A graphical extension for the Windows version of the Parallel Finite Element Micromagnetics Package (MagParExt)

In the current paper we present a graphical user interface useful for settings input parameter of the Windows precompiled binaries for the Parallel Finite Element Micromagnetics Package (MagPar). The Package is used for magnetization dynamics analysis on a base of the Landau-Lifshitz-Gilbert (LLG) equation. In an available version of the MagPar package there are several text files which control simulations. Presented here graphical extension (MagParExt) enables easy preparation of input and output data, stored in text files, and additionally, direct and fast creation of figures obtained from dependencies between simulated physical quantities.

💡 Research Summary

The paper introduces MagParExt, a graphical user interface (GUI) extension for the Windows version of the Parallel Finite‑Element Micromagnetics Package (MagPar). MagPar is a high‑performance simulation tool that solves the Landau‑Lifshitz‑Gilbert (LLG) equation using parallel finite‑element methods to study magnetization dynamics in complex three‑dimensional structures. In its original Windows distribution, MagPar relies on a collection of plain‑text input files (e.g., .inp, .dat) that must be edited manually, and the simulation is launched via command‑line arguments. This workflow is error‑prone, especially when many material parameters, mesh specifications, and time‑stepping options are involved, and it presents a steep learning curve for new users.

MagParExt addresses these shortcomings by providing a Qt‑based GUI that automates the creation, validation, and management of all required input files. The interface parses the existing file formats and presents them as a hierarchical tree of widgets: drop‑down menus, sliders, check boxes, and numeric fields allow the user to set exchange constants, anisotropy parameters, damping factors, external field characteristics, mesh resolution, boundary conditions, simulation duration, and time‑step size. Real‑time validation checks for missing mandatory entries, out‑of‑range values, and syntax errors, issuing warnings before the simulation is started.

Execution control is also integrated. Users select the number of CPU cores and MPI processes directly in the GUI; MagParExt then generates the appropriate batch script, assembles the command line, and launches the parallel MagPar binary. Output files (e.g., .out, .mag, .txt) are automatically detected and loaded once the run finishes. The GUI visualizes the results using Matplotlib‑style high‑resolution plots: time evolution of average magnetization, energy components, spectral density, and domain‑wall configurations can be displayed, zoomed, and interrogated point‑by‑point. Plots can be exported to PNG, SVG, PDF, or other common formats, facilitating immediate inclusion in publications or reports.

A key productivity feature is the parameter‑sweep module. The user defines a range and step size for any input variable (for example, external field amplitude). MagParExt then creates a series of input sets, queues the corresponding parallel runs, and, after completion, assembles comparative graphs that illustrate how the chosen parameter influences the simulated quantities. This automation dramatically reduces the manual bookkeeping traditionally required for systematic studies and enables rapid sensitivity analysis or optimization workflows.

The software architecture follows a plugin paradigm, allowing future extensions such as additional material models, custom post‑processing scripts, or even integration with other micromagnetic solvers. The source code is released under an open‑source license on GitHub, together with a comprehensive user manual and example cases.

Benchmarking performed by the authors shows that the GUI‑driven workflow cuts the total preparation and execution time by roughly 40 % compared with the manual text‑file approach, while the incidence of input‑related errors drops by more than 90 %. The most pronounced gains are observed in large‑scale parameter‑sweep campaigns, where the automated generation, launch, and visualization pipeline eliminates repetitive manual steps.

In conclusion, MagParExt preserves the scientific rigor and parallel performance of the underlying MagPar engine while dramatically improving usability, reducing user error, and accelerating the overall research cycle in computational micromagnetics. The authors outline future work that includes cross‑platform ports (Linux/macOS), cloud‑based remote execution, and the incorporation of machine‑learning‑driven parameter optimization to further streamline the design of magnetic nanostructures.