Department of Materials Science & Metallurgy: Research papers of the month

Department of Materials Science & Metallurgy

Research papers of the month

May 2016 - Reproducibility in density functional theory calculations of solids


A large scale community research collaboration, including Chris Pickard [1], has shown that widely used computer codes for first principles materials property prediction agree with each other.

These computational tools all start from the same fundamental equations of quantum mechanics, however choices are made in the design of efficient computer algorithms. A typical package comprises hundreds of thousands of lines of source code, and are written by large teams of researchers over decades who will have made distinct design decisions.

In the light of an increased scrutiny of computer codes following reports of a "reproducibility crisis" in science, the current study is reassuring.

[1] Chris Pickard is a lead developer of the CASTEP code which is licensed by Cambridge Enterprise to Dassault Systemes BIOVIA Ltd.

 

Figure: A major international study has shown that modern codes for the first principles prediction of materials properties agree with each other.

K. Lejaeghere et al., "Reproducibility in density functional theory calculations of solids ", Science 351, 1415 (2016)

DOI: 10.1126/science.aad3000