LSSA

LSSA (Lost Sheep Search Algorithm) is an adaptive continuous black-box optimisation algorithm.

This package provides the frozen research implementation used for out-of-sample validation. Development candidates such as the B20 branch are kept separate from the stable package release.

Installation

pip install lssaopt

The PyPI distribution is named lssaopt; the Python import remains lssa.

Quick start

import numpy as np
from lssa import minimize_lssa

class Sphere:
    d = 5
    lo = -5.0
    hi = 5.0

    def eval(self, x):
        x = np.asarray(x, dtype=float)
        return float(np.dot(x, x))

x_best, f_best = minimize_lssa(Sphere(), max_fes=2000, seed=0)
print(f_best, x_best)

The problem object must provide:

• d: dimensionality
• lo: scalar lower bound
• hi: scalar upper bound
• eval(x): objective function returning a scalar

API

minimize_lssa(problem, max_fes=4000, population=30, seed=0, return_metadata=False)

LSSA performs minimisation. For reproducibility, set seed explicitly.

Status

Initial PyPI release: 0.1.0.