FW: [ODE] Replacing the ODE's LCP solver

[Sergio Valverde]

I'm sorry, there was a mistake in the previous e-mail:

"I'm interested in experimentation with alternative LCP solvers within ODE 
because this is the bottleneck of the physics simulation. The main problem
is that I *CAN'T* easily figure out how express the current Dantzig hi-lo
LCP 
problem in terms of (more general) mixed LCP. That is:"

-----Original Message-----
From: Sergio Valverde 
Sent: martes, 14 de enero de 2003 15:56
To: ode@q12.org
Subject: [ODE] Replacing the ODE's LCP solver


Hello,

I'm interested in experimentation with alternative LCP solvers within ODE 
because this is the bottleneck of the physics simulation. The main problem
is that I can easily figure out how express the current Dantzig hi-lo LCP 
problem in terms of (more general) mixed LCP. That is:

Solve for w,z

| 0 |   | P  R | | x |   | u |
|   | - |      | |   | = |   |
| w |   | Q  S | | z |   | v |
 
under complementary constraints  w>=0, z>=0, wz = 0

It is possible? and in case it is,  how can be done?

Best,
Sergi