Lixin Sun

Timeline

1. 2015

   Dislocations in cerium dioxide

   As an MIT graduate student, she is lead author of a 2015 Nature Communications paper whose atomic-level simulations show that dislocations in cerium dioxide, a solid electrolyte for fuel cells, cause a “traffic jam” for oxide ions. Associate professor Bilge Yildiz is the paper's senior author.

2. 2017

   An MRS student award

   As a member of Bilge Yildiz's group at MIT, she wins a student award at the MRS Spring 2017 Meeting, where her labmate Qiyang Lu also receives one.

3. 2020

   FLARE force fields at Harvard

   After defending her doctoral thesis in Yildiz's group at MIT, she moves to Harvard University as a postdoctoral researcher. At Harvard's John A. Paulson School of Engineering and Applied Sciences, she coauthors FLARE (Fast Learning of Atomistic Rare Events) with Boris Kozinsky and colleagues in March 2020, an open-source method that uses Bayesian active learning to train interatomic force fields on the fly from molecular dynamics simulations.

4. 2022–2023

   NequIP and Allegro potentials

   In May 2022, she coauthors NequIP, an E(3)-equivariant neural network that learns interatomic potentials from ab-initio calculations and outperforms existing models with up to three orders of magnitude fewer training data. Allegro, a strictly local equivariant interatomic potential she coauthors in February 2023, runs a simulation of 100 million atoms.

5. December 2023

   Building MatterGen at Microsoft

   In December 2023, Microsoft Research AI4Science announces MatterGen, a generative model for broad property-guided materials design. Sun works on it with the AI4Science Materials Design team.

6. May 2024

   The MatterSim atomistic model

   In May 2024, she coauthors MatterSim, a deep learning model actively learned from large-scale first-principles computations. It simulates materials across the periodic table at temperatures from 0 to 5000 K and pressures up to 1000 GPa.

7. January 2025

   MatterGen in Nature

   Nature publishes the MatterGen paper in January 2025, with Sun among its authors at Microsoft Research AI for Science in Cambridge, UK. The model generates stable, diverse inorganic materials across the periodic table and, after fine-tuning, materials with desired chemistry, symmetry and mechanical, electronic and magnetic properties.

8. As of October 2026

   Senior researcher at Microsoft Research

   She is a senior researcher in the AI4Science team at Microsoft Research in Cambridge, UK, developing machine learning methods to accelerate materials R&D for sustainable technologies. In October 2025, she coauthors SimPoly, a machine learning force field that predicts polymer densities and glass transition temperatures from first principles. With Gregor Simm, she presents the work at Cambridge's Yusuf Hamied Department of Chemistry in March 2026.