This paper concerns a comprehensive description of a very popular software package called GROMACS (GROningen MAchine for Chemical Simulations), a molecular dynamics simulation package originally developed at the University of Groningen. It is highly cited because so many scientists use the GROMACS program.

There are a few reasons for the popularity of this software. First, we have put it in the public domain under GPL (GNU Public license) for anyone to download and use. There is only one restriction: a user can only incorporate (part of) the software in his program if (s)he also makes that program publicly available under the same conditions. Second, it is much faster than other programs because one of the authors (Erik Lindahl) has written assembly code for the most time-critical parts, exploiting the full capability of modern Intel processors with graphics pipelines. Furthermore, it can be applied to clusters of many processors that are nowadays commonplace. Third, it is scientifically up-to-date and comes with a large variety of options and analysis tools.

Molecular dynamics simulations mimic the motion of atoms and molecules. It can handle up to millions of particles during time spans up to microseconds. It produces "macroscopic" results based on the microscopic interactions between particles. Thus, we can predict and understand the properties of materials on a nanoscale, especially for biological processes involving proteins, nucleic acids, and membranes. We all hope that we can fold proteins into their native shape by simulation.

The roots of the research that led to this article lie in a scientific project we were engaged in during the early 1990s. The project concerned a computing machine consisting of 24 processors coupled in a ring, tailor-made for molecular simulations. Parallel computing was just coming about at the time. The project was quite successful and the software was completely rewritten for the parallel machine (hence the name GROningen MAchine for Chemical Simulation). But the software could also be used on processor clusters and was continuously developed since that project. Because it in the public domain, many people have contributed to its development. The developers are scattered among several institutions in Europe: including the Universities of Groningen, Uppsala, Stockholm, and the Max Planck Institute for Polymer Research in Mainz.

Scientifically, simulations are essential in order to understand biological processes and processes in nanoscale devices. Understanding means the possibility to innovate and predict new properties. But also, the fact that we have placed the software in the public domain has political implications. We believe that products of publicly funded research should be in the public domain. This not only applies to software, but also to scientific data. No publisher should be allowed to protect public access to scientific data, such as measured properties of materials, structures of molecules, and genetic nucleotide sequences. It is only in this way that international scientific endeavors can truly flourish.

Related Links:           [return]