Emergent Physical Properties?
Author
Just for discussion purposes at the moment... Im not building this (it would be highly inefficient).
Suppose we simulate rigid bodies as particles (simple point masses with position and mass, but no other properties) joined by springs to one another.
The springs would act like macroscale atomic bonds
Are physical properties such as moment of inertia, and behaviours such as gyroscopic precession going to be emergent from such a system?
Or are such high level features intangibles that need to be represented explicitly?
perhaps this delves into the nature of how such things work in reality?
should be interesting discussion...
Quote:Original post by haphazardlynamed
Are physical properties such as moment of inertia, and behaviours such as gyroscopic precession going to be emergent from such a system?
in one word: yes.
Quote:Original post by haphazardlynamed
Just for discussion purposes at the moment... Im not building this (it would be highly inefficient).
Suppose we simulate rigid bodies as particles (simple point masses with position and mass, but no other properties) joined by springs to one another.
The springs would act like macroscale atomic bonds
Are physical properties such as moment of inertia, and behaviours such as gyroscopic precession going to be emergent from such a system?
Or are such high level features intangibles that need to be represented explicitly?
Yes, as Eelco noted, point masses with springs give rise to correct motion. However, at higher angular velocites, steps must be taken to keep the simulation accurate. The integrator must be symplectic if accuracy is desired.
One of the ways I validated and tested my rigid body physics engine was to create tetrahedral point-mass systems with distributions that exactly matched the inertia tensor and total mass of my rigid bodies. I would then switch between the two methods in real-time, verifying no significant changes.
Here's an old demo that shows correct precession behavior (not explicitly modeled; is emergent).
You could view a rigid body as a collection of point masses connected by massless sticks which do not change length. If the sticks can change their length slightly but prefer to be at some restlength (a massless stiff spring), it looks reasonable to say that both systems will behave similar but slightly different.
Here's a 2D particle based physics engine with no actual rotation equations, even though the emergent system has full dynamics and deformable objects.
http://www.aidspan.org/alec/physical/overview_physics.htm
http://www.aidspan.org/alec/physical/overview_physics.htm
yeah, [google] "Advanced character physics", by Jakobsen. It does just that. stiff constraints + particles. I've got a few demoes about that.
This topic is closed to new replies.
Advertisement
Popular Topics
Advertisement
