1 / 6

Binh research 08 Dec 2009

Binh research 08 Dec 2009. Working on ST and non-convex time stepper within Bullet DONE with initial solver for frictionless ST. Working on non-convex time stepper (required changes in collision detection). Solver for frictionless ST.

xiu
Télécharger la présentation

Binh research 08 Dec 2009

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Binh research 08 Dec 2009 • Working on ST and non-convex time stepper within Bullet • DONE with initial solver for frictionless ST. • Working on non-convex time stepper (required changes in collision detection)

  2. Solver for frictionless ST • Right now use OOQP+MA57 to solve a simple QP from frictionless ST • Future: • OOQP+Cholmod(should be faster and more robust than MA57) • OOQP+UMFPACK(work with indefinite system)+LUMOD(sparse update)

  3. Solver for frictionless ST • Once done, need to gather data on how much KKT system change during 1 step. If small enough then can use sparse update (LUMOD) • If sparse update not efficient then may have to work on way to warm-start OOQP (warm start Interior Point Method is tricky) • Simple optimization = group column in same contact and pivot them at the same time should help A LOT (need to try)

  4. ST with linearized friction • Should be similar to frictionless case • Structure is different so may need a slightly different optimization

  5. ST with quadratic friction • Not spend much time on this subject yet • OOQP implements Mehrota-Gondzio predictor corrector Interior Point Method and we should be able to extend it to solve SQP problem (I had some papers on it but havent read them yet)

  6. This week • Get frictionless OOQP solver done and get some statistical data on a big physics scene. • Work on OOQP-Cholmod as it should be 3-5x faster (benchmark on convex QP showed that )

More Related