• Project description
• Cluster configuration
• Earthquakes

• Zacharie Duputel, IPGS - CNRS - Université de Strasbourg
• Mark Simons, Seismological Laboratory, California Institute of Technology
• Rodrigo Ibata, Observatoire Astronomique - CNRS - Université de Strasbourg
• James L. Beck, EAS Division, California Institute of Technology
• Luis Rivera, IPGS - CNRS - Université de Strasbourg
• Alessia Maggi, IPGS - CNRS - Université de Strasbourg
• Romaric David, Pôle HPC - Université de Strasbourg
• Fernando Niño, LEGOS - CNRS - Université Paul Sabatier

Despite the ever-increasing amounts observations available in the age of big data, observations are often limited to smaller spatial extents or timescales than the studied processes. In this context, the answers to many problems lie in models that cannot be uniquely constrained given available data. Such an inference process leaves unanswered basic questions about any model: How unique is a given model? How much does each observation contribute to the model? What is the impact of errors in the model? How reliable are predictions made from the model?

To address these questions, this project focuses on applying full Bayesian analysis techniques to large ill-posed inverse problems in solid earth geophysics including but not limited to earthquake source inversions and seismic imaging techniques. This project also investigates other fields such as volcanology, glaciology and astrophysics. Bayesian analysis is a powerful tool to combine theoretical knowledge with measurements in order to address scientific problems probabilistically. This project is particularly challenging given the high-dimensionality and ill-posed nature of investigated inversion problems. Another significant challenge is to develop appropriate stochastic models that will reflect uncertainties in our theoretical predictions, which are currently neglected

• publicgpu or grantgpu queue:
  • 2 nodes with 4 K20 GPU cards
  • 1 nodes with 2 K20 and one K40 GPU card
• public queue:
  • M2050, M2070 and K20 GPUs.

• Must be compatible with existing NEC nodes of the HPC cluster
• The GPUs must be able to host G and M matrices larger than 5GB.
• 6 compute nodes
  • 4 GPU NVIDIA K40 per node (total: 24 GPUs)
  • Each GPU will have its own dedicated PCI-E slot
• 4 GPU cards to be installed on existing nodes (2 GPU/node)
  • NVIDIA K40
  • Each GPU card will have its own dedicated PCI-E slot
• We also request 3 years of software and hardware support
• Budget: 147 000 euros VAT included
• Currently working with Transtech, Nec and Dell for pre-pricings :
  • Nec/Transtech working on the same nodes as the ones installed
  • Dell working on a solution based on a C8000 chassis and C8220x nodes. If the C8000 chassis is full, could be the same density as the NEC nodes. Otherwise may be less interresting

• Initiative d’Excellence (IdEx) 2014: projet attractivité
  • 150 000 euros
  • PI: Z. Duputel (Strasbourg)
  • project: Bringing a Bayesian Perspective to Large Dimensional Problems in Geophysics and Astrophysics
• CNRS INSU project: Terre Solide (ALEAS) 2014
  • 6 000 euros
  • PI: Z. Duputel (Strasbourg)
  • Project: Vers une nouvelle génération de modèles de source via une prise en compte réaliste des incertitudes