A comprehensive suite of hypervolume-related algorithms designed for efficient multi-objective optimization
The library documentation is available at ReadTheDocs: https://moda-put.readthedocs.io/en/latest The installation process is described in the documentation.
After the installation, you should be able to include the library files in any project.
The MODA library follows a Data-Solver-Result pattern. DataSets hold the objective function values, Solvers encapsulate the algorithms (like Hypervolume calculation), and Results return the metrics.
This example demonstrates loading a sample dataset, configuring the IQHVSolver (Improved Quick Hypervolume), and calculating the resulting hypervolume.
// 1. Load Data
// Loads sample data from file (file must be accessible in execution directory)
DataSet* ds = moda::DataSet::LoadFromFilename("../../sample-file/data_6_500_convex_triangular_1");
// Display data points one by one
std::cout << "Loaded Data Points:\n";
for (auto p : ds->points)
{
for (int i = 0; i < ds->getParameters()->NumberOfObjectives; i++)
std::cout << p->ObjectiveValues[i] << " ";
std::cout << "\n";
}
std::cout << "---------------------------------\n";
// 2. Prepare Data and Solver
ds->typeOfOptimization = moda::DataSet::minimization; // Set optimization type
ds->normalize(); // Apply normalization
moda::IQHVSolver solver;
moda::IQHVParameters params;
// 3. Set Solver Parameters
// Sets calculation formulas for the reference points (zeroone substitues zeroes and ones vectors for reference points)
using RefStyle = moda::IQHVParameters::ReferencePointCalculationStyle;
params.BetterReferencePointCalculationStyle = RefStyle::zeroone;
params.WorseReferencePointCalculationStyle = RefStyle::zeroone;
// 4. Run Solve and Display Result
// Solve returns a pointer to a HypervolumeResult
moda::HypervolumeResult* result = solver.Solve(ds, params);
std::cout << "Calculated Hypervolume: " << result->HyperVolume << std::endl;
std::cout << "Elapsed Time (ms): " << result->ElapsedTime << std::endl;
// Cleanup (optional but recommended)
delete ds;
delete result;This project follows a structured branching model designed to support active development, feature expansion, and Python API integration while keeping the main branch stable for releases and packaging.
 |
|---|
Purpose: The main branch serves as the source of truth for stable library releases.
It contains code intended for:
- Versioning
- Packaging
- Distribution (e.g., vcpkg, PyPI, internal deployment)
Characteristics: - Always stable - Updated through pull requests from dev or python-wrap - Tagged for official releases
Purpose: The dev branch is the central hub for active development of the C++ library.
Characteristics:
- Contains ongoing improvements
- Integrates completed feature branches
- Eventually merged into main once stable
- All standard development flows into dev before being prepared for release.
Purpose: Feature branches are used for isolated, focused development of new features or fixes.
Characteristics:
Named based on their feature
e.g., 1-matrix-ops, 2-simd-optimizations
Branched from dev
Merged back into dev after completion via pull request
Purpose: The python-wrap branch is dedicated solely to the Python C-API wrapper of the C++ library.
Characteristics:
Used for independent development of Python bindings
Branched from main
Merged back into main when new wrapper features are complete