How to use FICO xpress Server to solve multiple optimization concurrently

Hello, you should be able to start multiple runs on a class 3 server license of Xpress.

Can you provide more details about what you are trying to achieve with concurrent solving?

It is recommended to set the THREADS parameter for each problem so that the number of concurrent runs * THREADS is less than or equal to the number of available threads.

We will have to run long running calculations in background so just want to understand limits ?

Hello Jitendra,

For Linux-based containers, the Solver will respect the container-imposed limitations for memory and CPU limits.

By default, there are no limits - the Solver can potentially use all the threads and memory made available to it.

You can specify additional limits e.g., with the THREADS parameter as suggested in my response above, and also memory limits: MAXMEMORYHARD, MAXMEMORYSOFT, etc.

Depending on the size of the container, you might consider limiting the number of concurrent solves if the memory footprint of the Solver is high.

In the Python API you can set a control parameter as follows.

prob.controls.threads=4 # prob is the Xpress problem object. 

Hope this helps.