benchmarkfcns.Composition

class benchmarkfcns.Composition(constant_C=2000.0)

Bases: Composition

A high-performance engine for creating hybrid/composition benchmark functions. This class allows blending multiple base functions using CEC-standard exponential weighting, shifting, and rotation.

__init__(constant_C=2000.0)

Methods

__init__([constant_C])
add(function, shift[, rotation, sigma, ...])	Adds a base function component to the composition.
add_component(self, fcn, , ]], ...)	Adds a base function component to the composition.
evaluate(x)	Evaluates the composed function for a batch of points.
set_constant_C(self, C)	Sets the C constant for height normalization (default 2000.0).

add(function, shift, rotation=None, sigma=1.0, lambda_val=1.0, bias=0.0, f_max=1.0)

Adds a base function component to the composition.

Parameters:

• function (str | callable) – Name of the built-in function (e.g. ‘ackley’) or a pointer.

• shift (np.ndarray) – 1D array of size N for shifting the optimum.

• rotation (np.ndarray | None) – 2D array of size N-by-N for coordinate rotation. Defaults to Identity matrix.

• sigma (float) – Convergence range / basin of attraction size.

• lambda_val (float) – Scaling parameter for the landscape.

• bias (float) – Internal bias for this component.

• f_max (float) – Estimated maximum value for height normalization.

evaluate(x)

Evaluates the composed function for a batch of points.

Parameters:

x (ndarray) – Matrix of size M-by-N (M points in N dimensions)

Return type:

ndarray

Returns:

1D array of size M containing the composed scores.