Spectral neighbor representation for vector fields: Machine learning potentials including spin

Domina, Michelangelo; Cobelli, Matteo; Sanvito, Stefano

doi:10.1103/physrevb.105.214439

Cited by 17 publications

(16 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Tailoring the Curie temperature during the fitting process is limited as it is not a directly trainable quantity and relates to the curvature of the exchange interactions. Improvements could be made by training future models using more complete feature representations such as spectral neighbours for vector fields derived recently by Domina et al 34 . One may also observe our model produces the classical profile of the magnetisation curve despite careful training to accurate quantum mechanical data from DFT calculations.…”

Section: Resultsmentioning

confidence: 99%

“…Novikov et al 38 developed a moment tensor spin-lattice potential that includes longitudinal fluctuation, but they limited their approach to collinear configurations near perfect crystal structures. Domina et al 34 extended the spectral-neighbour representation to be applicable to non-unit vector fields such as spin. Whilst no dynamics was performed, their approach shows an excellent ability to predict the energies of non-collinear states relative to a prototype model of iron for configurations with small atomic displacements from the perfect BCC lattice.…”

Section: Introductionmentioning

confidence: 99%

“…Machine-learned potentials for spin-lattice dynamics that go beyond the need of a well defined functional form could be a viable solution 10,34 . While the number of machine-learned potentials for iron has rapidly increased over the past decade 3,[35][36][37] , applications including explicit spin degrees of freedoms are very limited.…”

mentioning

confidence: 99%

See 2 more Smart Citations

A machine-learned spin-lattice potential for dynamic simulations of defective magnetic iron

Chapman

2022

Sci Rep

View full text Add to dashboard Cite

A machine-learned spin-lattice interatomic potential (MSLP) for magnetic iron is developed and applied to mesoscopic scale defects. It is achieved by augmenting a spin-lattice Hamiltonian with a neural network term trained to descriptors representing a mix of local atomic configuration and magnetic environments. It reproduces the cohesive energy of BCC and FCC phases with various magnetic states. It predicts the formation energy and complex magnetic structure of point defects in quantitative agreement with density functional theory (DFT) including the reversal and quenching of magnetic moments near the core of defects. The Curie temperature is calculated through spin-lattice dynamics showing good computational stability at high temperature. The potential is applied to study magnetic fluctuations near sizable dislocation loops. The MSLP transcends current treatments using DFT and molecular dynamics, and surpasses other spin-lattice potentials that only treat near-perfect crystal cases.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

A machine-learned spin-lattice potential for dynamic simulations of defective magnetic iron

Chapman

2022

Sci Rep

View full text Add to dashboard Cite

show abstract

“…As a result, the ML model with noncollinear magnetic degrees of freedom is significant. M. Domina et al present machine learning potentials including spin with a spectral neighbor representation for the vector field [43]. The spin is treated as a vector transformed with the bond vector and achieves SE(3) invariant local representation.…”

Section: Main Textmentioning

confidence: 99%

Time-reversal equivariant neural network potential and Hamiltonian for magnetic materials

Zhong

et al. 2022

Preprint

View full text Add to dashboard Cite

This work presents Time-reversal Equivariant Neural Network (TENN) framework. With TENN, the time-reversal symmetry is considered in the equivariant neural network (ENN), which generalizes the ENN to consider physical quantities related to time-reversal symmetry such as spin and velocity of atoms. TENN-e3, as the time-reversal-extension of E(3) equivariant neural network, is developed to keep the Time-reversal E(3) equivariant with consideration of whether to include the spin-orbit effect for both collinear and non-collinear magnetic moments situations for magnetic material. TENN-e3 can construct spin neural network potential and the Hamiltonian of magnetic material from ab-initio calculations. Time-reversal-E(3)-equivariant convolutions for interactions of spinor and geometric tensors are employed in TENN-e3. Compared to the popular ENN, TENN-e3 can describe the complex spin-lattice coupling with high accuracy and keep time-reversal symmetry which is not preserved in the existing E(3)-equivariant model. Also, the Hamiltonian of magnetic material with time-reversal symmetry can be built with TENN-e3. TENN paves a new way to spin-lattice dynamics simulations over long-time scales and electronic structure calculations of large-scale magnetic materials.

show abstract

“…The recent advancement of machine learning has had a significant impact in uncovering hidden correlations in the field of condensed matter physics [1][2][3][4][5][6][7][8][9]. This technology has also been applied to the study of magnetism, enabling for the prediction of physical quantities without the need for direct measurement or calculations, [10][11][12][13][14][15][16][17][18][19][20][21][22][23] or probing orders from the data [24][25][26][27][28][29][30].…”

Section: Introductionmentioning

confidence: 99%

Classification of magnetic order from electronic structure by using machine learning

Jang

Kim

2023

Preprint

View full text Add to dashboard Cite

Identifying the magnetic state of materials is of great interest in a wide range of applications, but direct identification is not always straightforward due to limitations in neutron scattering experiments. In this work, we present a machine-learning approach using decision-tree algorithms to identify magnetism from the spin-integrated excitation spectrum, such as the density of states. The dataset was generated by Hartree-Fock mean-field calculations of candidate antiferromagnetic orders on a Wannier Hamiltonian, extracted from first-principle calculations targeting BaOsO3. Our machine learning model was trained using various types of spectral data, including local density of states, momentum-resolved density of states at high-symmetry points, and the lowest excitation energies from the Fermi level. Although the density of states shows good performance for machine learning, the broadening method had a significant impact on the model's performance. We improved the model's performance by designing the excitation energy as a feature for machine learning, resulting in excellent classification of antiferromagnetic order, even for test samples generated by different methods from the training samples used for machine learning.

show abstract

Spectral neighbor representation for vector fields: Machine learning potentials including spin

Cited by 17 publications

References 22 publications

A machine-learned spin-lattice potential for dynamic simulations of defective magnetic iron

A machine-learned spin-lattice potential for dynamic simulations of defective magnetic iron

Time-reversal equivariant neural network potential and Hamiltonian for magnetic materials

Classification of magnetic order from electronic structure by using machine learning

Contact Info

Product

Resources

About