This thesis is an investigation for the application of the Frank-Wolfe Method to the Lagrangian Relaxation of the Set Cover Problem, a classical combinatorial optimization problem with several applications. The study goes about with a literature review starting from the foundational papers on the Frank-Wolfe al- gorithm and on Lagrangian Relaxation. The Frank-Wolfe method has numerous applications and is often preferred for it s fast iterations and for being projection free, resulting in lower compu- tational requirements. Lagrangian Relaxation gives a generic framework for computing lower bounds on the optimal value of an optimization problem. The thesis then introduces the Set Cover Problem, its mathematical formu- lation and its practical relevance in real world applications. Finally, the thesis shows how to apply the Frank-Wolfe method to the Lagrangian dual of the set cover problem, yielding an alternative method for computing the linear pro- gramming relaxation of the set cover problem.
Questa tesi e unindagine sullapplicazione del metodo di Frank-Wolfe al rilassa- mento lagrangiano del problema del set cover, un problema classico di ottimiz- zazione combinatoria con molteplici applicazioni. Lo studio prosegue con una revisione della letteratura a partire dai concetti fondamentali di algoritmo di Frank-Wolfe e rilassamento lagrangiano. Il Metodo di Frank-Wolfe ha numerose applicazioni e viene spesso scelto per le sue iterazioni veloci e l’assenza di proiezioni, che risultano in minori requisiti computazionali. Il rilassamento lagrangiano fornisce invece un metodo per il calcolo di approssimazioni inferiori al valore ottimo di un problema di ottimiz- zazione. La tesi prosegue introducendo il problema del set cover, la sua formulazione matematica e la sua rilevanza pratica nel mondo reale. Infine viene mostrato come applicare il metodo di Frank-Wolfe per la risoluzione del lagrangiano duale del problema del set cover, ottenendo un metodo alternativo per risolvere il suo rilassamento lineare.
Frank Wolfe Method for Set Cover Problem with Lagrangian Relaxation
BRAL, ASHWINI
2023/2024
Abstract
This thesis is an investigation for the application of the Frank-Wolfe Method to the Lagrangian Relaxation of the Set Cover Problem, a classical combinatorial optimization problem with several applications. The study goes about with a literature review starting from the foundational papers on the Frank-Wolfe al- gorithm and on Lagrangian Relaxation. The Frank-Wolfe method has numerous applications and is often preferred for it s fast iterations and for being projection free, resulting in lower compu- tational requirements. Lagrangian Relaxation gives a generic framework for computing lower bounds on the optimal value of an optimization problem. The thesis then introduces the Set Cover Problem, its mathematical formu- lation and its practical relevance in real world applications. Finally, the thesis shows how to apply the Frank-Wolfe method to the Lagrangian dual of the set cover problem, yielding an alternative method for computing the linear pro- gramming relaxation of the set cover problem.File | Dimensione | Formato | |
---|---|---|---|
Computer_Engineering_MsC_Thesis__UniPD_24jan2024.pdf
accesso riservato
Dimensione
359.82 kB
Formato
Adobe PDF
|
359.82 kB | Adobe PDF |
The text of this website © Università degli studi di Padova. Full Text are published under a non-exclusive license. Metadata are under a CC0 License
https://hdl.handle.net/20.500.12608/66467