FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W2136124799> ?p ?o ?g. }

• W2136124799 abstract "In this dissertation, we investigate a class of multi-level optimization problems, in which discrete variables are present at multiple stages. Such problems arise in many practical settings, and they are notoriously difficult to optimize. Benders decomposition, which is a well-known decomposition method for solving large-scale mixed-integer programming problems, cannot be utilized for the class of problems that we consider due to the existence of discrete variables at lower levels. Cutting plane algorithms such as those proposed by Laporte and Louveaux have been designed for use in bi-level integer programming problems with binary variables at both levels. However, these are based on generic cuts, which do not utilize any problem specific structures, and hence often result in weak convergence. Our goal in this dissertation is to propose novel formulation and solution strategies for several multi-level optimization problems to solve these problems to optimality within practical computational limits. We first consider an edge-partition problem. The motivation for this study is provided by a Synchronous Optical Network (SONET) design application. In the SONET context, each edge represents a demand pair between two client nodes, and the weight of each edge represents the number of communication channels needed between the client nodes. We consider a stochastic version of the problem, in which the edge weights are not deterministic, but their underlying probability distribution is known. The problem is to design a set of SONET “rings” at minimum cost, while ensuring that the resulting network can handle the random demand with a prespecified level of service. We first model the problem as a large-scale integer program, and attempt to solve it via a bi-level decomposition approach, in which both levels contain binary variables. We then propose a three-level solution approach for the problem, which is based on a hybrid integer programming/constraint programming decomposition algorithm. We show computationally that the hybrid algorithm significantly outperforms the other approaches. Next, we focus on a matrix decomposition problem, which arises in Intensity Modulated Radiation Therapy (IMRT) treatment planning. The problem input is a matrix of intensity values that needs to be delivered to a patient, which must be decomposed into a collection of apertures and corresponding intensities. In a feasible decomposition the sum of binary shape matrices multiplied by corresponding intensity values is equal to the original intensity matrix. We consider two variants of the problem: (i) a variant in which the shape matrices used in the decomposition have to satisfy the “consecutive-ones” property, and (ii) a variant in which the shape matrices have to be rectangular. For the first variant, we start by investigating an integer programming model proposed in the literature, and show how the formulation can be strengthened. We then formulate the problem as a bi-level optimization problem that has discrete variables at both stages, and suggest a hybrid integer programming/constraint programming decomposition algorithm similar to our algorithm for the edge-partition problem. Our tests on data obtained from patients show that our algorithm is capable of solving problem instances of clinically relevant dimensions within practical computational limits. We then turn our attention to the second variant of the matrix decomposition problem. We formulate the problem as a mixed-integer program, and investigate a decomposition method for solving it. Unlike the first variant, the second-level problem turns out to be a linear programming problem, and therefore we are able to derive a Benders decomposition algorithm for solving this variant. Finally, we investigate a class of graph search problems. In this class of problems, an intruder is located at an unknown node on the input graph, and a group of searchers needs to be coordinated to detect the intruder within a limited amount of time. This problem arises in settings such as search-and-rescue and search-and-capture operations, and patrol route design. We investigate three variants of the problem: (i) a hide-and-seek version, in which a stationary intruder is hiding at an unknown node, (ii) a pursuit evasion version, in which the intruder can move across the edges of the graph to avoid being detected, and (iii) a patrol problem, in which the searchers are assigned to recurring patrol routes to protect the graph from intrusion. We model each problem as a large-scale integer program, and propose a branch-cut-price algorithm to find the minimum number of searchers needed, and a route for each searcher. In our formulation, both the master problem and the subproblems corresponding to the searchers and the intruder contain binary variables. We model the master problem as a set covering problem and propose a solution approach that is based on dynamic column and row generation." @default.
• W2136124799 created "2016-06-24" @default.
• W2136124799 creator A5006099809 @default.
• W2136124799 creator A5047632649 @default.
• W2136124799 date "2009-01-01" @default.
• W2136124799 modified "2023-09-27" @default.
• W2136124799 title "Algorithms for solving multi-level optimization problems with discrete variables at multiple levels" @default.
• W2136124799 cites W1483563708 @default.
• W2136124799 cites W1501957312 @default.
• W2136124799 cites W1508945008 @default.
• W2136124799 cites W1536408070 @default.
