[論文レビュー] soliton_solver: A GPU-based finite-difference PDE solver for topological solitons in two-dimensional non-linear field theories

Paul Leask|arXiv (Cornell University)|Mar 25, 2026

Scientific Research and Discoveries被引用数 0

ひとこと要約

GPU-accelerated, Python framework with a theory-agnostic finite-difference core for simulating and visualizing 2D topological solitons across multiple models via modular theory components.

ABSTRACT

This paper introduces soliton_solver, an open-source GPU-accelerated software package for the simulation and real-time visualization of topological solitons in two-dimensional non-linear field theories. The software is structured around a theory-agnostic numerical core implemented using Numba CUDA kernels, while individual physical models are introduced through modular theory components. This separation enables a single computational framework to be applied across a broad class of systems, from nanoscale magnetic spin textures in condensed matter physics to cosmic strings spanning galaxies in high energy physics. The numerical backend provides finite-difference discretization, energy minimization, and GPU-resident evaluation of observables. A CUDA--PyOpenGL rendering pipeline allows direct visualization of evolving field configurations without staging full arrays through host memory. The package is distributed in Python via PyPI and supports both reproducible batch simulations and interactive exploration of metastable configurations, soliton interactions, and model-dependent initial states. We describe the software architecture, numerical workflow, and extensibility model, and we present representative example applications. We also outline how additional theories can be incorporated with minimal modification of the shared numerical infrastructure.

研究の動機と目的

• Motivate the study of topological solitons across diverse physical systems and emphasize the need for a reusable computational tool.
• Provide a theory-agnostic GPU backend that can drive multiple nonlinear field theories via modular theory components.
• Enable real-time visualization and interactive exploration of metastable configurations and soliton interactions.
• Deliver an open-source package (PyPI) that supports both batch simulations and rapid prototyping of new models.

提案手法

• GPU-accelerated finite-difference engine for two-dimensional fields with a shared parameter system and halo-augmented stencils.
• Arrested Newton flow for energy minimization to efficiently relax to static soliton configurations.
• CUDA kernels implemented via Numba CUDA with a theory registry to load models at runtime.
• CUDA–PyOpenGL rendering pipeline for interactive, host-memory-free visualization of evolving fields.
• Modular theory interface where each model provides field definitions, energy functionals, parameters, initializations, and optional visualizers.

Figure 1 : Typical usage of soliton_solver for a micromagnetic simulation of anti-skyrmions in Heusler compounds, modelled by a chiral ferromagnet with the Dzyaloshinskii–Moriya interaction (DMI), and including the effects of demagnetization.

実験結果

リサーチクエスチョン

• RQ1 How can a single computational framework support diverse two-dimensional nonlinear field theories with topological solitons?
• RQ2 What is the impact of arrested Newton flow on the speed and robustness of soliton relaxation compared with standard gradient descent?
• RQ3 Can real-time GPU-resident visualization be integrated with the solver to support interactive exploration across different theories?
• RQ4 How easily can new theories be added to the framework with minimal modification to the shared numerical infrastructure?

主な発見

• The package delivers a reusable computational framework with a theory-agnostic core and a modular theory interface.
• Built-in theories cover Abelian Higgs cosmic strings, vortices in Ginzburg–Landau superconductors, anyons, skyrmions in various media, and baby skyrmions, among others.
• The framework provides a GPU-resident, CUDA–PyOpenGL visualization backend suitable for real-time inspection of evolving configurations.
• Arrested Newton flow accelerates relaxation by exploiting a fictitious time dynamics and stopping when energy increases, improving efficiency for multi-soliton problems.
• New models can be added by implementing a compact theory module; the solver core handles grid construction, differentiation, time-stepping, and rendering.
• The workflow supports interactive exploration and batch simulations, with installation and usage via PyPI or source.

Figure 2 : A single anti-skyrmion in the bulk of a Heusler compound, with the associated Dzyaloshinskii–Moriya interaction and backreaction of the demagnetization field. All panels are in dimensionless units. The left panel shows the energy density $\mathcal{E}(\vec{x})$ , the middle shows the magne