arXiv subjects

Thomas Schrefl

Publications and source records attributed to Thomas Schrefl.

At least 19 recordsLinked to original sources

arXiv2026

A Community-Developed Domain Ontology for Magnetic Materials

Magnetic materials play a crucial role in energy-related technologies, mobility, and sensing, but their complex multiscale behaviour and the coexistence of multiple unit systems pose persistent challenges for data exchange and interpretation. This paper presents a domain ontology for magnetic materials, developed within the European Union funded Magnetic Multiscale Modelling Suite (MaMMoS) project and aligned with the Elementary Multiperspective Material Ontology (EMMO). The ontology formalises intrinsic, hysteretic, and microstructural properties across multiple length scales and supports semantic interoperability between simulation tools, databases, and experimental workflows. One key feature of the ontology is its code-based and human-readable structure, enabled through the EMMOntoPy framework, which allows for direct manipulation and versioning without relying on opaque .owl or .ttl files. This facilitates collaborative development and improves transparency. The ontology supports FAIR (Findable, Accessible, Interoperable, Reusable) principles and is openly available for extension by the community. The aim of the MaMMoS project is to foster reproducibility, improve traceability, and promote the adoption of ontology in the magnetism domain. The ontology that has been developed already serves as the foundation for multiple software tools that handle all kinds of magnetic material data.

cond-mat.mtrl-sci↗

arXiv2026

Beyond Stoner-Wohlfarth: Machine-Learning Models and Symbolic Regression of Hard-Magnet Properties

Predicting the extrinsic properties from hysteresis loops of a magnetic grain, namely the coercive field, remanent magnetisation, and maximum energy product, from its intrinsic micromagnetic parameters is a central problem in permanent-magnet modelling. Established analytical models provide useful estimates but often neglect nonuniform magnetisation processes, whereas direct micromagnetic simulations are computationally expensive. In this work, we train machine-learning models on 12012 micromagnetic simulations of an idealised cubic grain, spanning broad ranges of the saturation magnetisation, exchange constant, and uniaxial anisotropy constant. Benchmarked against the analytical models on identical held-out data, the machine-learning models predict all three extrinsic properties with substantially lower errors. Symbolic regression recovers the Kronm\"uller form of the coercive field, with an effective demagnetising factor that depends on the material, and finds new closed-form expressions for the remanence and maximum energy product. Each law contains at most two fitted constants yet approaches the accuracy of the machine-learning models. We also investigate the inverse problem of recovering the intrinsic parameters from the three extrinsic properties. The saturation magnetisation and anisotropy constant are recovered accurately, whereas the exchange constant is not, because it influences the extrinsic properties only weakly. The trained models are released through the mammos-ai Python package, enabling thousands of candidate parameter sets to be screened in seconds rather than the hours or days required by direct micromagnetic simulation.

cond-mat.str-el↗

arXiv2026

Modelling magnetic material properties with uncertainty-aware neural networks

Machine learning is increasingly applied to accelerate the discovery of novel materials by exploring large compositional and structural design spaces. Yet, the scarcity of high-quality data and the frequent need for out-of-distribution prediction introduce substantial uncertainty, making the assessment of model reliability essential. In this work, we investigate uncertainty quantification as a means to evaluate model confidence in the context of permanent magnet research. In a first study, we benchmark classical and modern machine learning models for predicting intrinsic magnetic properties, focusing on the quality of their uncertainty estimates. We apply Gaussian negative log-likelihood loss and dropout-based Bayesian approximation as practical strategies for estimating predictive uncertainty. In a second study, we transfer these architectural features for uncertainty estimation to a more complex task: predicting coercivity from microstructural information using a graph neural network. Together, these studies demonstrate that uncertainty quantification not only enhances the trustworthiness of predictions but is also transferable across different modeling tasks.

cond-mat.mtrl-sci↗

arXiv2026

Tailoring the material properties, nanostructure and grain alignment of Alnico magnets through micromagnetic simulations

