[CLPEX memo] Run DO Python API in COS installed environment

At the beginning

The standard development environment for IBM's optimization solution CPLEX is ** IBM ILOG CPLEX Optimization Studio ** (abbreviated as ** COS **). On a PC with COS installed, model development is usually done in an optimization-specific language called OPL using an Eclipse plug-in, but development using the Python API is also possible. I will write a memo of the procedure.

Reference link

Setting up an optimization engine

Premise

Check the operation

MacOS: 10.14.6 CPLEX Optimization Studio: 12.9

It was done in the environment of.

Setup steps

Introduction of COS

As for COS, all installations are done by default in the usual way.

Introduction of anaconda

The test was done with Jupyter Notebook. As for anaconda, which is the premise, everything is done by default in the usual way.

Confirmation of COS / cplex installation path

Check the COS installation path with the following command.

$ ls -l /Applications | grep CPLEX

In the environment at hand

drwxrwxr-x  11 [email protected]  admin  352  9 13 10:19 CPLEX_Optimizer129

It was the result.

Check Python version

Check with the following command.

$ python -V

In my environment, I got the following results:

Python 3.7.3

Setting the environment variable PYTHONPATH

Based on the results so far, add a line like the following to ~ / .bash_profile.

export PYTHONPATH=/Applications/CPLEX_Optimizer129/cplex/python/3.7/x86-64_osx

Please modify the part of 3.7 / x86-64_osx according to the Python version and platform.

Introducing the library

Install the DO Python API library with the following pip command.

$ pip install docplex

Operation check

This completes the necessary preparations. With the environment variable PYTHONPATH set

$ jupyter notebook

Then, start Jupyter Notebook.

Solving the traveling salesman problem with Watson Download and load the Jupyter Notebook file introduced in.

After that, Shift + Enter to execute the cells in order, and CPLEX should find the best route as shown in the figure below.

Relationship between functional constraints and licenses

COS also has an evaluation version that is available for free. If you use the evaluation version, make the above settings, and try to solve the traveling sales problem, you will see a message like the one below.

CPLEX Error  1016: Community Edition. Problem size limits exceeded. Purchase at https://ibm.co/2s0wqSa.

This is literally an error caused by the functional restrictions of the evaluation version (up to 1000 rules). In the case of the evaluation version, if you correct the number of points (N) = 5 in the traveling sales problem and execute it, the functional constraint will not be hit and it will end normally.