Results 1 
7 of
7
Pricing commodities, or how to sell when buyers have restricted valuations
 IN 5TH WORKSHOP ON APPROXIMATION AND ONLINE ALGORITHMS
, 2007
"... How should a seller price his goods in a market where each buyer prefers a single good among his desired goods, and will buy the cheapest such good, as long as it is within his budget? We provide efficient algorithms that compute nearoptimal prices for this problem, focusing on a commodity market, ..."
Abstract

Cited by 4 (2 self)
 Add to MetaCart
How should a seller price his goods in a market where each buyer prefers a single good among his desired goods, and will buy the cheapest such good, as long as it is within his budget? We provide efficient algorithms that compute nearoptimal prices for this problem, focusing on a commodity market, where the range of buyer budgets is small. We also show that our technique (which is based on LProunding) easily extends to a different scenario, in which the buyers want to buy all the desired goods, as long as they are within budget.
Randomized rounding for sensor placement problems
, 2007
"... The recent emphasis on homeland security in the U. S. has led to a number of new applications involving sensor placement in physical networks. We will describe some of these sensor placement problems including sensor placement in municipal water networks to minimize health effects from accidental or ..."
Abstract

Cited by 2 (2 self)
 Add to MetaCart
The recent emphasis on homeland security in the U. S. has led to a number of new applications involving sensor placement in physical networks. We will describe some of these sensor placement problems including sensor placement in municipal water networks to minimize health effects from accidental or malicious contamination and sensor placement for intruder detection in transportation networks or buildings. We have addressed these problems using parallel integer programming. In this talk we present a parallelizable heuristic method for finding approximate solutions to general sensor placement problems. An Integer program (IP) is the optimization (maximization or minimization) of a linear function subject to linear constraints and integrality constraints on some or all of the variables. IPs naturally model NPhard combinatorial optimization problems. Thus integer programming is itself NPcomplete, but one can frequently solve instances in practice using branch and bound via commercial or research solvers. The sensor placement problems we consider have n binary decision variables corresponding to the possible sensor locations. The IP must chose at most k sensors (set at most k decision variables to 1). The remainder of the IP sets (integral and/or rational) dependent variables that calculate the objective. Removing the integrality constraints gives the linearprogramming relaxation of an integer program. This is tractible both theoretically and in practice. IP solvers solve this LP relaxation to bound (sub)problems
Pricing commodities
, 2009
"... How should a seller price her goods in a market where each buyer prefers a single good among his desired goods, and will buy the cheapest such good, as long as it is within his budget? We provide efficient algorithms that compute nearoptimal prices for this problem, focusing on a commodity market, ..."
Abstract

Cited by 2 (0 self)
 Add to MetaCart
How should a seller price her goods in a market where each buyer prefers a single good among his desired goods, and will buy the cheapest such good, as long as it is within his budget? We provide efficient algorithms that compute nearoptimal prices for this problem, focusing on a commodity market, where the range of buyer budgets is small. We also show that our LP rounding based technique easily extends to a different scenario, in which the buyers want to buy all the desired goods, as long as they are within budget.
The Network Testbed Mapping Problem
"... Abstract. The Network Testbed Mapping Problem is the problem of mapping an emulated network into a test cluster such as Emulab or DETER. In this paper, we demonstrate that the Network Testbed Mapping Problem is N Pcomplete when there is constrained bandwidth between cluster switches. We demonstrat ..."
Abstract

Cited by 2 (0 self)
 Add to MetaCart
Abstract. The Network Testbed Mapping Problem is the problem of mapping an emulated network into a test cluster such as Emulab or DETER. In this paper, we demonstrate that the Network Testbed Mapping Problem is N Pcomplete when there is constrained bandwidth between cluster switches. We demonstrate that the problem is trivial when bandwidth is unconstrained, and note that a number of new proposals for data center networking have removed this barrier. Finally, we consider new heuristics in the bandwidthlimited case. Key words: network emulation, network embedding, graph partitioning 1
SinglePath Routing of Timevarying Traffic
"... Abstract — We consider the problem of finding a singlepath intradomain routing for timevarying traffic. We characterize the traffic variations by a finite set of traffic profiles with given nonzero fractions of occurrence. Our goal is to optimize the average performance over all of these traffic ..."
Abstract
 Add to MetaCart
Abstract — We consider the problem of finding a singlepath intradomain routing for timevarying traffic. We characterize the traffic variations by a finite set of traffic profiles with given nonzero fractions of occurrence. Our goal is to optimize the average performance over all of these traffic profiles. We solve the optimal multipath version of this problem using linear programming and develop heuristic singlepath solutions using randomized rounding and iterated rounding. We analyze our singlepath heuristic (finding the optimal singlepath routing is NPHard), and prove that the randomized rounding algorithm has a worst case performance bound of O(log(KN) / log(log(KN))) compared to the optimal multipath routing with a high probability, where K is the number of traffic profiles, and N the number of nodes in the network. Further, our simulations show the iterated rounding heuristics perform close to the optimal multipath routing on a wide range of measured ISP topologies, in both the average and the worstcase. Overall, these results are extremely positive since they show that in a widerange of practical situations, it is not necessary to deploy multipath routing; instead, an appropriately computed singlepath routing is sufficient to provide good performance. I.
Business Administration
"... Metaheuristic for Flexibility Design Matching uncertain demand with capacities is notoriously hard. Operations managers can use mixflexible resources to shift excess demands to unused capacities. To find the optimal configuration of a mixflexible production network, a flexibility design problem (F ..."
Abstract
 Add to MetaCart
Metaheuristic for Flexibility Design Matching uncertain demand with capacities is notoriously hard. Operations managers can use mixflexible resources to shift excess demands to unused capacities. To find the optimal configuration of a mixflexible production network, a flexibility design problem (FDP) is solved. Existing literature on FDPs provides qualitative structural insights, but work on solution methods is rare. We contribute the first metaheuristic which integrates these structural insights and is specifically tailored to solve FDPs. Our genetic algorithm is compared to commercial solvers on instances of up to 15 demand types, resources, and 500 demand scenarios. Experimental evidence shows that in the realistic case of flexible optimal configurations, it dominates the comparison methods regarding runtime and solution quality. 1