ICTP-EAIFR Colloquium on "Machine learning and molecular dynamics"
Speaker: Michele Parrinello (ETH-Z, USI Lugano, IIT Genoa)
Atom based computer simulation is one of the most important tools of contemporary physical chemistry. In spite of its many successes, it suffers from severe limitations. Here we show how machine-learning techniques can help in solving at least two different problems. The first one is the accuracy of current interatomic potential models; the second is the limited time scale that simulations can explore. In order to solve the first problem we train a neural network on a set of accurate but expensive quantum chemical calculations. In this way, it is possible to obtain an accurate description of the system at a relatively low computational cost. Crucial for the success of this program has been the design of the neural work and the selection of the training set. We apply this approach to study a metal non-metal transition and to chemical reactions in condensed phases. These applications would not have been possible without the use of efficient sampling methods capable of lifting the time scale barrier. To this effect, we have developed two very efficient sampling methods, metadynamics and variationally enhanced sampling. Both methods are based on the identification of appropriate collective variables, or slow modes, whose sampling needs to be accelerated. Machine learning can be used also for the construction of efficient collective variables based on a modification of the well-known linear discriminant analysis classification method. Finally, we use the variational enhanced sampling approach and a deep neural network to further increase our sampling ability.
Видео ICTP-EAIFR Colloquium on "Machine learning and molecular dynamics" канала Int'l Centre for Theoretical Physics
Atom based computer simulation is one of the most important tools of contemporary physical chemistry. In spite of its many successes, it suffers from severe limitations. Here we show how machine-learning techniques can help in solving at least two different problems. The first one is the accuracy of current interatomic potential models; the second is the limited time scale that simulations can explore. In order to solve the first problem we train a neural network on a set of accurate but expensive quantum chemical calculations. In this way, it is possible to obtain an accurate description of the system at a relatively low computational cost. Crucial for the success of this program has been the design of the neural work and the selection of the training set. We apply this approach to study a metal non-metal transition and to chemical reactions in condensed phases. These applications would not have been possible without the use of efficient sampling methods capable of lifting the time scale barrier. To this effect, we have developed two very efficient sampling methods, metadynamics and variationally enhanced sampling. Both methods are based on the identification of appropriate collective variables, or slow modes, whose sampling needs to be accelerated. Machine learning can be used also for the construction of efficient collective variables based on a modification of the well-known linear discriminant analysis classification method. Finally, we use the variational enhanced sampling approach and a deep neural network to further increase our sampling ability.
Видео ICTP-EAIFR Colloquium on "Machine learning and molecular dynamics" канала Int'l Centre for Theoretical Physics
Показать
Комментарии отсутствуют
Информация о видео
22 октября 2020 г. 21:15:49
01:20:42
Другие видео канала
Machine learning in drug discovery: what can go wrong? — Vikram SundarGems of Ramanujan and their Lasting Impact on MathematicsTime Crystals: Past, Present, and FutureApplications of Machine Learning in the Supply ChainJohn Kitchin: Using Machine Learning to Improve Molecular SimulationsEnhanced Sampling Methods - chapter 1: Free Energy and SamplingDecoherence and the Quantum Theory of the Classical – ICTP ColloquiumMachine Learning Basics | What Is Machine Learning? | Introduction To Machine Learning | SimplilearnInteratomic potentials from first principlesMolecular Dynamics in 5 MinutesGeorg Kresse: Finite temperature materials modelling, is machine learning the way to go?Molecular Dynamics Simulations - Introduction to BeginnersML4Science Seminar: Alexandre Tkatchenko (University of Luxembourg)Machine Learning Fundamentals: Cross ValidationDay 4: Theoretical Physics Session, Edward WittenDeep learning on graphs: successes, challenges, and next steps | Graph Neural NetworksConvenient and efficient development of Machine Learning Interatomic PotentialsJuliaCon 2020 | Doing Scientific Machine Learning (SciML) With JuliaDr. Satchin Panda on Time-Restricted Feeding and Its Effects on Obesity, Muscle Mass & Heart Health