Comments | particleswarm: Faster than real-time

particleswarm

Faster than real-time

Jun 08, 2004 12:48

Will someone please tell DARPA and the engineers here at Lockheed, that "faster than real-time" simulation is IMPOSSIBLE!

Comments 2

2bev July 12 2004, 15:18:12 UTC

It is? Howso? Do you mean writing a simulation model can't be done in realtime?

Bev

particleswarm July 13 2004, 07:26:31 UTC

I should note that I was having a bad day when I wrote that entry. "Faster than real-time" is the latest buzz-phrase to make the rounds. DARPA is demanding 100-1000x speedup in battlefield sims, while maintaining fidelity. With such complex sims, this is a definite case of sensitivity to initial conditions. The faster one pushes the sim, the shorter the "reliability window" becomes; approaching zero in the limit. Eventually the simulation would have to be so abstract that it becomes unreliable or useless. At least that's what a group of us think at ATL. We could be off-base, as we are not "simulation theory" experts. I realize it isn't as black and white as my entry suggests. It all depends on scale, fidelity requirements, and "satisfiablity."