Lipschitz-inspired HALRECT algorithm for derivative-free global optimization

This article considers a box-constrained global optimization problem for Lipschitz-continuous functions with an unknown Lipschitz constant. Motivated by the famous DIRECT (DIviding RECTangles), a new HALRECT (HALving RECTangles) algorithm is introduced. A new deterministic approach combines halving...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:Journal of global optimization Ročník 88; číslo 1; s. 139 - 169
Hlavní autoři: Stripinis, Linas, Paulavičius, Remigijus
Médium: Journal Article
Jazyk:angličtina
Vydáno: New York Springer US 01.01.2024
Springer
Témata:
Algorithms
Computer Science
Mathematics
Mathematics and Statistics
Operations Research/Decision Theory
Optimization
Real Functions
DIRECT-type algorithm
Global optimization
Lipschitz optimization
65K05
78M50
90C99
Sampling-based algorithm
65K10
74P99
Derivative-free optimization
ISSN:0925-5001, 1573-2916
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:This article considers a box-constrained global optimization problem for Lipschitz-continuous functions with an unknown Lipschitz constant. Motivated by the famous DIRECT (DIviding RECTangles), a new HALRECT (HALving RECTangles) algorithm is introduced. A new deterministic approach combines halving (bisection) with a new multi-point sampling scheme in contrast to trisection and midpoint sampling used in most existing DIRECT-type algorithms. A new partitioning and sampling scheme uses more comprehensive information on the objective function. Four different strategies for selecting potentially optimal hyper-rectangles are introduced to exploit the objective function’s information effectively. The original algorithm HALRECT and other introduced HALRECT variations (twelve in total) are tested and compared with the other twelve recently introduced DIRECT-type algorithms on 96 box-constrained benchmark functions from DIRECTGOLib v1.1, and 96 perturbed their versions. Extensive experimental results are advantageous compared to state-of-the-art DIRECT-type global optimization. New HALRECT approaches offer high robustness across problems of different degrees of complexity, varying from simple—uni-modal and low dimensional to complex—multi-modal and higher dimensionality.
ISSN:0925-5001
1573-2916
DOI:10.1007/s10898-023-01296-7