evobench 0.1.0

Benchmarks for large scale, model-based optimization.

Evobench

Evobench is a collection of benchmark problems dedicated for model-based large scale optimization.

Overview

This package contains following problems.

Discrete

• trap
• step trap
• bimodal
• step bimodal
• HIFF
• Ising Spin Glass

Continuous

• trap
• multimodal
• step multimodal
• sawtooth

Getting started

pip install evobench

import evobench


trap = evobench.discrete.Trap(block_size=5, repetitions=3)
initialization = evobench.discrete.initialization.Uniform(population_size=4e3)

population = initialization.initialize_population(trap.genome_size)
fitness = trap.evaluate_population(population)

You can also evaluate single solution.

fitness = trap.evaluate_solution(population.solutions[0])

Every time you're evaluating solutions we increment ffe counter. You can access it through benchmark instance.

print(trap.ffe)

Ising Spin Glass

To instantiate ISG you need to pass specific problem configuration.

from evobench.discrete import IsingSpinGlass


isg = IsingSpinGlass('IsingSpinGlass_pm_16_0')

You can find 5,000 instances at evobench\discrete\isg\data folder. Instances vary in length and complexity.

How to implement your own function

Fully separable

You need to inherit Separable class from evobench.separable. Then just implement def evaluate_block(self, block: np.ndarray) -> int method. Best follow evobench.discrete.trap implementation.

Other

Inherit Benchmark class from evobench.benchmark. Then implement def _evaluate_solution(self, solution: Solution) -> float method.

Coming soon

We'll be adding more problems in near future. If you're looking for any particular problem, please mail us or open an issue.

We're thinking about interactive visualizations, so you can sample the space and check how it looks. It's easier than digging through definitions.

We're working on linkage quality metrics. Once they're published, we'll incorporate them to this package.