Projects
Ab initio water using machine learning models
The advent of machine learning models has made possible to study water with unprecedented accuracy in the description of the interatomic interactions. In this series of papers we study different aspects of water science using one such model, ranging from the intriguing hypothesis of a liquid-liquid separation at deep supercoolings, to the homogeneous nucleation of ice.
Signatures of a liquid–liquid transition in an ab initio deep neural network model for water, TE Gartner, L Zhang, PM Piaggi, R Car, AZ Panagiotopoulos, PG Debenedetti, Proc. Natl. Acad. Sci. 117, 26040-26046 (2020)
Phase equilibrium of water with hexagonal and cubic ice using the SCAN functional, PM Piaggi, AZ Panagiotopoulos, PG Debenedetti, R Car, J. Chem. Theory Comput. (2021)
Homogeneous ice nucleation in an ab initio machine-learning model of water, PM. Piaggi, J Weis, AZ Panagiotopoulos, PG Debenedetti, and R Car, Proc. Natl. Acad. Sci. 119, 33 (2022)
Entropy as a tool for crystal discovery
The prediction of crystal structures using the computer is a challenging problem that has received considerable attention in the last two decades. Despite the great progress in this area, most available methods search for structures neglecting the effects of temperature. In this project we propose a computational method to search for polymorphs in free energy surface therefore automatically including the effects of temperature. This allows to find structures stabilized by entropic effects.
This research is described in these papers:
Enhancing entropy and enthalpy fluctuations to drive crystallization in atomistic simulations, Pablo Piaggi, Omar Valsson, and Michele Parrinello, Phys. Rev. Lett. 119 , 015701 (2017)
Predicting polymorphism in molecular crystals using orientational entropy, Pablo Piaggi and Michele Parrinello, Proc. Natl. Acad. Sci. USA 115 (41), 10251-10256 (2018)
Multithermal-multibaric and generalized ensembles simulations
Can you imagine performing a single experiment in the lab and obtaining information at all temperatures and pressures? Well, this might not be possible in the lab, but the tools of Statistical Mechanics allow us to do it in the computer. Much like in The Matrix, in the computer we can bend the rules of nature and use this to our advantage. In this way, in a single computer simulation we can obtain properties of a substance in a big chunk of the temperature pressure phase diagram.
This research is described in these papers:
Multithermal-multibaric simulations from a variational principle, Pablo Piaggi and Michele Parrinello, Phys. Rev. Lett. 122, 050601 (2019)
Calculation of phase diagrams in the multithermal-multibaric ensemble, Pablo Piaggi and Michele Parrinello, J. Chem. Phys. 150, 244119 (2019)
Unified Approach to Enhanced Sampling, Michele Invernizzi, Pablo Piaggi, and Michele Parrinello, Phys. Rev. X 10, 041034 (2020)
And some very nice applications can be found here:
Signatures of a liquid–liquid transition in an ab initio deep neural network model for water, Thomas E Gartner et al, PNAS 117 (2020)
Ab initio phase diagram and nucleation of gallium, Haiyang Niu, Luigi Bonati, Pablo M. Piaggi, and Michele Parrinello, Nature Communications 11 (2020)
Development of order parameters and fingerprints
Order parameters are an essential ingredient of free energy calculations. In the past five years I have worked in the development of several orders parameters and fingerprints. Here are some of them:
Enhancing entropy and enthalpy fluctuations to drive crystallization in atomistic simulations, Pablo Piaggi, Omar Valsson, and Michele Parrinello, Phys. Rev. Lett. 119 , 015701 (2017)
Predicting polymorphism in molecular crystals using orientational entropy, Pablo Piaggi and Michele Parrinello, Proc. Natl. Acad. Sci. USA 115 (41), 10251-10256 (2018)
Calculation of phase diagrams in the multithermal-multibaric ensemble, Pablo Piaggi and Michele Parrinello, J. Chem. Phys. 150, 244119 (2019)
Molecular dynamics simulations of liquid silica crystallization, Haiyang Niu, Pablo M. Piaggi, Michele Invernizzi, and Michele Parrinello, PNAS 115 ( 2018)