Niall McMahon

Home  |  Blog  |  Contact  |  Powered by DuckDuckGo

There is Theory, There is Experiment and There is Simulation


Niall McMahon

Taking up the challenge to build an argument that simulation will someday replace almost all animal experiments, I thought of Sandia National Laboratories and a news report in February this year about the launch of Red Storm, an IBM Blue Gene supercomputer and the world's fastest machine. Sandia is a laboratory of the National Nuclear Security Administration (NNSA).

Sandia has long used computer simulation as a research tool. Faced with budget cuts and regulatory constraints in the 90s, however, Sandia accelerated the development of its computer simulation capability to replace costly physical experimental programmes. Carl Peterson, speaking at last year's Third MIT Conference on Fluid and Solid Mechanics, gave a vivid overview of Sandia's transition from having relatively little experience in advanced computer simulation to Red Storm, the cutting edge of simulation technology. It was Sandia who first proposed, based on computer models, that a small piece of foam might have caused the Columbia tragedy in 2003 [PDF 1.0MB]. Red Storm's simulations will probe nuclear processes that could never be performed physically.

Sandia's rapid development of simulation technology over the past 20 years has been driven by: (1) the pursuit of understanding, (2) the reduction of costs and (3) regulatory and accepted moral constraints. Sandia's simulation capability in Red Storm was the stuff of experimentalists' dreams not 20 years ago. Replace nuclear technology with systems biology and you have a striking parallel.

At the launch of Red Storm, the NNSA Administrator Linton Brooks commented [PDF 1.8MB], "When I grew up, there were two ways to think about science: theory and experiment. Some of my colleagues in the scientific community now say that scientists in the future will grow up thinking there is theory, there is experiment, and there is simulation - three ways in which we advance scientific knowledge." I like this summary.

The idea that simulation will someday replace almost all animal experiments is not so far-fetched: the argument continues next time.