Online makespan minimization in MapReduce-like systems with complex reduce tasks

In the MapReduce processing, since map tasks output key-value pairs, and reduce tasks take the pairs output by the map tasks and compute the final results. Therefore, reduce tasks are unknown until their map tasks are finished. Also, we assume that map tasks are preemptive and parallelizable, but reduce tasks are non-parallelizable. With these assumptions, we study the scheduling of minimizing makespan. Both preemptive and non-preemptive reduce tasks are considered. We prove that no matter if preemption is allowed or not, any algorithm has a competitive ratio at least \(2-\frac{1}{h}\), we then give two optimal algorithms for these two versions.     

On-line supply chain scheduling problems with preemption

We consider supply chain scheduling problems where customers release jobs to a manufacturer that has to process the jobs and deliver them to the customers. The jobs are released on-line, that is, at any time there is no information on the number, release and processing times of future jobs; the processing time of a job becomes known when the job is released. Preemption is allowed. To reduce the total costs, processed jobs are grouped into batches, which are delivered to customers as single shipments; we assume that the cost of delivering a batch does not depend on the number of jobs in the batch. The objective is to minimize the total cost, which is the sum of the total flow time and the total delivery cost. For the single-customer problem, we present an on-line two-competitive algorithm, and show that no other on-line algorithm can have a better competitive ratio. We also consider an extension of the algorithm for the case of m customers, and show that its competitive ratio is not greater than 2m if the delivery costs to different customers are equal.     

两台可拒绝同型机半在线排序问题

本文讨论一个两台可拒绝同型机半在线排序问题.当工件到达时,可以被拒绝,但要付出一定的罚值,也可以被接收加工,消耗一定的加工时间.其目标是要使所有加工工件生成的makespan和被拒绝工件的总罚值之和最小.加工不允许中断.进一步,机器带有两个并行处理子系统,可以提供两种排序方案,最后选取较好的一种.这是第一个在可拒绝同型机排序模型中使用半在线信息,我们设计出一个近似算法,其竞争比为3/2,另外又给出一个√3+1/2≈1.366的下界.     

Competitive analysis of preemptive single-machine scheduling

We give an online algorithm for minimizing the total weighted completion time on a single machine where preemption of jobs is allowed and prove that its competitive ratio is at most 1.57.     

Online scheduling of parallel jobs on two machines is 2-competitive

We consider online scheduling of parallel jobs on parallel machines. For the problem with two machines and the objective of minimizing the makespan, we show that 2 is a tight lower bound on the competitive ratio. For the problem with m machines, we derive lower bounds using an ILP formulation.     

Two-machine open shop scheduling with special transportation times

The paper considers a problem of scheduling n jobs in a two-machine open shop to minimise the makespan, provided that preemption is not allowed and the interstage transportation times are involved. In general, this problem is known to be NP-hard. We present a linear time algorithm that finds an optimal schedule if no transportation time exceeds the smallest of the processing times. We also describe an algorithm that creates a heuristic solution to the problem with job-independent transportation times. Our algorithm provides a worst-case performance ratio of 8/5 if the transportation time of a job depends on the assigned processing route. The ratio reduces to 3/2 if all transportation times are equal.     

Optimal on-line flow time with resource augmentation

We study the problem of scheduling n jobs that arrive over time. We consider a non-preemptive setting on a single machine. The goal is to minimize the total flow time. We use extra resource competitive analysis: an optimal off-line algorithm which schedules jobs on a single machine is compared to a more powerful on-line algorithm that has ? machines. We design an algorithm of competitive ratio , where Δ is the maximum ratio between two job sizes, and provide a lower bound which shows that the algorithm is optimal up to a constant factor for any constant ?. The algorithm works for a hard version of the problem where the sizes of the smallest and the largest jobs are not known in advance, only Δ and n are known. This gives a trade-off between the resource augmentation and the competitive ratio.We also consider scheduling on parallel identical machines. In this case the optimal off-line algorithm has m machines and the on-line algorithm has ?m machines. We give a lower bound for this case. Next, we give lower bounds for algorithms using resource augmentation on the speed. Finally, we consider scheduling with hard deadlines, and scheduling so as to minimize the total completion time.     

Approximability of single machine scheduling with fixed jobs to minimize total completion time

We consider the single machine scheduling problem to minimize total completion time with fixed jobs, precedence constraints and release dates. There are some jobs that are already fixed in the schedule. The remaining jobs are free to be assigned to any free-time intervals on the machine in such a way that they do not overlap with the fixed jobs. Each free job has a release date, and the order of processing the free jobs is restricted by the given precedence constraints. The objective is to minimize the total completion time. This problem is strongly NP-hard. Approximability of this problem is studied in this paper. When the jobs are processed without preemption, we show that the problem has a linear-time n-approximation algorithm, but no pseudopolynomial-time (1 − δ)n-approximation algorithm exists even if all the release dates are zero, for any constant δ > 0, if P ≠ NP, where n is the number of jobs; for the case that the jobs have no precedence constraints and no release dates, we show that the problem has no pseudopolynomial-time (2 − δ)-approximation algorithm, for any constant δ > 0, if P ≠ NP, and for the weighted version, we show that the problem has no polynomial-time 2^q(n)-approximation algorithm and no pseudopolynomial-time q(n)-approximation algorithm, where q(n) is any given polynomial of n. When preemption is allowed, we show that the problem with independent jobs can be solved in O(n log n) time with distinct release dates, but the weighted version is strongly NP-hard even with no release dates; the problems with weighted independent jobs or with jobs under precedence constraints are shown having polynomial-time n-approximation algorithms. We also establish the relationship of the approximability between the fixed job scheduling problem and the bin-packing problem.     

