Linear Optimization and Using Solvers in DSS

delrussa

I am working on a Supply Chain Optimization solution. The ETL is complete and python code written to converge the optimization. Normally, the solver (an executable file on windows such as glpsol.exe, Gurobi, etc..) exists on my machines path and is found by the executing program.

Is there a way to integrate a solver in a linear optimization solution with Dataiku? and if so how?

---

Operating system used: Windows

Sarina

Hi @delrussa
,

Are you encountering any specific errors when attempting to use one of the solvers?

In general, you'll need to ensure that any relevant packages are installed on the DSS server, and that the solver exists in the PATH for DSS. You will need to work with your DSS admin to ensure that the relevant steps are taken on the DSS server.

For example, for glpk, there are several system packages that must be installed on the DSS server in order to use it.

glpk-utils libglpk-dev glpk-doc python-glpk

In addition, the executable glpsol is installed and in the PATH for DSS. Your admin can also check that glpsol is installed and ensure that the PATH for DSS includes the path to glpsol:

which glpsol

If the PATH for DSS needs to be updated, your admin can add the path to glpsol to the <DATADIR>/bin/env-site.sh file like so:

export PATH=$PATH:/path/to/glpsol

Similarly, with Gurobi, your admin could follow the instructions to install Gurobi as described here: https://www.gurobi.com/documentation/9.5/quickstart_linux/software_installation_guid.html. The environment variables can be added to the <DATADIR>/bin/env-site.sh file similar to above.

I hope that information is helpful. If your admin runs into any issues installing these, or if you run into any issues using them in DSS, you can open a ticket with us.

Thanks,
Sarina