquely, this silicon-carbon MLIP is adept at modeling complex high-temperature phenomena, including the peritectic decomposition of SiC and carbon dimer formation during SiC surface reconstruction, which cannot be captured with prior classical interatomic potentials for this material.

cond-mat.mes-hall↗

arXiv2024

Validation Workflow for Machine Learning Interatomic Potentials for Complex Ceramics

The number of published Machine Learning Interatomic Potentials (MLIPs) has increased significantly in recent years. These new data-driven potential energy approximations often lack the physics-based foundations that inform many traditionally-developed interatomic potentials and hence require robust validation methods for their applicability, accuracy, computational efficiency, and transferability to the intended applications. This work presents a sequential, three-stage workflow for MLIP validation: (i) preliminary validation, (ii) static property prediction, and (iii) dynamic property prediction. This material-agnostic procedure is demonstrated in a tutorial approach for the development of a robust MLIP for boron carbide (B4C), a widely employed, structurally complex ceramic that undergoes a deleterious deformation mechanism called "amorphization" under high-pressure loading. It is shown that the resulting B4C MLIP offers a more accurate prediction of properties compared to the available empirical potential.

cond-mat.mtrl-sci↗

arXiv2023

Fluid-mediated impact of soft solids

A viscous, lubrication-like response can be triggered in a thin film of fluid squeezed between a rigid and flat surface and the tip of an incoming projectile. We develop a comprehensive theory for this viscous approach stage of fluid-mediated normal impact, applicable to soft impactors. Under the assumption of mediating fluid being incompressible, the impacting solid displays two limit regimes: one dominated by elasticity and the other by inertia. The transition between the two is predicted by a dimensionless parameter, which can be interpreted as the ratio between two time scales that are the time that it takes for the surface waves to warn the leading edge of the impactor of the forthcoming impact, and the characteristic duration of the final viscous phase of the approach. Additionally, we assess the role of solid compressibility and elucidate why nearly-incompressible solids feature (a) substancial "gliding" prior to contact at the transition between regimes, (b) the largest size of entrapped bubble between the deformed tip of the impactor and the flat surface, and (c) a sudden drop in entrapped bubble radius past the transition between regimes. Finally, we argue that the above time scale ratio (a dimensionless number) can govern the different dynamics reported experimentally for a fluid droplet as a function of its viscosity and surface tension.

cond-mat.soft↗