BBOB Functions

Examples using BBOB benchmark functions.

---

Basic BBOB Usage

import numpy as np
from surfaces.test_functions.benchmark.bbob import (
    Sphere,
    RosenbrockRotated,
    RastriginRotated,
    Schwefel,
)

# Create 10D functions
functions = [
    Sphere(n_dim=10),
    RosenbrockRotated(n_dim=10),
    RastriginRotated(n_dim=10),
    Schwefel(n_dim=10),
]

for func in functions:
    space = func.search_space
    sample = {k: np.random.choice(v) for k, v in space.items()}
    result = func(sample)
    print(f"{func.__class__.__name__}: {result:.4f}")

---

BBOB Function Groups

"""Evaluate functions from different BBOB groups."""

import numpy as np
from surfaces.test_functions.benchmark.bbob import (
    # Separable
    Sphere,
    EllipsoidalSeparable,
    # Multimodal
    RastriginRotated,
    SchaffersF7,
    # Weak structure
    Schwefel,
    Gallagher101,
)

groups = {
    "Separable": [Sphere, EllipsoidalSeparable],
    "Multimodal": [RastriginRotated, SchaffersF7],
    "Weak structure": [Schwefel, Gallagher101],
}

for group_name, func_classes in groups.items():
    print(f"\n{group_name}:")
    for func_class in func_classes:
        func = func_class(n_dim=5)
        space = func.search_space
        sample = {k: np.random.choice(v) for k, v in space.items()}
        result = func(sample)
        print(f"  {func_class.__name__}: {result:.4f}")

---

Note

BBOB functions are designed for rigorous algorithm comparison. See the COCO platform documentation for standard benchmarking protocols.