Alnico magnets have gained renewed interest in the search for rare-earth free permanent magnets due to their high thermal stability and magnetisation. However, the limited coercivity of these shape-anisotropy-based alloys constrains their performance. Starting from a reference Alnico sample, we realised a finite elements micromagnetic study of exchange-decoupled rods by varying their dimensions and interrod spacing across those observed experimentally. We computed the hysteresis properties by progressing from micromagnetic simulations of a small number of rods within the magnetostatic field of their neighbours to large systems treated statistically based on the distribution of orientations of the grains. We compared the coercivity of an isolated rod with that of the exchange-decoupled system to highlight the effect of magnetostatic interactions. We computed analytically the stray field acting on a single rod as a consequence of its surrounding rods in order to confirm the scaling of the coercivity with the packing fraction p. We explored how intrinsic material properties influence magnetic behaviour by examining materials with different magnetocrystalline anisotropy constants and saturation polarisation values. Results from several hundred simulations were used to train a multi-layer perceptron regressor and predict the magnetic properties as function of the dimensions of the rods, interrod spacing and orientation of the grains. With this approach, we highlight the underlying trends by which nanoscale structuring, intrinsic material properties and grain alignment can be tailored to improve the magnetic properties of Alnico alloys.

cond-mat.mtrl-sci↗

arXiv2026

Thermodynamic Approach for Deciphering Magneto-Structural Phase Transitions: Proof of Concept in Heusler Alloys

Ferromagnetic solids acquire nontrivial magnetic and caloric properties when the temperature of the structural phase transition approaches the Curie point. Deciphering magneto-structural transitions, i.e. determining their characteristic temperatures and elucidating the related properties, remains challenging. In the present paper, three types of transformational behaviour of Ni50Mn25-xCuxGa25 (x = 6.25, 6.5, 6.75, 7) and Ni50.5Mn18.5Cu6.5Ga24.5 alloys have been identified, arising from small variations in chemical composition: (i) structural martensitic transformation (MT) in the ferromagnetic phase; (ii) magneto-structural phase transition from paramagnetic austenite to ferromagnetic martensite; (iii) MT in paramagnetic phase. The temperature-dependent values of magnetization, M(T), and of magnetic susceptibility, $\chi(T)$, were measured for each alloy. A novel thermodynamic analysis was used to determine the Curie points and MT temperatures. The novelty lies in considering the interplay between structural and magnetic characteristics of the alloys through the impact of the structural transition on the spin-exchange parameter. The theoretical analysis of experimental data revealed that this impact results in a large difference ($\geq$ 50 K) between the Curie temperatures computed for the austenitic and martensitic states of each alloy. The characteristic temperatures, corresponding to the extrema of the dM(T)/dT and $\chi(T)$ functions, were calculated. The correlation of these temperatures with the Curie temperatures and the MT temperatures is not straightforward and depends strongly on the type of transformational behaviour (i) - (iii). The proposed approach provides a robust framework for extracting unmeasurable characteristic temperatures from standard magnetization data, applicable to ferromagnetic Heusler systems and other multiferroic ferromagnetic materials.

cond-mat.mtrl-sci↗

arXiv2025

Graph Neural Networks to Predict Coercivity of Hard Magnetic Microstructures

Graph neural networks (GNN) are a promising tool to predict magnetic properties of large multi-grain structures, which can speed up the search for rare-earth free permanent magnets. In this paper, we use our magnetic simulation data to train a GNN to predict coercivity of hard magnetic microstructures. We evaluate the performance of the trained GNN and quantify its uncertainty. Subsequently, we reuse the GNN architecture for predicting the maximum energy product. Out-of-distribution predictions of coercivity are also performed, following feature engineering based on the observed dependence of coercivity on system size.

physics.comp-ph↗

arXiv2025

Modeling liquid-mediated interactions for close-to-substrate magnetic microparticle transport in dynamic magnetic field landscapes

