High Performance Computing
Theoretical chemistry and modelling
|A combined theory/experiment approach|
The evolution of research tools in the field of modeling and representation of complex chemical systems, as well as in the field of structural analysis, characterization and data processing, requires the acquisition and pooling of heavy computing resources and access to an infrastructure of efficient network services, meeting the intrinsic constraints related to the types of calculations carried out in chemistry.
Theoretical chemistry and modeling are indispensable in the study of the properties of matter, at the molecular or solid scale. The theoretical study contributes to the structural determination of chemical systems and provides a better understanding of the electronic properties that determine the physical and chemical properties measured experimentally. It is also particularly powerful in the study of chemical transformations and is the indispensable tool for the representation of chemical systems.
|Large and diversified computing resources for maximum reliability.|
Theoretical chemistry is still far from having achieved its ultimate goal of relating the electronic properties of groups of atoms to macroscopic quantities that can be measured. This need is particularly important for nanosystems whose properties vary with size. The modelling of these poses real challenges and significant computational resources are necessary. In addition, these means of calculation must be diversified because the methodological approaches are themselves very varied.
In other words, the material requirements for conducting current studies vary from subject to subject. Thus, some calculations depend on the speed of access to the storage space, others require gigantic amounts of RAM, still others need communication between nodes with very low latency. Finding hardware that simultaneously meets the needs of all types of computing is not only difficult but above all unprofitable, because some features will only be used temporarily. It is recognized by the community of theorists around the world that the diversification of computational means is essential and intrinsic to the discipline.
These means of high-performance computing are useful in the fields of:
- New energy sources: hydrogen generation, methanol.
- Storage: lithium-ion batteries, supercapacitors
- Conversion materials and devices.
- Fuel cells, photovoltaic cells, light-emitting diodes.
- Active ingredients and vectorization: new molecules, nano-encapsulation.
- Diagnostic assistance: medical imaging, biosensors, biochips.
- Materials for bio-applications: biomimetic materials, bioactive surfaces.
- Detection and treatment of pollution: chemical sensors, selective adsorbents.
- Valorization of agro/biological resources: agro-fuels.
- Green chemistry: syntheses with atom savings, homogeneous and heterogeneous catalysis.
Members of the Information Systems Department of the Charles Gerhardt Montpellier Institute:
- Fabrice Boyrie
- Cyril Bourgogne
- Mourad Guermache
- Pierre Sans
In order to be able to deal with new problems and open up computing resources to new users, the ICGM has increased its computing power from year to year, allowing access to a total resource of more than 5000 cores.
Some of the ICGM's high-performance computing capabilities
- 174 nodes based on bi-XEON X5560, (or equivalent), 8 cores, 24GB RAM/node.
- 102 nodes based on bi Xeon E52670v1 (or equivalent), 16 cores, 64GB RAM/node
- 28 nodes based on bi Xeon E52697v3 (or equivalent), 28 cores, 128GB RAM/node
Most of these machines are interconnected by low-latency "infiniband" networks.
Some software used
- Gaussian 2003 and 2009
- Vasp v 5.4 (quantum solid-state chemistry)
- Wien 2k (quantum solid-state chemistry)
- Materials Studio (quantum solid-state chemistry)
- Demon, Gamess, Crystal