The applications are closed.
New postdoc positions are available for 5-year JST/CREST project "Modularity for Supercomputing". (see the project description below for more details), conducted jointly at the University of Tokyo, Kyushu University, and Kyoto University.
The appointment can start as early as in April 2012 (the starting date is negotiable). The contract of appointment will be renewed for each academic year, and can be extended up to March 2017, subject to performance. Salary will be about 4,000,000 upto 5,000,000 JPY (Japanese yen) per year.
Applicants should have a PhD in computer science, software engineering, or related fields, and have strong background in at least one (preferably two or more) of the following topics: programming languages, software engineering, high-performance computing, and theoretical computer science.
Interested candidates are invited to send a detailed CV via email to Shigeru Chiba (see below for his contact info.), no later than December 15, 2011. As mentioned above, the whole project consists of four research groups distributed over Japan; we will decide which group he or she will belong to, according the applicant's expertise and preference.
Software development for supercomputing (SC) is extremely hard, mainly because application programmers require deep knowledge about the architecture, operating system, and middleware of the supercomputer, not to mention the application domain. The goal of our project is to address the difficulties by applying language-based technologies---more specifically, technology based on product line architectures---to software development for supercomputing.
Towards our goal, we study programming languages and software development environments in which each individual developer or researcher can easily build their own product lines for their favorite supercomputing applications.
Design and implementation of a new programming language that provides functionality for advanced modularity and frameworks in high-performance computing. In particular, with minimal execution overheads and static typing.
Design and implementation of domain-specific languages (DSLs) for prototyping highly-parallel scientific applications and for experimenting optimization techniques. Particular goals are: (1) DSLs embedded into dynamic programming languages such as Ruby and executable highly-parallel architectures such as GPGPU, and (2) design and implementation of the DSLs' framework to support modular optimization techniques.
Domain-specific language (DSL) construction methods based on product-line architecture. We plan to develop a DSL for open source repository mining, one of the important research fields requiring high performance computing.
Type systems for efficiently checking safety of highly modularized software components, supported by the language devoloped by Chiba's group. Customizable type systems for domain-specific languages.
Tokyo Institute of Technology
Back to the top