Re: Problems with AI-boxing

From: Michael Wilson (
Date: Sat Aug 27 2005 - 04:07:04 MDT

Phil Goetz wrote:
> I think it's interesting that, were one to build a simulation of
> a universe, it would be impossible to use Newtonian physics.
> First, you'd have roundoff errors for objects moving at
> very high speeds - but implementing relativity theory
> could solve that problem.

On what are you basing this statement? I have some experience of
physical simulation as I wrote a couple of physics engines when
I used to work as a professional games developer. If you are
using an integrative model, i.e. a model with a discrete state
for every frame, then low speeds are actually much more prone
to roundoff errors than high speeds, because the distance moved
each frame is closer to the co-ordinate resolution. A continuous
model essentially doesn't suffer from this issue, because the
velocity vector is multiplied by the entire elapsed time since
the last inertial event every time you compute a position or solve
for a collision. But in any case, as far as I can see relativity
would make things harder, not easier.

> Then, you'd be unable to implement physical interactions
> to the infinitely-fine level of resolution required by
> Newtonian physics. So you would probably choose some very fine
> level of resolution, and say that for interactions at that
> resolution or less, instead of simulating them completely,
> you'd use a probabilistic approximation.

Aren't you talking about quantum physics, which does indeed have
both quantisation and probabilistic results, not relativity?

 * Michael Wilson

To help you stay safe and secure online, we've developed the all new Yahoo! Security Centre.

This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:52 MDT