MultiObjectiveProblems.jl is a curated Julia library of benchmark problems for multiobjective optimization. It provides source-traceable problem implementations, a consistent evaluation API, registered analytical derivatives, and metadata-based catalog queries.
import Pkg
Pkg.add("MultiObjectiveProblems")See the installation guide for the development version, project environments, and local checkouts.
using MultiObjectiveProblems
using Random
prob = DTLZ2()
lower, upper = recommended_bounds(prob)
rng = MersenneTwister(1234)
α = rand(rng, prob.nvar)
x = lower .+ α .* (upper .- lower)
values = eval_f(prob, x)
J = eval_jacobian(prob, x)
names = filter_problems(has_jacobian = true)The MultiObjectiveProblems.jl documentation
contains the quick start, task-oriented guides, mathematical formulations,
API reference, and bibliography. Instructions for building it locally are in
docs/README.md.
- Curated, source-traceable benchmark implementations.
- Allocating and in-place APIs for objectives and registered derivatives.
- Catalog filters for dimensions, bounds, constraints, and derivative support.
- Registered variable bounds, and a documented finite box for the problems whose formulation declares none.
- Per-family formulations, metadata, usage examples, and references.
Contributions are welcome. See CONTRIBUTING.md for the
development workflow, testing requirements, and source-evidence checklist.
If you use MultiObjectiveProblems.jl in research, please cite the software using the
metadata in CITATION.cff.
MultiObjectiveProblems.jl is distributed under the MIT License.