Competitive online scheduling of perfectly malleable jobs with setup times

We study how to efficiently schedule online perfectly malleable parallel jobs with arbitrary arrival times on m ? 2 processors. We take into account both the linear speedup of such jobs and their setup time, i.e., the time to create, dispatch, and destroy multiple processes. Specifically, we define the execution time of a job with length p_j running on k_j processors to be p_j/k_j + (k_j − 1)c, where c > 0 is a constant setup time associated with each processor that is used to parallelize the computation. This formulation accurately models data parallelism in scientific computations and realistically asserts a relationship between job length and the maximum useful degree of parallelism. When the goal is to minimize makespan, we show that the online algorithm that simply assigns k_j so that the execution time of each job is minimized and starts jobs as early as possible has competitive ratio 4(m − 1)/m for even m ? 2 and 4m/(m + 1) for odd m ? 3. This algorithm is much simpler than previous offline algorithms for scheduling malleable jobs that require more than a constant number of passes through the job list.     

Parallel machine covering with limited number of preemptions简

In this paper,we investigate the i-preemptive scheduling on parallel machines to maximize the minimum machine completion time,i.e.,machine covering problem with limited number of preemptions. It is aimed to obtain the worst case ratio of the objective value of the optimal schedule with unlimited preemptions and that of the schedule allowed to be preempted at most i times. For the m identical machines case,we show the worst case ratio is 2m.i.1 m,and we present a polynomial time algorithm which can guarantee the ratio for any 0 ≤ i ≤ m. 1. For the i-preemptive scheduling on two uniform machines case,we only need to consider the cases of i = 0 and i = 1. For both cases,we present two linear time algorithms and obtain the worst case ratios with respect to s,i.e.,the ratio of the speeds of two machines.     

Preemptive multiprocessor order scheduling to minimize total weighted flowtime

Consider m identical machines in parallel, each of which can produce k different product types. There is no setup cost when the machines switch from producing one product type to another. There are n orders each of which requests various quantities of the different product types. All orders are available for processing at time t = 0, and preemption is allowed. Order i has a weight w_i and its completion time is the time when its last requested product type finishes. Our goal is to find a preemptive schedule such that the total weighted completion time ∑_wiCi$\sum w_i{C}_i$ is minimized. We show that this problem is NP-hard even when all jobs have identical weights and there are only two machines. Motivated by the computational complexity of the problem, we propose a simple heuristic and show that it obeys a worst-case bound of 2 − 1/m. Finally, empirical studies show that our heuristic performs very well when compared with a lower bound of the optimal cost.     

Improved algorithm for a generalized on-line scheduling problem on identical machines

This paper considers the problem of on-line scheduling a list of independent jobs in which each job has an arbitrary release time on m parallel identical machines. In this problem, jobs arrive in form of order before its release time and decisions have to be made whenever an order is placed and the orders arrive according to any sequence. A heuristic algorithm, NMLS, better than MLS is given for any m ? 2. The competitive ratio is improved from 2.93920 to 2.78436.     

On-line scheduling of multi-core processor tasks with virtualization

We consider an on-line list scheduling problem of multi-core processor tasks with virtualization to minimize makespan. The competitive ratio of an on-line algorithm is shown for every specific m, where m is the number of processors. Better on-line algorithms are presented for a small number of processors.     

Scheduling with target start times

We address the single-machine problem of scheduling n independent jobs subject to target start times. Target start times are essentially release times that may be violated at a certain cost. The objective is to minimize a bicriteria objective function that is composed of total completion time and maximum promptness, which measures the observance of these target start times. We show that in case of a linear objective function the problem is solvable in O(n⁴) time if preemption is allowed or if total completion time outweighs maximum promptness.     

From Single Agent to Multi-Agent via Hypersequents

In this paper we present a sequent calculus for the multi-agent system S5 _m . First, we introduce a particularly simple alternative Kripke semantics for the system S5 _m . Then, we construct a hypersequent calculus for S5 _m that reflects at the syntactic level this alternative interpretation. We prove that this hypersequent calculus is theoremwise equivalent to the Hilbert-style system S5 _m , that it is contraction-free and cut-free, and finally that it is decidable. All results are proved in a purely syntactic way and the cut-elimination procedure yields an upper bound of ip ₂ (n, 0) where ip ₂ is an hyperexponential function of base 2.     