Understanding the on-chip motion of magnetic particles in a microfluidic environment is key to realizing magnetic particle-based Lab-on-a-chip systems for medical diagnostics. In this work, a simulation model is established to quantify the trajectory of a single particle moving close to a polymer surface in a quiescent liquid. The simulations include hydrodynamic, magnetostatic, and Derjaguin-Landau-Verwey-Overbeek (DLVO) interactions. They are applied to particle motion driven by a dynamically changing magnetic field landscape created by engineered parallel-stripe magnetic domains superposed by a homogeneous, time-varying external magnetic field. The simulation model is adapted to experiments in terms of fluid-particle interactions with the magnetic field landscape approximated by analytic equations under the assumption of surface charges. Varying simulation parameters, we especially clarify the impact of liquid-mediated DLVO interactions, which are essential for diagnostic applications, on the 3D trajectory of the particle. A comparison to experimental results validates our simulation approach.

physics.flu-dyn↗

arXiv2025

Effect of interface on magnetic exchange coupling in Co/Ru/Co trilayer: from ab-initio simulations to micromagnetics

Interfaces play a substantial role for the functional properties of structured magnetic materials and magnetic multilayers. Modeling the functional behavior of magnetic materials requires the treatment of the relevant phenomena at the device level. Properties predicted from the electronic structure and spin dynamics at the atomistic level have to be properly transferred into a continuum level treatment. In this work we show how Co/Ru/Co three layers can be simulated with the continuum theory of micromagnetism, with interface coupling energies and bulk intrinsic properties properly derived from the results of \emph{ab initio} and spin dynamics simulations at different temperatures.

cond-mat.mtrl-sci↗

arXiv2024

Realization of inverse-design magnonic logic gates

Magnonic logic gates represent a crucial step toward realizing fully magnonic data processing systems without reliance on conventional electronic or photonic elements. Recently, a universal and reconfigurable inverse-design device has been developed, featuring a 7$\times$7 array of independent current loops that create local inhomogeneous magnetic fields to scatter spin waves in a Yttrium-Iron-Garnet film. While initially used for linear RF components, we now demonstrate key non-linear logic gates, NOT, OR, NOR, AND, NAND, and a half-adder, sufficient for building a full processor. In this system, binary data ("0" and "1") are encoded in the spin-wave amplitude. The contrast ratio, representing the difference between logic states, achieved values of 34, 53.9, 11.8, 19.7, 17, and 9.8 dB for these gates, respectively.

cond-mat.other↗

arXiv2024

NeuralMag: an open-source nodal finite-difference code for inverse micromagnetics

We present NeuralMag, a flexible and high-performance open-source Python library for micromagnetic simulations. NeuralMag leverages modern machine learning frameworks, such as PyTorch and JAX, to perform efficient tensor operations on various parallel hardware, including CPUs, GPUs, and TPUs. The library implements a novel nodal finite-difference discretization scheme that provides improved accuracy over traditional finite-difference methods without increasing computational complexity. NeuralMag is particularly well-suited for solving inverse problems, especially those with time-dependent objectives, thanks to its automatic differentiation capabilities. Performance benchmarks show that NeuralMag is competitive with state-of-the-art simulation codes while offering enhanced flexibility through its Python interface and integration with high-level computational backends.

physics.comp-ph↗

arXiv2024

Physics aware machine learning for micromagnetic energy minimization: recent algorithmic developments

In this work, we explore advanced machine learning techniques for minimizing Gibbs free energy in full 3D micromagnetic simulations. Building on Brown's bounds for magnetostatic self-energy, we revisit their application in the context of variational formulations of the transmission problems for the scalar and vector potential. To overcome the computational challenges posed by whole-space integrals, we reformulate these bounds on a finite domain, making the method more efficient and scalable for numerical simulation. Our approach utilizes an alternating optimization scheme for joint minimization of Brown's energy bounds and the Gibbs free energy. The Cayley transform is employed to rigorously enforce the unit norm constraint, while R-functions are used to impose essential boundary conditions in the computation of magnetostatic fields. Our results highlight the potential of mesh-free Physics-Informed Neural Networks (PINNs) and Extreme Learning Machines (ELMs) when integrated with hard constraints, providing highly accurate approximations. These methods exhibit competitive performance compared to traditional numerical approaches, showing significant promise in computing magnetostatic fields and the application for energy minimization, such as the computation of hysteresis curves. This work opens the path for future directions of research on more complex geometries, such as grain structure models, and the application to large scale problem settings which are intractable with traditional numerical methods.

