Augmented Lagrangian coordination for distributed optimal design in MDO

S. Tosserams, L.F.P. Etman, J.E. Rooda

Onderzoeksoutput: Bijdrage aan tijdschriftTijdschriftartikelAcademicpeer review

56 Citaties (Scopus)

Uittreksel

Quite a number of coordination methods have been proposed for the distributed optimal design of large-scale systems consisting of a number of interacting subsystems. Several coordination methods are known to have numerical convergence difficulties that can be explained theoretically. The methods for which convergence proofs are available have mostly been developed for so called quasi-separable problems (i.e. problems with individual subsystems coupled only through a set of linking variables, not through constraints and/or objectives). In this paper we present a new coordination approach for MDO problems with linking variables as well as coupling objectives and constraints. Two formulation variants are presented offering a large degree of freedom in tailoring the coordination algorithm to the design problem at hand. The first, centralized variant introduces a master problem to coordinate coupling of the subsystems. The second, distributed variant coordinates coupling directly between subsystems. Our coordination approach employs an augmented Lagrangian penalty relaxation in combination with a block coordinate descent method. The proposed coordination algorithms can be shown to converge to Karush-Kuhn-Tucker (KKT) points of the original problem by using existing convergence results. We illustrate the flexibility of the proposed approach by showing that the Analytical Target Cascading method of [18] and theaugmented Lagrangian method for quasi-separable problems of [20] are subclasses of the proposed formulations.
TaalEngels
Pagina's1885-1910
TijdschriftInternational Journal for Numerical Methods in Engineering
Volume73
Nummer van het tijdschrift13
DOI's
StatusGepubliceerd - 2008

Vingerafdruk

Augmented Lagrangians
Augmented Lagrangian
Subsystem
Large scale systems
Linking
Coordinate Descent
Descent Method
Lagrangian Method
Formulation
Large-scale Systems
Optimal design
Convergence Results
Penalty
Degree of freedom
Flexibility
Converge
Target

Citeer dit

@article{cde02b4967404764aefbf5e7d6d97aef,
title = "Augmented Lagrangian coordination for distributed optimal design in MDO",
abstract = "Quite a number of coordination methods have been proposed for the distributed optimal design of large-scale systems consisting of a number of interacting subsystems. Several coordination methods are known to have numerical convergence difficulties that can be explained theoretically. The methods for which convergence proofs are available have mostly been developed for so called quasi-separable problems (i.e. problems with individual subsystems coupled only through a set of linking variables, not through constraints and/or objectives). In this paper we present a new coordination approach for MDO problems with linking variables as well as coupling objectives and constraints. Two formulation variants are presented offering a large degree of freedom in tailoring the coordination algorithm to the design problem at hand. The first, centralized variant introduces a master problem to coordinate coupling of the subsystems. The second, distributed variant coordinates coupling directly between subsystems. Our coordination approach employs an augmented Lagrangian penalty relaxation in combination with a block coordinate descent method. The proposed coordination algorithms can be shown to converge to Karush-Kuhn-Tucker (KKT) points of the original problem by using existing convergence results. We illustrate the flexibility of the proposed approach by showing that the Analytical Target Cascading method of [18] and theaugmented Lagrangian method for quasi-separable problems of [20] are subclasses of the proposed formulations.",
author = "S. Tosserams and L.F.P. Etman and J.E. Rooda",
year = "2008",
doi = "10.1002/nme.2158",
language = "English",
volume = "73",
pages = "1885--1910",
journal = "International Journal for Numerical Methods in Engineering",
issn = "0029-5981",
publisher = "Wiley",
number = "13",

}

Augmented Lagrangian coordination for distributed optimal design in MDO. / Tosserams, S.; Etman, L.F.P.; Rooda, J.E.

In: International Journal for Numerical Methods in Engineering, Vol. 73, Nr. 13, 2008, blz. 1885-1910.

Onderzoeksoutput: Bijdrage aan tijdschriftTijdschriftartikelAcademicpeer review

TY - JOUR

T1 - Augmented Lagrangian coordination for distributed optimal design in MDO

AU - Tosserams,S.

AU - Etman,L.F.P.

AU - Rooda,J.E.

PY - 2008

Y1 - 2008

N2 - Quite a number of coordination methods have been proposed for the distributed optimal design of large-scale systems consisting of a number of interacting subsystems. Several coordination methods are known to have numerical convergence difficulties that can be explained theoretically. The methods for which convergence proofs are available have mostly been developed for so called quasi-separable problems (i.e. problems with individual subsystems coupled only through a set of linking variables, not through constraints and/or objectives). In this paper we present a new coordination approach for MDO problems with linking variables as well as coupling objectives and constraints. Two formulation variants are presented offering a large degree of freedom in tailoring the coordination algorithm to the design problem at hand. The first, centralized variant introduces a master problem to coordinate coupling of the subsystems. The second, distributed variant coordinates coupling directly between subsystems. Our coordination approach employs an augmented Lagrangian penalty relaxation in combination with a block coordinate descent method. The proposed coordination algorithms can be shown to converge to Karush-Kuhn-Tucker (KKT) points of the original problem by using existing convergence results. We illustrate the flexibility of the proposed approach by showing that the Analytical Target Cascading method of [18] and theaugmented Lagrangian method for quasi-separable problems of [20] are subclasses of the proposed formulations.

AB - Quite a number of coordination methods have been proposed for the distributed optimal design of large-scale systems consisting of a number of interacting subsystems. Several coordination methods are known to have numerical convergence difficulties that can be explained theoretically. The methods for which convergence proofs are available have mostly been developed for so called quasi-separable problems (i.e. problems with individual subsystems coupled only through a set of linking variables, not through constraints and/or objectives). In this paper we present a new coordination approach for MDO problems with linking variables as well as coupling objectives and constraints. Two formulation variants are presented offering a large degree of freedom in tailoring the coordination algorithm to the design problem at hand. The first, centralized variant introduces a master problem to coordinate coupling of the subsystems. The second, distributed variant coordinates coupling directly between subsystems. Our coordination approach employs an augmented Lagrangian penalty relaxation in combination with a block coordinate descent method. The proposed coordination algorithms can be shown to converge to Karush-Kuhn-Tucker (KKT) points of the original problem by using existing convergence results. We illustrate the flexibility of the proposed approach by showing that the Analytical Target Cascading method of [18] and theaugmented Lagrangian method for quasi-separable problems of [20] are subclasses of the proposed formulations.

U2 - 10.1002/nme.2158

DO - 10.1002/nme.2158

M3 - Article

VL - 73

SP - 1885

EP - 1910

JO - International Journal for Numerical Methods in Engineering

T2 - International Journal for Numerical Methods in Engineering

JF - International Journal for Numerical Methods in Engineering

SN - 0029-5981

IS - 13

ER -