Good morning,
I’m approaching to use NVIDIA libraries to solve linear optimization problems. I see that potentially this could be done with 2 approaches: CuOpt and CuSolver.
CuOpt seems a more complete framework, but I don’t understand if it is free for unlimited time using the container that could be downloaded on GTC or it needs a license to be used
Furthermore, is CuOpt based under the hood on CuSolver or they are different approches?
Best regards
Thank you for your interest in cuOpt.
Yes, cuOpt is a higher level application for large scale optimization where as cuSolver is a primitive offering a collection of dense and sparse direct linear solvers to solve specifically linear systems.
cuOpt solvers for mathematical optimization are not available today. There is a VRP solver that is available for free and as part of NVAIE for enterprises. Please watch out for GTC updates related to cuOpt.
Link to get the 90 day trial for Vehicle Routing Solver - NVIDIA Enterprise Account Registration
Link to access NIM catalog for Vehicle Routing Solver - Try NVIDIA NIM APIs
Thank You,
cuOpt Team
Hi @preethim ,
thanks for reply. I have a couple of further questions:
- in order to use cuOpt service an Nvidia AI Enterprise account (not-free) is needed. Is it?
- from Example section of CuOpt user guide seems that is can be used also to solve general LP and MLP models (Linear Programming — NVIDIA cuOpt)
Regards
Hi,
could someone please confirm me that CuOpt could be used also to solve general LP and MLP problems and not only VRP problems?
Regards
Hi @liviolima80 yes absolutely you can use CuOpt for LP and MLP problems - and it seems like you’ve already found the examples section in our docs! Linear Programming — NVIDIA cuOpt
Hi @sophwats ,
if CuOpt could be used for LP and MLP, what the meanning of the note in the Intro section of user guide?
Note
Linear Programming (LP) and Mixed Integer Linear Programming (MILP) are Early Access features and are currently open to only select customers.
Regards