physics.comp-ph↗

arXiv2024

Experimental realisation of a universal inverse-design magnonic device

In the field of magnonics, which uses magnons, the quanta of spin waves, for energy-efficient data processing, significant progress has been made leveraging the capabilities of the inverse design concept. This approach involves defining a desired functionality and employing a feedback-loop algorithm to optimise the device design. In this study, we present the first experimental demonstration of a reconfigurable, lithography-free, and simulation-free inverse-design device capable of implementing various RF components. The device features a square array of independent direct current loops that generate a complex reconfigurable magnetic medium atop a Yttrium-Iron-Garnet (YIG) rectangular film for data processing in the gigahertz range. Showcasing its versatility, the device addresses inverse problems using two algorithms to create RF notch filters and demultiplexers. Additionally, the device holds promise for binary, reservoir, and neuromorphic computing applications.

physics.app-ph↗

arXiv2023

Physics-informed machine learning and stray field computation with application to micromagnetic energy minimization

We study the full 3d static micromagnetic equations via a physics-informed neural network (PINN) ansatz for the continuous magnetization configuration. PINNs are inherently mesh-free and unsupervised learning models. In our approach we can learn to minimize the total Gibbs free energy with additional conditional parameters, such as the exchange length, by a single low-parametric neural network model. In contrast, traditional numerical methods would require the computation and storage of a large number of solutions to interpolate the continuous spectrum of quasi-optimal magnetization configurations. We also consider the important and computationally expensive stray field problem via PINNs, where we use a basically linear learning ansatz, called extreme learning machines (ELM) within a splitting method for the scalar potential. This reduces the stray field training to a linear least squares problem with precomputable solution operator. We validate the stray field method by means of numerical example comparisons from literature and illustrate the full micromagnetic approach via the NIST $\mu$MAG standard problem $\#3$.

physics.comp-ph↗

arXiv2022

FORC Diagram Features of Co Particles due to Reversal by Domain Nucleation

First Order Reversal Curve (FORC) diagrams are a popular tool in geophysics and materials science for the characterization of magnetic particles of natural and synthetic origin. However, there is still a lot of controversy about the rigorous interpretation of the origin of certain features in a FORC diagram. In this study, we analyze FORCs computed by micromagnetic simulations of Co cubes with dimensions of 50, 100 and 150 nm and uniaxial magnetocrystalline anisotropy. For the larger cubes we observe the formation of a stable two-domain state. The nucleation of a reversed domain and its subsequent annihilation are clearly visible as separate peaks in the FORC diagram. They spread out along the coordinate axis in the FORC diagram, which is associated with the bias field $H_U$ of a Preisach hysteron. Based on our findings, we state that a FORC diagram peak spreading along the $H_U$ axis can have its origin in the step-wise magnetization reversal driven by nucleation of domains in a single particle. This means that we have identified another mechanism apart from the well-known magnetostatic interaction between a set of particles that leads to features in the FORC diagram extending along the $H_U$-axis. Our study demonstrates that if FORCs shall be used as a quantitative tool to assess the microstructure of samples containing magnetic material, more information from other methods will be required to identify the correct physical mechanism by which a certain "fingerprint" in a FORC diagram is produced.

