2012 INCITE Awards
Title: “Towards the Phase Diagram of Water and Ice with Quantum Monte Carlo”
Principal Investigator: Dario Alfè, University College London
Co-Investigators: Mike Towler, University of Cambridge
Wissam Al-Saidi, University of Pittsburgh
Michael Gillan, University College London
Kenneth Jordan, University of Pittsburgh
Michaelides Angelos, University College London
Scientific Discipline: Materials Science: Condensed Matter and Materials
INCITE Allocation: 23,000,000 processor hours
Site: Oak Ridge National Laboratory
Machine (Allocation): Cray XK6 (23,000,000 processor hours)
Water is one of the most important substances on Earth, being crucial in fields ranging from biology to the Earth sciences, environmental sciences and medicine. The interaction of water molecules with various surfaces is an outstandingly important issue, as almost any surface in contact with the atmosphere is covered with at least a thin film of water molecules. Because of its importance, water is also one of the most studied substances, with early attempts to simulate its properties dating back nearly 80 years.
However, it is fair to say that the state of the art in computer simulation is still not sufficiently accurate. In particular, even seemingly simple properties like the diffusion coefficient, or the room pressure/temperature density, are difficult to calculate even with modern techniques based on density functional theory (DFT) with sophisticated exchange-correlation (XC) functionals including exact exchange and/or empirical van der Waals interactions. Our INCITE 2010 and 2011 projects have shown how quantum Monte Carlo (QMC) techniques can efficiently exploit petascale resources, and provide the much needed accuracy for water’s interaction with various surfaces.
The crystalline phases of ice provide an excellent platform to assess and quantify the
importance of weak van der Waals and hydrogen bond forces by presenting a large variety of distinct fixed geometries, with experimental data available for a number of properties including structural data and the cohesive energies of the various phases. Comparing cohesive energies and structural data obtained with QMC will provide a stringent test for the accuracy of the method, and early results obtained during this project’s 2010 and 2011 projects are very encouraging, showing that QMC is able to achieve much better than chemical accuracy. The techniques to be applied are completely general, and the project’s work will raise awareness of what it is possible to do with QMC methods on machines like Jaguar, which will become increasingly available in the future.