[ODE] Constraint theorizing

[Adam D. Moss]

Henri Hakl wrote:
> I was just wondering, if there is a theoretical reason why all the
> physics are computed into one giant contraint matrix.
> 
> Consider the following simple generic case:
> 
> The floor, and 2 boxes falling onto it. The boxes are quite far apart
> and won't even come remotely close to touching each other.
> 
> Instead of solving the constraints in one (for argument's sake) 20x20
> matrix, why not use two 10x10 matrices that handle the boxes
> seperately?

This is pretty much what ODE does.

From section 3.2.1 in the manual:
"Each island in the world is treated separately when the
simulation step is taken. This is useful to know: if there
are N similar islands in the simulation then the step
computation time will be O(N)."

An island is defined in the manual as a group of bodies connected
via joints, though it should be noted that two or more islands can
temporarily become a single larger island for the duration of a
simulation step if contact joints have been created between them
(~O(N^3) havoc then ensues).

HTH,
--Adam
-- 
Adam D. Moss   . ,,^^   adam@gimp.org   http://www.foxbox.org/   co:3
busting makes me feel good
kthx bye