Skip to main content

From graphics for computer games to simulations for science


Synchrotron light can tell you the shape of the electrons on the outside of a molecule – but what lies inside the wrapping is harder to find out. Now, a program called BigDFT allows scientists to calculate what complicated structures would look like, in record time. The “father” – both main contributor and “babysitter” - of BigDFT is Luigi Genovese, from the Theory Group, and he has just received the French national prize Bull-Joseph Fourier for it.

  • Share

The prize is awarded for writing a physical simulation which uses the power of parallel processing – where the programme handles many tasks at once and then knits them back together, rather than taking things strictly linearly. Luigi Genovese has used this technique to optimize a programme capable of assembling the electron density pattern for groups of thousands of atoms.

The software, called Big DFT – or Big Density Functional Theory – makes its calculations based on pure quantum theory, with no adjustable parameters (ab-initio calculation). Using parallel processing techniques, the software can quickly predict the electronic structure properties of a given system. Like all the ab-initio approaches, since this data has been generated from first principles, it is eminently suitable for comparing to experimental data. This way, both the effectiveness of theory and of the experimental measurements can be tested. The level of detail which that BigDFT calculations can reach makes them relevant to real systems. For example, both nano-electronics and many small bio-molecules are made of a few thousands of atoms, so their properties can be calculated with BigDFT.

Luigi Genovese has been working on this project since 2006 as a post-doctoral researcher at the Commissariat à l'Energie Atomique (CEA). When he started at the ESRF in September 2008, he set about porting BigFTP to make use of the new capabilities of an unusual type of processor – a graphical processing unit (GPU), traditionally used for computer game graphics. However, nowadays GPUs are being repurposed for scientific simulation thanks to their accuracy and the speed of  their parallel processing ability. Now that BigDFT is updated to work on the newest generation of GPUs, it can handle even greater numbers of electrons – even for tricky materials like metals, where each electron interacts with every other. Thanks to the work of programmers like Luigi Genovese, the capabilities of density functional theory continue to increase. This helps researchers to design and test new materials, molecules and medicines, turning their knowledge of the world into improvements in people’s lives.


Top image: Luigi Genovese received his prize on 30 June during the Ter@tec'09 Forum, an event on in the field of high performance simulation and computing. Credits: CEA/Ter@atec.