• W2136124799 cites W1544328127 @default.
• W2136124799 cites W1561296992 @default.
• W2136124799 cites W1572663059 @default.
• W2136124799 cites W1692778824 @default.
• W2136124799 cites W1734243449 @default.
• W2136124799 cites W177005447 @default.
• W2136124799 cites W1919364911 @default.
• W2136124799 cites W1931244053 @default.
• W2136124799 cites W1964974282 @default.
• W2136124799 cites W1966876292 @default.
• W2136124799 cites W1968143413 @default.
• W2136124799 cites W1968143987 @default.
• W2136124799 cites W1971232334 @default.
• W2136124799 cites W1972562189 @default.
• W2136124799 cites W1973619523 @default.
• W2136124799 cites W1975387421 @default.
• W2136124799 cites W1975598755 @default.
• W2136124799 cites W1977545325 @default.
• W2136124799 cites W1979396664 @default.
• W2136124799 cites W1982984907 @default.
• W2136124799 cites W1983843138 @default.
• W2136124799 cites W1984280574 @default.
• W2136124799 cites W1989099434 @default.
• W2136124799 cites W1998007106 @default.
• W2136124799 cites W2000005960 @default.
• W2136124799 cites W2001456707 @default.
• W2136124799 cites W2003071413 @default.
• W2136124799 cites W2003662489 @default.
• W2136124799 cites W2004087121 @default.
• W2136124799 cites W2005332233 @default.
• W2136124799 cites W2006510951 @default.
• W2136124799 cites W2011039300 @default.
• W2136124799 cites W2011911122 @default.
• W2136124799 cites W2013405119 @default.
• W2136124799 cites W2014158133 @default.
• W2136124799 cites W2017244914 @default.
• W2136124799 cites W2024644980 @default.
• W2136124799 cites W2031044433 @default.
• W2136124799 cites W2031168329 @default.
• W2136124799 cites W2038195012 @default.
• W2136124799 cites W2040570464 @default.
• W2136124799 cites W2042443501 @default.
• W2136124799 cites W2042642264 @default.
• W2136124799 cites W2046635105 @default.
• W2136124799 cites W2049181072 @default.
• W2136124799 cites W2051461436 @default.
• W2136124799 cites W2053664185 @default.
• W2136124799 cites W2055427220 @default.
• W2136124799 cites W2066474750 @default.
• W2136124799 cites W2067834214 @default.
• W2136124799 cites W2068534561 @default.
• W2136124799 cites W2068956636 @default.
• W2136124799 cites W2070526994 @default.
• W2136124799 cites W2072099942 @default.
• W2136124799 cites W2076322499 @default.
• W2136124799 cites W2082455082 @default.
• W2136124799 cites W2084991719 @default.
• W2136124799 cites W2087205557 @default.
• W2136124799 cites W2088703618 @default.
• W2136124799 cites W2096073429 @default.
• W2136124799 cites W2096800361 @default.
• W2136124799 cites W2097731353 @default.
• W2136124799 cites W2100387313 @default.
• W2136124799 cites W2101186575 @default.
• W2136124799 cites W2104649084 @default.
• W2136124799 cites W2106518318 @default.
• W2136124799 cites W2111165162 @default.
• W2136124799 cites W2114358880 @default.
• W2136124799 cites W2117801253 @default.
• W2136124799 cites W2124947072 @default.
• W2136124799 cites W2126218710 @default.
• W2136124799 cites W2134656724 @default.
• W2136124799 cites W2137766675 @default.
• W2136124799 cites W2139721783 @default.
• W2136124799 cites W2143514150 @default.
• W2136124799 cites W2145157516 @default.
• W2136124799 cites W2146311953 @default.
• W2136124799 cites W2153572980 @default.
• W2136124799 cites W2154788235 @default.
• W2136124799 cites W2155417411 @default.
• W2136124799 cites W2155511526 @default.
• W2136124799 cites W2157771622 @default.
• W2136124799 cites W2159272988 @default.
• W2136124799 cites W2163241124 @default.
• W2136124799 cites W2167580124 @default.
• W2136124799 cites W2167937316 @default.
• W2136124799 cites W2169315729 @default.
• W2136124799 cites W2169427658 @default.
• W2136124799 cites W2183888585 @default.

• next