Improving the Solution Interface and the Problem Conversion Methods

[kwesiRutledge]

There's a couple of things that have become apparent as we try to make our first solver ready for "production". Here are some of the things that I noticed that seem to be difficult.

• When solving an LP, we apply some transforms and then solve the transformed problem.
  • We thus have optimal values for variables that the user has never seen/doesn't have access to.
• When trying to map the transformed problem to the original problem, we do not have a mapping that explains which which transformed variable makes up which original variable.

How can we make sure that we can easily transform all data back to the original problem?