On the structure of solutions to a class of quasilinear elliptic Neumann problems

We study the structure of positive solutions to the equation ?^mΔ_mu-u^m-1+f(u)=0 with homogeneous Neumann boundary condition. First, we show the existence of a mountain-pass solution and find that as ?→0⁺ the mountain-pass solution develops into a spike-layer solution. Second, we prove that there is an uniform upper bound independent of ? for any positive solution to our problem. We also present a Harnack-type inequality for the positive solutions. Finally, we show that if 1<m?2 holds and ? is sufficiently large, any positive solution must be a constant.     

Construction of systems of sets related to the plurality functions

In the present paper we provide a way of construction of all m-elements consistent systems. These kinds of families of sets are the parameters determining the solutions of some functional equation, which express the consistency condition appearing in characterizing the plurality functions. First, we formulate the idea of extending p-elements family to such m-tuples families which are m-elements consistent systems. Then we study some of their properties and we use them in the constructing of extending p-elements families to m-elements consistent systems and all such systems. Finally, we deal with m-elements consistent systems which satisfy an additional condition.     

Scheduling uniform machines on-line requires nondecreasing speed ratios

We consider the following on-line scheduling problem. We have to schedulen independent jobs, wheren is unknown, onm uniform parallel machines so as to minimize the makespan; preemption is allowed. Each job becomes available at its release date, and this release date is not known beforehand; its processing requirement becomes known at its arrival. We show that if only a finite number of preemptions is allowed, there exists an algorithm that solves the problem if and only ifs _i–1/s_i s_i/s_i+1 for alli = 2,,m – 1, wheres _i denotes theith largest machine speed. We also show that if this condition is satisfied, then O(mn) preemptions are necessary, and we provide an example to show that this bound is tight. © 1998 The Mathematical Programming Society, Inc. Published by Elsevier Science B.V.     

Preemptive Semi-online Algorithms for Parallel Machine Scheduling with Known Total Size

This paper investigates preemptive semi-online scheduling problems on m identical parallel machines, where the total size of all jobs is known in advance. The goal is to minimize the maximum machine completion time or maximize the minimum machine completion time. For the first objective, we present an optimal semi-online algorithm with competitive ratio 1. For the second objective, we show that the competitive ratio of any semi-online algorithm is at least （2m-3）/（m-1） for any m〉2 and present optimal semi-online algorithms for m = 2, 3.     

On the geometry,preemptions and complexity of multiprocessor and shop scheduling

In this paper we study multiprocessor and open shop scheduling problems from several points of view. We explore a tight dependence of the polynomial solvability/intractability on the number of allowed preemptions. For an exhaustive interrelation, we address the geometry of problems by means of a novel graphical representation. We use the so-called preemption and machine-dependency graphs for preemptive multiprocessor and shop scheduling problems, respectively. In a natural manner, we call a scheduling problem acyclic if the corresponding graph is acyclic. There is a substantial interrelation between the structure of these graphs and the complexity of the problems. Acyclic scheduling problems are quite restrictive; at the same time, many of them still remain NP-hard. We believe that an exhaustive study of acyclic scheduling problems can lead to a better understanding and give a better insight of general scheduling problems. We show that not only acyclic but also a special non-acyclic version of periodic job-shop scheduling can be solved in polynomial (linear) time. In that version, the corresponding machine dependency graph is allowed to have a special type of the so-called parti-colored cycles. We show that trivial extensions of this problem become NP-hard. Then we suggest a linear-time algorithm for the acyclic open-shop problem in which at most m−2 preemptions are allowed, where m is the number of machines. This result is also tight, as we show that if we allow one less preemption, then this strongly restricted version of the classical open-shop scheduling problem becomes NP-hard. In general, we show that very simple acyclic shop scheduling problems are NP-hard. As an example, any flow-shop problem with a single job with three operations and the rest of the jobs with a single non-zero length operation is NP-hard. We suggest linear-time approximation algorithm with the worst-case performance of ( , respectively) for acyclic job-shop (open-shop, respectively), where (‖ℳ‖, respectively) is the maximal job length (machine load, respectively). We show that no algorithm for scheduling acyclic job-shop can guarantee a better worst-case performance than . We consider two special cases of the acyclic job-shop with the so-called short jobs and short operations (restricting the maximal job and operation length) and solve them optimally in linear time. We show that scheduling m identical processors with at most m−2 preemptions is NP-hard, whereas a venerable early linear-time algorithm by McNaughton yields m−1 preemptions. Another multiprocessor scheduling problem we consider is that of scheduling m unrelated processors with an additional restriction that the processing time of any job on any machine is no more than the optimal schedule makespan C _max^*. We show that the (2m−3)-preemptive version of this problem is polynomially solvable, whereas the (2m−4)-preemptive version becomes NP-hard. For general unrelated processors, we guarantee near-optimal (2m−3)-preemptive schedules. The makespan of such a schedule is no more than either the corresponding non-preemptive schedule makespan or max {C _max^*,p _max}, where C _max^* is the optimal (preemptive) schedule makespan and p _max is the maximal job processing time. E.V. Shchepin was partially supported by the program “Algebraical and combinatorial methods of mathematical cybernetics” of the Russian Academy of Sciences. N. Vakhania was partially supported by CONACyT grant No. 48433.