[SOLVED] ForceField with implicit iterative solver

[Bassam]

Hi,

I have written a custom ForceField and implemented both addForce and addDForce. The simulation runs fine when using an explicit solver, but when I switch to an implicit iterative solver the object is totally rigid and I cannot pull on any vertex. When debugging, the dx input argument to addDForce seems to always be 0.

In case it might matter, I am using a 1D ForceField with a 1D MechanicalObject custom mapped to another 3D MechanicalObject as outlined in my previous post Using model reduction in SOFA

Any ideas what I might be doing wrong?

Thanks!
—
Bassam

[Hugo]

Hey @beejay ,

alright then, if the explicit simulation gives you the right behaviour/result (are your sure about it?), then the addForce function is well implemented.

About the implicit simulation:
– are you sure the object is not moving (even small value in the velocity field of the MechanicalObject) ?
– If so, are you sure to have both the integration scheme (ODESolver, e.g. EulerImplicitSolver) AND a linear solver (e.g. CGLinearSolver) ?
– what are the value of iterations / threshold / tolerance in the CG (if you are using one) ?
– finally, the problem could also come from your mapping

Could you share us the explicit and implicit scene ?

Hugo

[Bassam]

Hi Hugo,

You nailed it, I had a problem in my mapping’s applyJT where I was overwriting the output value instead of adding to it, therefore overwriting my ForceField’s contribution. The overall simulation still ran with an explicit solver, albeit with no internal forces, and with a heavy dose of wishful thinking I assumed it worked fine..

I have got it working now, thank you!

Bassam

[Bassam]

Closing this topic!

[Author]

Posts