Basics

Table of Contents

• What is idol_cl?
• Solving Your First MILP
• Bilevel Problems
• Two-Stage Robust Problems

Describes the basic usage of the idol_cl tool.

What is idol_cl?

idol_cl is a command-line tool for solving mathematical optimization problems, including:

• Mixed-Integer Linear Programs (MILPs);
• Bilevel optimization problems;
• Robust and two-stage robust optimization problems.

It allows you to model, solve, and analyze optimization problems directly from the command line.

To install it, please refer to the Installation page.

Solving Your First MILP

To solve a given MILP stored in an .lp file, all you need to do is

idol_cl solve model.lp

If you want to set a time limit and use a specific method, run

idol_cl solve model.lp --time-limit 3600 --method GLK

To see which solution methods are available for a given problem, use the list-methods subcommand.

idol_cl list-methods model.lp

Sampled Output:

-- No configuration file loaded
-- The main input file is model.lp
-- Detected: problem type is MILP
-- Detected: applicable methods are [GUROBI, CPLEX, HIGHS, GLPK, JUMP]
 
Details (all methods):
 
 - HIGHS: High Performance Software for Linear Optimization [https://highs.dev/]
 - JUMP: Calls the Julia package JuMP [https://jump.dev/]
 - GLPK: GLPK (GNU Linear Programming Kit) [https://www.gnu.org/software/glpk/]
 - CPLEX: IBM ILOG CPLEX Optimization Studio [https://www.ibm.com/fr-fr/products/ilog-cplex-optimization-studio]
 - GUROBI: Gurobi Optimizer [https://www.gurobi.com/]

If no method is specified, idol_cl automatically selects a suitable solver for the detected problem type.

Bilevel Problems

Bilevel problems are represented using a pair of files:

• an .mps file describing the main model,
• an .aux file describing the follower problem.

For more information, see the dedicated page on Input Format for Bilevel Problems.

Given a bilevel problem, you can solve it by running

idol_cl solve model.mps --bilevel follower.aux

Some commands require more parameters than others. For instance, here is how to solve an LP-LP bilevel problem using its KKT reformulation using big-M values.

idol_cl solve model.mps --bilevel follower.aux --method KKT-BIGM --bound-provider bounds.txt

For more information, see our tutorial on Solving Bilevel Problems via its KKT-Based Single-Level Reformulation.

Here again, to list the available methods for your problem class, you may use

idol_cl list-methods model.mps --bilevel follower.aux

Finally, if the bilevel problem does not have coupling constraints, it is possible to address its pessimistic variant by appending the --pessimistic flag. If so, the relaxation-and-correction scheme from Zeng (2025) will be applied.

idol_cl solve model.lp --bilevel follower.aux --method KKT-SOS1 --pessimistic

Two-Stage Robust Problems

This section is still a work in progress.