Synthetic Benchmarks

The Synthetic benchmark collection includes classic mathematical test functions commonly used in optimization research.

Available Problems

• bocode.Synthetics.Ackley

• bocode.Synthetics.Bukin

• bocode.Synthetics.DixonPrice

• bocode.Synthetics.Goldstein

• bocode.Synthetics.Goldstein_Discrete

• bocode.Synthetics.Griewank

• bocode.Synthetics.Levy

• bocode.Synthetics.Michalewicz

• bocode.Synthetics.Powell

• bocode.Synthetics.Rastrigin

• bocode.Synthetics.Rosenbrock

• bocode.Synthetics.Styblinski-Tang

• bocode.Synthetics.Beale

• bocode.Synthetics.Cosine8

• bocode.Synthetics.DropWave

• bocode.Synthetics.EggHolder

• bocode.Synthetics.Hartmann3D

• bocode.Synthetics.Hartmann6D

• bocode.Synthetics.HolderTable

• bocode.Synthetics.Shekelm5

• bocode.Synthetics.Shekelm7

• bocode.Synthetics.Shekelm10

• bocode.Synthetics.Shekel

• bocode.Synthetics.SixHumpCamel

• bocode.Synthetics.ThreeHumpCamel

• bocode.Synthetics.ConstrainedGramacy

• bocode.Synthetics.ConstrainedHartmann

• bocode.Synthetics.ConstrainedHartmannSmooth

• bocode.Synthetics.PressureVessel

• bocode.Synthetics.WeldedBeamSO

• bocode.Synthetics.TensionCompressionString

• bocode.Synthetics.SpeedReducer

• bocode.Synthetics.SVM

Example Usage

import bocode
import torch

# Create a synthetic benchmark problem
problem = bocode.Synthetics.Goldstein_Discrete()

# Evaluate at a point
x = torch.Tensor([[0.0] * problem.dim])
values, constraints = problem.evaluate(x)

print(f"Goldstein Discrete function value at origin: {values[0]}")

Output:

Goldstein Discrete function value at origin: tensor([-600.])