Generalized Moving Peaks Benchmark (dynamic minimization wrapper)

A stateful dynamic optimization benchmark that wraps GMPB’s native maximization objective and negates it to fit this package’s minimization-oriented optimization API.

See Optimization Problem Catalog for the optimization family index.

Quick Facts

Field	Value
Problem ID	gmpb_default_dynamic_min
Problem Family	optimization
Implementation	design_research_problems.problems.optimization._gmpb:GMPBOptimizationProblem
Capabilities	baseline-solver, bounded-variables, external-adapter, statement-markdown
Study Suitability	none
Tags	optimization, continuous, dynamic, benchmark, single-objective

Taxonomy

Formulation

nonlinear_program

Convexity

not_guaranteed

Design Variable Type

continuous

Is Dynamic

yes

Orientation

mathematical

Objective Mode

single

Constraint Nature

hard

Bounds Summary

five continuous variables bounded on [-100, 100]

Tags

optimization, continuous, dynamic, benchmark, single-objective

Statement

This packaged optimization problem wraps the external Generalized Moving Peaks Benchmark (GMPB). GMPB is defined as a dynamic maximization benchmark whose evaluation counter advances through changing environments.

To fit the optimization API in this package, the wrapper reports the negated native GMPB objective, so lower values are better. Evaluations remain stateful: each call consumes benchmark budget in the current environment, and the environment changes automatically when the configured change frequency is exhausted.

Problem Shape

Field	Value
Design Variable Count	5
Bound Summary	five continuous variables bounded on [-100, 100]
Total Constraint Count	0
Equality Constraint Count	0
Inequality Constraint Count	0

Variable Bounds

Variable	Lower Bound	Upper Bound
x[0]	-100	100
x[1]	-100	100
x[2]	-100	100
x[3]	-100	100
x[4]	-100	100

Manifest Parameters

Key	Value
change_frequency	1000
component_count	10
dimension	5
environment_count	100
lower_bound	-100
seed	7
upper_bound	100

Library Interface

• generate_initial_solution(seed=None)

• objective(x)

• evaluate(x)

• solve(initial_solution=None, seed=None, maxiter=200)