A Lagrangian decomposition approach combined with metaheuristics for the Knapsack Constrained Maximum Spanning Tree problem
Year: 2006
Type: dissertation
Abstract: This master's thesis deals with solving the Knapsack Constrained Maximum Spanning Tree (KCMST) problem, which is a so far less addressed N P-hard combinatorial optimization problem belonging to the area of network design.Thereby sought is a spanning tree whose profit is maximal, but at the same time its total weight must not exceed a specified value.For this purpose a Lagrangian decomposition approach, which is a special variant of Lagrangian relaxation, is applied to derive upper bounds.In the course of the application the problem is split up in two subproblems, which are likewise to be maximized but easier to solve on its own.The subgradient optimization method as well as the Volume Algorithm are deployed to solve the Lagrangian dual problem.To derive according lower bounds, i.e. feasible solutions, a simple Lagrangian heuristic is applied which is strengthened by a problem specific local search.Furthermore an Evolutionary Algorithm is presented which uses a suitable encoding for the solutions and appropriate operators, whereas the latter are able to utilize heuristics based on defined edge-profits.It is shown that simple edge-profits, derived straightforward from the data given by an instance, are of no benefit.Afterwards the Lagrangian and the Evolutionary algorithm are combined to form a hybrid algorithm, within both algorithms exchange as much information as possible.Thereby the two algorithms can be combined either in sequential or intertwined order, whereas this choice depends on the desired requirements.Extensive tests were made on specially created instances of different types.The Lagrangian algorithm turns out to derive the optimal upper bound for all instances except for a few, and to yield, particularly in conjunction with local search, very good and for the most part optimal feasible solutions.Least optimal solutions are derived for maximal plane graphs, where a further improvement is expected by applying the hybrid algorithm.The Lagrangian algorithm is generally superior to previously applied solution methods, especially when considering its run-time.The Hybrid Lagrangian Evolutionary Algorithm can nevertheless partly improve these results.Whereas the sequential variant actually finds more provably optimal solutions for large maximal plane graphs with up to 12000 nodes, the intertwined variant allows to obtain high-quality solutions usually earlier during a run, in particular on complete graphs. DanksagungenIch möchte meinen beiden Betreuern Prof. Günther Raidl und Dr. Jakob Puchinger dafür danken, dass sie mich, wie auch schon in einem Praktikum davor, für dieses Thema begeistert haben, mir immer die nötige Unterstützung gegeben haben und mich vor allem in die faszinierende Welt der kombinatorischen Optimierung eingeführt haben.Ein ganz besonderer Dank gebührt meinen Eltern und Großeltern.Sie haben mich immer in meinem Studium unterstützt und ohne sie wäre nicht ein so sorgloses Weiterkommen möglich gewesen.Nicht zuletzt will ich mich ganz herzlich bei Marlene für ihr Verständnis, ihre Unterstützung und ihre Liebe bedanken.
Source: reposiTUm (TU Wien)
Author: Pirkwieser, Sandro
Language: English
Subfield: Computational Theory and Mathematics
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7029517789
