Drastically transforming parallel program development through “POP”

Professor Masato Takeichi
(Department of Mathematical Informatics)

Professor Takeichi works on the creative research of writing parallel programs from the perspective of “POP = Parallelism-Oblivious Programming.” It aims to establish “next generation parallel programming methods” that support any parallel computer to emerge in the future.

Standard methods for developing parallel programs use the library called MPI (Message Passing Interface). Since programmers specify where to use parallelism and it is dependent on their skills, fundamental problems will remain unsolved and even the negative legacy may increase as long as this kind of methods are used to develop parallel programs. The goal of POP is to establish a programming methodology that is based on a programming language targeted at parallel computers and particularly oblivious of computer parallelism.

Professor Takeichi aspires to realize algorithms expressed in the functional language Haskell using Fortress developed by Sun Microsystems in US. Fortress was designed as a successor language to Fortran used in scientific computing, and has embedded parallelism. He applied skeleton parallel programming developed by his laboratory to it and is currently developing program styles specific to each problem area.

Scientific and technological development is driven by computer science, which systematically handles algorithms, and computing science, which applies computational techniques to the phenomena to be solved. However, advanced achievements of parallel programming in computer science are not fully transferred to program development in computing science. To promote scientific and technological development in the 21st century by removing such barrier with POP ---- this is the grand challenge of Professor Takeichi as a computer scientist.