PL seminar - Aart Bik

To: plseminar@cs.indiana.edu
Subject: PL seminar - Aart Bik
From: Y. Annie Liu <liu@cs.indiana.edu>
Date: Mon, 7 Oct 1996 16:55:07 -0500 (EST)
Cc: ajcbik@cs.indiana.edu, cs.students.grad@moose, cs.colloq@moose

		    Programming Languages Seminar
 		        Thursday, Oct.10, 1996	<--- DATE change, this wk only!
  		         10am-11am, LH 101

	  Compiler Support for Sparse Matrix Computations
	  -------- ------- --- ------ ------ ------------

				Aart Bik
			Computer Science Department
			   Indiana University

				Abstract
				--------

As restructuring compilers are useful in automatically detecting and 
exploiting implicit parallelism in a serial program, the question arises 
whether it is also possible to let a restructuring compiler convert code that 
operates on simple data structures into a format that exploits certain 
characteristics of the data operated on. In contrast to conventional 
restructuring compilers, mainly focusing on program transformations, this 
approach must allow for the application of data structure transformations
as well.

For applications involving sparse matrices, this approach implies that
all computations on these matrices may simply be defined on two-dimensional 
arrays. A special kind of restructuring compiler, which we will refer to as a 
sparse compiler, transforms these simple data structures into more
complex sparse data structures, thereby reducing storage requirements and 
computational time.  

This talk will present the principles of behind the sparse compiler and discuss
the results obtained with it.

				Biography
				---------
Aart J.C. Bik was born in 1969 in Gouda, The Netherlands. He received the M.S.
degree (cum laude) in computer science from the Utrecht University, The 
Netherlands, in 1992. In 1996, he received his Ph.D. degree from the Leiden 
University, The Netherlands. His recent research interests include
restructuring compilers and sparse computations. Currently, he is working on 
restructuring compiler techniques for Java programs.

--
Aart's Ph.D. dissertation won the best thesis award of this year in the
Netherlands!