Témakiírások
Machine Learning Interaction Potentials for simulations of condensed-phase systems
témakiírás címe
Machine Learning Interaction Potentials for simulations of condensed-phase systems
intézmény
doktori iskola
témakiíró
tudományág
témakiírás leírása
"Next-generation machine learning interaction potentials (MLIPs) to model aqueous-phase chemistry with quantum chemical accuracy. The project aims to integrate physically grounded representations with data-driven models trained on highly accurate local coupled-cluster (CC) reference calculations, ensuring transferability and chemical reliability.
Utilization of path-integral molecular dynamics (PIMD) to capture nuclear quantum effects (NQEs) and anharmonicity in hydrogen-bonded networks, and to describe isotope effects and tunneling.
Exploration of algorithmic strategies to couple MLP training pipelines with systematic CC benchmarks, including data selection, active learning, and handling long-range electrostatics. Acceleration of simulations via multiple time step and ring-polymer contraction-based methods.
Target applications include:
Detailed mechanistic understanding of chemical reactions in solution.
Accurate prediction of structural, thermodynamic, and dynamic properties of aqueous systems.
Extension of MLPs to describe complex species in solution with chemical accuracy.
This project bridges quantum chemistry, path-integral molecular dynamics simulations, and machine learning."
Utilization of path-integral molecular dynamics (PIMD) to capture nuclear quantum effects (NQEs) and anharmonicity in hydrogen-bonded networks, and to describe isotope effects and tunneling.
Exploration of algorithmic strategies to couple MLP training pipelines with systematic CC benchmarks, including data selection, active learning, and handling long-range electrostatics. Acceleration of simulations via multiple time step and ring-polymer contraction-based methods.
Target applications include:
Detailed mechanistic understanding of chemical reactions in solution.
Accurate prediction of structural, thermodynamic, and dynamic properties of aqueous systems.
Extension of MLPs to describe complex species in solution with chemical accuracy.
This project bridges quantum chemistry, path-integral molecular dynamics simulations, and machine learning."
felvehető hallgatók száma
2 fő
helyszín
ELTE Department of Organic Chemistry
jelentkezési határidő
2026-05-31