Researchers in race for computer speed 


Some of the sharpest minds in sophisticated computer modelling will gather at Charles Darwin University this week to consider questions with potentially dramatic implications globally.

The 19 experts from nine countries research in a field known as parameterized complexity, which involves the pre-programming, mathematical foundation that builds tools for modelling programs, designing algorithms for solving them, and analysing their robustness and speed.

They will attend a workshop to consider how computer speeds can continue to be increased using better algorithms rather than relying on continual improvement in computer hardware.

Algorithms are applied in a range of environments where massive amounts of data are involved, for example how Facebook handles 40 billion photos, Wal-Mart deals with more than a million customer transactions each hour and Google provides 10 top responses almost immediately.

The workshop is being organised by Mike Fellows, an Australian Professorial Fellow at CDU and one of the founders of parameterized complexity in the mid-1990s.

Professor Fellows said it was doubtful that computer hardware could continue to deliver great improvements in speed.

“Wires can only get so close before molecules begin to jump from one to another, and speeds can only get so great before the user is exposed to x-rays,” he said.

“In the race for speed, the parameterized complexity researchers are betting on better algorithms over computer hardware.”

The mathematicians will consider a fundamental change in how algorithms are done and where this change might take their application.

The gathering will include Professor Mike Langston, who conducts his research on one of the largest super computers in the world at Oak Ridge National Laboratories in the United States.

Professor Fellows is a Humboldt Research Award recipient, a prestigious international award made to a maximum of 100 researchers globally each year for the impact of their discoveries.