Proposed the first neural-network-only approach to solve the chemical master equation
Characterized a type of dynamical phase transition in nonequilibrium statistical mechanics
Learned noise-induced transitions using multi-scaling reservoir computing
Developed a computational framework to quantify dynamical mutual information in intracellular signaling processes
Discovered that free energy change via Jarzynski equality is magnetic-field-independent in classical systems but amplifiable by magnetic fields in driven quantum systems
Developed a scalable path-integral-based numerical method to compute transition rates for Langevin dynamics, robust to noise intensity beyond the small-noise limit
In quantitative biology: identified a molecular circuit controlling necroptosis decisions yielding bimodal death-time distribution; modeled how chemotaxis in nutrient-replete conditions promotes bacterial population expansion