Abstract:Motivated by issues of saving energy in data centers we define a collection of new problems referred to as "machine activation" problems. The central framework we introduce considers a collection of m machines (unrelated or related) with each machine i having an activation cost of ai. There is also a collection of n jobs that need to be performed, and pi,j is the processing time of job j on machine i. Standard scheduling models assume that the set of machines is fixed and all machines are available. However, i… Show more
“…In addition to its theoretical significance, one of the motivations for this problem comes from the need to minimize energy consumption in large data centers, such as those used by Google and Amazon (see [12,29] for the practical significance of this problem). This problem has been studied earlier in both the offline [19,29] and online [17,27,28] models. While near-optimal algorithms were known in the offline model, we give the first near-optimal online algorithm for this problem.…”
Section: Unrelated Machine Scheduling With Startupmentioning
confidence: 99%
“…The analysis for this algorithm involves proving a bound on the value of the cumulative potential function over all the facilities. We then adapt randomized rounding techniques used for offline machine scheduling with startup costs [29] and online set cover [15] to round the fractional solution and obtain an integral assignment of jobs to machines.…”
Section: Theorem 12mentioning
confidence: 99%
“…The (offline) UMSC problem was first studied in [19,29]. They give offline algorithms for the problem using different techniques.…”
Recent work has shown that the classical framework of solving optimization problems by obtaining a fractional solution to a linear program (LP) and rounding it to an integer solution can be extended to the online setting using primal-dual techniques. The success of this new framework for online optimization can be gauged from the fact that it has led to progress in several longstanding open questions. However, to the best of our knowledge, this framework has previously been applied to LPs containing only packing or only covering constraints, or minor variants of these. We extend this framework in a fundamental way by demonstrating that it can be used to solve mixed packing and covering LPs online, where packing constraints are given offline and covering constraints are received online. The objective is to minimize the maximum multiplicative factor by which any packing constraint is violated, while satisfying the covering constraints. Our results represent the first algorithm that obtains a polylogarithmic competitive ratio for solving mixed LPs online.We then consider two canonical examples of mixed LPs: unrelated machine scheduling with startup costs, and capacity constrained facility location. We use ideas generated from our result for mixed packing and covering to obtain polylogarithmic-competitive algorithms for these problems. We also give lower bounds to show that the competitive ratios of our algorithms are nearly tight.
“…In addition to its theoretical significance, one of the motivations for this problem comes from the need to minimize energy consumption in large data centers, such as those used by Google and Amazon (see [12,29] for the practical significance of this problem). This problem has been studied earlier in both the offline [19,29] and online [17,27,28] models. While near-optimal algorithms were known in the offline model, we give the first near-optimal online algorithm for this problem.…”
Section: Unrelated Machine Scheduling With Startupmentioning
confidence: 99%
“…The analysis for this algorithm involves proving a bound on the value of the cumulative potential function over all the facilities. We then adapt randomized rounding techniques used for offline machine scheduling with startup costs [29] and online set cover [15] to round the fractional solution and obtain an integral assignment of jobs to machines.…”
Section: Theorem 12mentioning
confidence: 99%
“…The (offline) UMSC problem was first studied in [19,29]. They give offline algorithms for the problem using different techniques.…”
Recent work has shown that the classical framework of solving optimization problems by obtaining a fractional solution to a linear program (LP) and rounding it to an integer solution can be extended to the online setting using primal-dual techniques. The success of this new framework for online optimization can be gauged from the fact that it has led to progress in several longstanding open questions. However, to the best of our knowledge, this framework has previously been applied to LPs containing only packing or only covering constraints, or minor variants of these. We extend this framework in a fundamental way by demonstrating that it can be used to solve mixed packing and covering LPs online, where packing constraints are given offline and covering constraints are received online. The objective is to minimize the maximum multiplicative factor by which any packing constraint is violated, while satisfying the covering constraints. Our results represent the first algorithm that obtains a polylogarithmic competitive ratio for solving mixed LPs online.We then consider two canonical examples of mixed LPs: unrelated machine scheduling with startup costs, and capacity constrained facility location. We use ideas generated from our result for mixed packing and covering to obtain polylogarithmic-competitive algorithms for these problems. We also give lower bounds to show that the competitive ratios of our algorithms are nearly tight.
“…al. [22], in which jobs must be assigned to machines, each machine has an associated cost and capacity, and subject to staying within a budget, the goal is to purchase a set of machines and assign jobs to them to minimize the maximum load on any machine. This problem is clearly Set Cover-hard.…”
In this work, we introduce the Cov-MECF framework, a special case of minimum-edge cost flow in which the input graph is bipartite. We observe that several important covering (and multi-covering) problems are captured in this unifying model and introduce a new heuristic LP O for any problem in this framework. The essence of LP O harnesses as an oracle the fractional solution in deciding how to greedily modify the partial solution. We empirically establish that this heuristic returns solutions that are higher in quality than those of Wolsey's algorithm. We also apply the analogs of Leskovec et. al. 's [25] optimization to LP O and introduce a further freezing optimization to both algorithms. We observe that the former optimization generally benefits LP O more than Wolsey's algorithm, and that the additional freezing step often corrects suboptimalities while further reducing the number of subroutine calls. We tested these implementations on randomly generated testbeds, several instances from the Second DIMACS Implementation Challenge and a couple networks modeling realworld dynamics.
“…T'kindt et al (2001) considered the problem to minimize the sum of machine cost plus balancedness defined by the maximum difference among the total processing times on individual machines. Most related work was done by Khuller et al (2010). They considered the offline scheduling problem with machine activation cost as well as with machine cost.…”
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.