Safety challenges for AI agents' ability to learn and act in desired ways in relation to biologically and economically relevant aspects. The benchmarks are implemented in a gridworld-based environment. The environments are relatively simple, just as much complexity is added as is necessary to illustrate the relevant safety and performance aspects.

Surrogate-Based Architecture Optimization toolbox

A dependency free library of standardized optimization test functions written in pure Python.

Benchmark library of vector-valued optimization problems in Julia, with analytic per-objective gradients, filtering functions, and a unified interface for testing and comparisons of multi-objective solvers.

A Julia package for solving multi-objective optimization problems with composite structure (F = f + h). Implements Conditional Gradient, Proximal Gradient, and Partially Derivative-Free algorithms that operate directly on the vector-valued objective, without scalarization or heuristics (direct / vector-optimization methods).

Multi-objective adversarial perturbations on LiDAR point clouds using NSGA-III to evaluate SLAM robustness. Integrates MOLA SLAM with Isaac Sim via ROS2.

Multi-objective Bayesian optimization with decoupled objectives using knowledge gradient and expectation over linear scalarisations

Systematic runaway-optimiser-like LLM failure modes on Biologically and Economically aligned AI safety benchmarks for LLM-s with simplified observation format. The benchmark themes include multi-objective homeostasis, (multi-objective) diminishing returns, complementary goods, sustainability, multi-agent resource sharing.

Extended, multi-agent, and multi-objective (MaMoRL / MoMaRL) gridworld environments building framework based on DeepMind's AI Safety Gridworlds. This is a suite of reinforcement learning environments illustrating various safety properties of intelligent agents. It is made compatible with OpenAI's Gym/Gymnasium and Farama Foundation PettingZoo.

A Julia wrapper for Bensolve

Explanation system for semi-supervised multi-objective optimization

A comprehensive Python implementation of MOEA/D (Multiobjective Evolutionary Algorithm based on Decomposition), a state-of-the-art algorithm for solving multiobjective optimization problems. This implementation is based on the seminal work by Zhang and Li (2007).

Multi-Objective Multi-Agent RL with non-linear utility functions

Paxplot is a Python visualization library for parallel axis, or parallel coordinate, plots.

Multi-Objective Reinforcement Learning algorithms implementations.

The cMIBACO implementation for lightly robust solutions in MOGenConVRP under uncertainty.

A Python 3 gradient-free optimization library

Multi-objective Gymnasium environments for reinforcement learning

2D and 3D plots of linear/integer programming models in R

Official Implementation: Boundary Decomposition for Nadir Objective Vector Estimation, NeurIPS 2024; Boundary Decomposition for Finding Nadir Objective Vector in Multi-Objective Discrete Optimization, AAAI 2025.