INSPIRED: Inelastic neutron scattering prediction for instantaneous results and experimental design

Abstract

Inelastic neutron scattering (INS) has unique advantages in probing how atoms vibrate and how the vibrations propagate and interact. Such dynamic information is crucial in understanding various material properties, from heat capacity, thermal conductivity, phase transitions, and chemical reactions to more exotic quantum behavior. The analysis and interpretation of the INS spectra often start from a model structure of the sample, followed by a series of calculations to obtain the simulated spectra to compare with experiments. The conventional way to perform such calculations usually requires significant time, computing resources, and specialized expertise. Here, we present a new program named INSPIRED (Inelastic Neutron Scattering Prediction for Instantaneous Results and Experimental Design), which enables users to perform rapid INS simulations in several different ways on their personal computers in just a few clicks, with the crystal structure as the only input file. Specifically, the users can choose a pre-trained symmetry-aware neural network (coupled with an autoencoder) to predict the phonon density of states (DOS), 1D S(E) and 2D S($\left|\text{Q}\right|$,E) spectra for any given structure. One can also choose an existing density functional theory (DFT) calculation from a database (containing over 12,000 crystals), and quickly obtain the simulated INS spectra for single crystals and powders. It is also possible to use pre-trained universal machine learning force fields to relax a given crystal structure, calculate the phonon dispersion and DOS, and, subsequently, the INS spectra. All these functions are implemented with a PyQt graphic user interface. We expect these new tools will benefit broad user communities and significantly improve the efficiency of experiment design, execution, and data analysis for INS.

Program summary

Program Title: INSPIRED

CPC Library link to program files: https://doi.org/10.17632/8g3s8f9n2p.1

Developer's repository link: https://github.com/cyqjh/inspired (software), https://doi.org/10.5281/zenodo.11478889 (database, models files, and virtual machine appliance file)

Licensing provisions: MIT

Programming language: Python

Nature of problem: How to easily and quickly assess the expected INS spectra for a given crystal structure has been a major challenge in the INS user community. It is a main bottleneck affecting almost every stage of the workflow, from experimental design and steering to data analysis and interpretation. The widely used approach involving DFT calculations is time-consuming, requires advanced computing resources, and has a steep learning curve. With the growing power of neutron sources and more high-throughput INS experiments, there is a pressing need to address this problem, preferably by taking advantage of the recent developments in machine learning and artificial intelligence.

Solution method: We take a data-driven approach to tackle the problem. A symmetry-aware neural network is trained to make direct predictions from the crystal structure to either 1D spectra or latent space vectors, which can be decoded to reconstruct 2D spectra. The database used for the training contains over ten thousand crystals, which can also be used to calculate INS spectra for single crystals and powders. The recently emerging universal machine learning force fields provide another venue to accelerate the simulation significantly. All these solutions are implemented in a graphic user interface so that a user with no modeling/programming background or access to powerful computers can still easily run the workflow.