Multi-Objective Optimization¶

This guide shows how to work with multi-objective optimization problems in BOCoDe.

Basic Example¶

Here’s an example of using BOCoDe with a multi-objective problem:

import bocode
import numpy as np
from pymoo.algorithms.moo.nsga2 import NSGA2
from pymoo.optimize import minimize
from pymoo.core.problem import Problem
import torch

# Create a multi-objective benchmark problem
problem = bocode.Engineering.CarSideImpact()

# Wrap the problem as a PyMOO Problem
class CarSideImpactProblem(Problem):
    def __init__(self):
        super().__init__(n_var=problem.dim,
                        n_obj=problem.num_objectives,
                        n_constr=problem.num_constraints,
                        xl=np.array([b[0] for b in problem.bounds], dtype=float),
                        xu=np.array([b[1] for b in problem.bounds], dtype=float),
                        )

    def _evaluate(self, x, out, *args, **kwargs):
        values, constraints = problem.evaluate(torch.Tensor(x))
        out["F"] = values.numpy()
        out["G"] = constraints.numpy()

# Setup the algorithm
algorithm = NSGA2(pop_size=100)

# Optimize
res = minimize(CarSideImpactProblem(),
              algorithm,
              verbose=True)

# Get Pareto front
pareto_front = res.F

Features for Multi-Objective Problems¶

Pareto front visualization
Performance metrics (hypervolume, IGD)
Constraint handling
Integration with multi-objective optimization frameworks