Searcharxiv⌕ Search

arXiv subjects

Hehua Zhu

Publications and source records attributed to Hehua Zhu.

3 recordsLinked to original sources

OpenPINN/PINO: An Extensible Open-Source Framework for Agent-Assisted Physics-Informed Neural Networks and Neural Operators

Substantial challenges in physics-informed learning have motivated many improvements, but fragmented implementations hinder systematic organization, integration, and comparison. We propose OpenPINN/PINO, an extensible open-source PyTorch framework for physics-informed neural networks and neural operators. Two shared specifications connect problem setup, local computation, and evaluation. Agents support problem formulation, solver planning, and diagnosis. ProblemSpec defines geometry, governing equations, constitutive relations, parameters, initial and boundary conditions, and optional data. MethodSpec configures the model, sampling, loss, optimization, and differentiation. Researchers can register new problem templates and method components, then combine or compare strategies such as Fourier features, residual-adaptive sampling, and self-adaptive loss weighting for a fixed problem. Agents translate natural-language descriptions into specifications, flag missing information, and propose solver plans. The framework validates these specifications and deterministically generates local training and testing code. A diagnostic agent reviews training logs and evaluation results to identify problems and suggest revisions. Three examples show how method choices and agent diagnostics affect accuracy. For viscous Burgers, Fourier features and residual-adaptive sampling each reduced relative $L^2$ error by about $75\%$ from a shared baseline. For periodic Poisson, spectral differentiation reduced mean-free relative $L^2$ error by about $68\%$ across 16 held-out forcings. For a plate with a hole, successive agent diagnoses guided nondimensionalization and optimization. After initial training failure, all three runs met the $10\%$ target, with relative $L^2$ displacement errors of $3.51\%$--$6.31\%$.

physics.comp-ph↗

Dual-support smoothed particle hydrodynamics in solid: variational principle and implicit formulation

In this paper, we derive the dual-support smoothed particle hydrodynamics (DS-SPH) in solid in the framework of variational principle. The tangent stiffness matrix of SPH is obtained with ease, which can be served as the basis for implicit SPH. We propose a hourglass energy functional, which allows the direct derivation of hourglass force and hourglass tangent stiffness matrix. The dual-support is identified in all derivation based on variational principles and is automatically satisfied in the assembling of stiffness matrix. The implementation of stiffness matrix comprises with two steps, the nodal assembly based on deformation gradient and global assembly. Several numerical examples are presented to validate the method.

physics.comp-ph↗

Phase field modelling of crack propagation, branching and coalescence in rocks

We present a phase field model (PFM) for simulating complex crack patterns including crack propagation, branching and coalescence in rock. The phase field model is implemented in COMSOL and is based on the strain decomposition for the elastic energy, which drives the evolution of the phase field. Then, numerical simulations of notched semi-circular bend (NSCB) tests and Brazil splitting tests are performed. Subsequently, crack propagation and coalescence in rock plates with multiple echelon flaws and twenty parallel flaws are studied. Finally, complex crack patterns are presented for a plate subjected to increasing internal pressure, the (3D) Pertersson beam and a 3D NSCB test. All results are in good agreement with previous experimental and numerical results.

cond-mat.soft↗