cond-mat.other↗

arXiv2022

Description of collective magnetization processes with machine learning models

This work introduces a latent space method to calculate the demagnetization reversal process of multigrain permanent magnets. The algorithm consists of two deep learning models based on neural networks. The embedded Stoner-Wohlfarth method is used as a reduced order model for computing samples to train and test the machine learning approach. The work is a proof of concept of the method since the used microstructures are simple and the only varying parameters are the magnetic anisotropy axes of the hard magnetic grains. A convolutional autoencoder is used for nonlinear dimensionality reduction, in order to reduce the required amount of training samples for predicting the hysteresis loops. We enriched the loss function with physical information about the underlying problem which increases the accuracy of the machine learning approach. A deep learning regressor is operating in the latent space of the autoencoder. The predictor takes a series of previously encoded magnetization states and outputs the next magnetization state along the hysteresis curve. To predict the complete demagnetization loop, we apply a recursive learning scheme.

physics.comp-ph↗

arXiv2022

Exploring the hysteresis properties of nanocrystalline permanent magnets using deep learning

We demonstrate the use of model order reduction and neural networks for estimating the hysteresis properties of nanocrystalline permanent magnets from microstructure. With a data-driven approach, we learn the demagnetization curve from data-sets created by grain growth and micromagnetic simulations. We show that the granular structure of a magnet can be encoded within a low-dimensional latent space. Latent codes are constructed using a variational autoencoder. The mapping of structure code to hysteresis properties is a multi-target regression problem. We apply deep neural network and use parameter sharing, in order to predict anchor points along the demagnetization curves from the magnet's structure code. The method is applied to study the magnetic properties of nanocrystalline permanent magnets. We show how new grain structures can be generated by interpolation between two points in the latent space and how the magnetic properties of the resulting magnets can be predicted.

cond-mat.mtrl-sci↗

arXiv2021

Full-Spin-Wave-Scaled Finite Element Stochastic Micromagnetism: Mesh-Independent FUSSS LLG Simulations of Ferromagnetic Resonance and Reversal

In this paper, we address the problem that standard stochastic Landau-Lifshitz-Gilbert (sLLG) simulations typically produce results that show unphysical mesh-size dependence. The root cause of this problem is that the effects of spin wave fluctuations are ignored in sLLG. We propose to represent the effect of these fluctuations by a "FUll-Spinwave-Scaled Stochastic LLG", or FUSSS LLG method. In FUSSS LLG, the intrinsic parameters of the sLLG simulations are first scaled by scaling factors that integrate out the spin wave fluctuations up to the mesh size, and the sLLG simulation is then performed with these scaled parameters. We developed FUSSS LLG by studying the Ferromagnetic Resonance (FMR) in Nd$_2$Fe$_{14}$B cubes. The nominal scaling greatly reduced the mesh size dependence relative to sLLG. We further discovered that adjusting one scaling exponent by less than 10% delivered fully mesh-size-independent results for the FMR peak. We then performed three tests and validations of our FUSSS LLG with this modified scaling. 1) We studied the same FMR but with magnetostatic fields included. 2) We simulated the total magnetization of the Nd$_2$Fe$_{14}$B cube. 3) We studied the effective, temperature- and sweeping rate-dependent coercive field of the cubes. In all three cases we found that FUSSS LLG delivered essentially mesh-size-independent results, which tracked the theoretical expectations better than unscaled sLLG. Motivated by these successful validations, we propose that FUSSS LLG provides marked, qualitative progress towards accurate, high precision modeling of micromagnetics in hard, permanent magnets.

physics.comp-ph↗

arXiv2021

Proposal for a micromagnetic standard problem: domain wall pinning at phase boundaries

We propose a novel micromagnetic standard problem calculating the coercive field for unpinning a domain wall at the interface of a multiphase magnet. This problem is sensitive to discontinuities in material parameters for the exchange interaction, the uniaxial anisotropy, and the spontaneous magnetization. We derive an explicit treatment of jump conditions at material interfaces for the exchange interaction in the finite-difference discretization. The micromagnetic simulation results are compared with analytical solutions and show good agreement. The proposed standard problem is well-suited to test the implementation of both finite-difference and finite-element simulation codes.

physics.comp-ph↗