The school targets both students and researchers interested in exploring the potential interactions between numerical simulation and compiler optimizations. This is a too vast area, in terms of types of applications, specifications, optimization mechanisms, and architectures, to cover it in its generality and be able to identify common topics of interest between specialists in numerical simulation from the maths side, and compiler/architecture researchers on the computer science side. The school will rather focus on applications whose regularity, use of data structures such as arrays, and arithmetical intensity may make the automation or semi-automation of optimizations both feasible and useful, concentrating in particular on polyhedral analysis and optimizations for GPGPUs and multicores. The goal is to discuss, in the light of the expertise from the application side, how recent developments in polyhedral techniques and tools can help (or not yet) improving the interaction between the users and the compilers.
The school being inter-disciplinary, the prerequisites for an attendee are therefore a good knowledge on the research problems on his/her specific domain and an open mind towards numerical simulation applications (for the compiler attendees) and compilers and architectures (for the numerical analysis attendees). From the computer science side, the school can be viewed as a follow-up to the 2013 spring school on polyhedral code analysis and optimizations (see all the material, slides and videos on http://labexcompilation.ens-lyon.fr/polyhedral-school) although attendance to this previous school is neither mandatory, nor redundant (different topics will be covered). Some knowledge on parallel programming languages and associated runtimes or libraries should help to have a higher-level view on the problems addressed in the school. The keynotes organized in 2013 on parallel languages for HPC (see all slides available on http://labexcompilation.ens-lyon.fr/hpc-languages) can be a helpful complement in this respect, as they will barely be addressed in the school.
