Two new Associate Teams and one renewed Team with California Universities


© Inria / Photo Kaksonen

Inria is glad to announce the selection of 2 new Associate Teams with California universities (University of California San Diego and University of California Santa Barbara) as part of the 2017 Inria Associated Team call. The two teams are created in 2017 for 3 years.

In addition, one team that was created in 2014 (SNOWBALL formerly SNOWFLAKE) with Stanford University has been renewed for 3 years.

The Inria@SiliconValley program counts 14 ongoing Associated Teams in 2017. See the Research Teams page for the full list.

2 new Associate Teams selected in 2017:

  • Num4SEP – “Numerics for Spherical Electroporation” – MONC Inria Team (Clair Poignard) and UC Santa Barbara (Frederic Gibou)

Electroporation-based therapies (EPTs) consist in applying high voltage short pulses to cells in order to create defects in the plasma membrane. They provide interesting alternatives to standard ablative techniques, for instance for deep seated badly located tumors. However their use is still limited due to a lack of knowledge of tissue electroporation. The goal of the associate team is to focus on the multiscale numerical modeling of spheroid electroporation, in order to provide new insights in electroporation at the mesoscopic scales (spheroids provide interesting tumor-like biological models). Benefiting from the expertise of F. Gibou’s team in HPC for multiphysics, and the expertise of the team MONC in tumor growth and cell electroporation modeling, the goal of the associate team Num4SEP is to obtain accurate and efficient numerical tools for the quantitative evaluation of the EPTs at the mesoscopic scale.
  • COMPOSITE – “Compositional System Integration” – TEA Inria Team (Jean-pierre Talpin) and UC San Diego (Rajesh Gupta)

Most applications that run somewhere on the internet are not optimized to do so. They execute on general purpose operating systems or on containers (virtual machines) that are built with the most conservative assumptions about their environment. While an application is specific, a large part of the system it runs on is unused, which is both a cost (to store and execute) and a security risk (many entry points). A unikernel, on the contrary, is a system program object that only contains the necessary operating system services it needs for execution. A unikernel is built from the composition of a program, developed using high-level programming language, with modules of a library operating system (libOS), to execute directly on an hypervisor. A unikernel can boot in milliseconds to serve a request and shut down, demanding minimal energy and resources, offering stealthiest exposure time and surface to attacks, making them the ideal platforms to deploy on sensor networks, networks of embedded devices, smart grids and clouds. The goal of COMPOSITE is to develop the mathematical foundations for sound and efficient composition in system programming: analysis, verification and optimization technique for modular and compositional hardware-system-software integration of unikernels. We intend to further this development with the prospect of an end-to-end co-design methodology to synthesize lean and stealth networked embedded devices.