GitHub

ModelCompare.jl

A Julia package for comparing two optimization models (LP/MPS format) and producing a detailed diff report covering variables, bounds, objective function, and constraints.

Installation

using Pkg
Pkg.add(url="https://github.com/rafabench/ModelCompare")

Quick Start

using ModelCompare

result = compare_models(
    "model1.lp",
    "model2.lp",
    outfile = "comparison.txt",
    tol = 1e-3,
)

The result is a NamedTuple with four fields:

  • variables — a VariablesDiff showing which variables are unique to each model vs shared
  • bounds — a BoundsDiff showing variable bound differences
  • objective — an ObjectiveDiff showing objective sense and coefficient differences
  • constraints — a ConstraintElementsDiff showing constraint coefficient and bound differences

Output Sections

Variable Names

Lists variables that belong only to Model 1, only to Model 2, or both.

Variable Bounds

Shows bound differences for shared variables and bounds for variables unique to each model.

Objective Function

Shows coefficient differences in the objective function. If the optimization senses differ (min vs max), that is reported as well.

Constraints

Compares constraints with matching names across both models. Unmatched constraint names are not reported. For each matched constraint, coefficient and set (bound) differences are shown.

Sorting Models

You can canonicalize a model file by sorting its variables and constraints alphabetically:

sort_model("model.lp")  # creates model.lp.sorted

CLI Usage

After compilation (see compile/compile.jl):

ModelCompare --file1 model1.lp --file2 model2.lp -o result.txt -t 0.001