Both flexible jobshop scheduling and parallel batch processing machine scheduling have been extensively considered;however,the flexible jobshop and parallel batch processing machine scheduling problem(FJPBPMSP)is prev...Both flexible jobshop scheduling and parallel batch processing machine scheduling have been extensively considered;however,the flexible jobshop and parallel batch processing machine scheduling problem(FJPBPMSP)is prevalent in real-life manufacturing processes and is seldom investigated.In this study,FJPBPMSP is examined,where flexible processing and batch processing are performed sequentially.An adaptive imperialist competitive algorithm with cooperation(CAICA)is proposed to minimize makespan and total energy consumption simultaneously.In CAICA,a four-string representation is adopted,and initial empires with novel structures are formed by uniformly dividing the population.An adaptive assimilation and revolution are designed.An adaptive assimilation and revolution are designed.An adaptive imperialist competition with cooperation is provided.Search strategies,imperialists,and colonies are also renewed by new procedures.Computational experiments are conducted on 50 instances.The computational results show that the new strategies of CAICA are effective,and CAICA can provide better results than its comparative algorithms in solving FJPBPMSP.展开更多
In the rescheduling on a single machine, a set of the original jobs has already been scheduled, in order to make a given objective function is optimal. The decision maker needs to insert the new jobs into the existing...In the rescheduling on a single machine, a set of the original jobs has already been scheduled, in order to make a given objective function is optimal. The decision maker needs to insert the new jobs into the existing schedule without excessively disrupting it. A batching machine is a machine that can handle up to some jobs simultaneously. In this paper,we consider the total completion time under a limit on the sequence disruptions for parallel batching based on rescheduling. For the parallel batching problem based on rescheduling, we research the properties of feasible schedules and optimal schedules on the total completion time under a limit on the maximum time disruptions or total time disruptions, in which the jobs are sequenced in SPT order, and give out the pseudo-polynomial time algorithms on the number of jobs and the processing time of jobs by applying the dynamic programming method.展开更多
The online scheduling on an unbounded parallel batch machine with delivery times and limited restarts is studied in this paper.Here,online means that jobs arrive over time and the characteristics of a job become known...The online scheduling on an unbounded parallel batch machine with delivery times and limited restarts is studied in this paper.Here,online means that jobs arrive over time and the characteristics of a job become known until it arrives.Limited restarts mean that once a running batch contains at least one restarted job,it cannot be restarted again.The goal is to minimize the time by which all jobs have been delivered.We consider a restricted model that the delivery time of each job is no more than its processing time.We present a best possible online algorithm with a competitive ratio of 3/2 for the problem.展开更多
This paper considers competitive project scheduling on two unbounded parallel batch machines.There are two competing firms,and each firm has an unbounded parallel batch machine.All projects must be performed in batche...This paper considers competitive project scheduling on two unbounded parallel batch machines.There are two competing firms,and each firm has an unbounded parallel batch machine.All projects must be performed in batches by Firms 1 and 2 on their machines,respectively.The profit that each firm obtains from each project depends on whether the firm finishes the job before or after its competitor.In the first problem,given a feasible schedule for Firm 1,the objective is to find an optimal schedule to maximize the total reward for Firm 2 under the given schedule for Firm 1.The corresponding total reward for Firm 1 is called the worst-case total reward of the given schedule for Firm 1.In the second problem,the objective is to find an optimal schedule for Firm 1 to maximize the worst-case total reward.We provide optimal algorithms for the two problems,respectively.展开更多
Training deep neural networks(DNNs)requires a significant amount of time and resources to obtain acceptable results,which severely limits its deployment in resource-limited platforms.This paper proposes DarkFPGA,a nov...Training deep neural networks(DNNs)requires a significant amount of time and resources to obtain acceptable results,which severely limits its deployment in resource-limited platforms.This paper proposes DarkFPGA,a novel customizable framework to efficiently accelerate the entire DNN training on a single FPGA platform.First,we explore batch-level parallelism to enable efficient FPGA-based DNN training.Second,we devise a novel hardware architecture optimised by a batch-oriented data pattern and tiling techniques to effectively exploit parallelism.Moreover,an analytical model is developed to determine the optimal design parameters for the DarkFPGA accelerator with respect to a specific network specification and FPGA resource constraints.Our results show that the accelerator is able to perform about 10 times faster than CPU training and about a third of the energy consumption than GPU training using 8-bit integers for training VGG-like networks on the CIFAR dataset for the Maxeler MAX5 platform.展开更多
We sttidy the problem of scheduling n jobs on m parallel bounded batch machines to minimize the sum of squared machine loads. Each batch contains at most B jobs, and the processing time of a batch is equal to the long...We sttidy the problem of scheduling n jobs on m parallel bounded batch machines to minimize the sum of squared machine loads. Each batch contains at most B jobs, and the processing time of a batch is equal to the longest processing time of the jobs in this batch. We prove this problem to be NP-hard. Furthermore, we present a polynomial time approximation scheme (PTAS) and a fully polynomial time approximation scheme (FPTAS) for this problem.展开更多
In parallel-batching machine scheduling, all jobs in a batch start and complete at the same time, and the processing time of the batch is the maximum processing time of any job in it. For the unbounded parallel-batchi...In parallel-batching machine scheduling, all jobs in a batch start and complete at the same time, and the processing time of the batch is the maximum processing time of any job in it. For the unbounded parallel-batching machine scheduling problem of minimizing the maximum lateness, denoted 1|p-batch|Lmax, a dynamic programming algorithm with time complexity O(n^2) is well known in the literature.Later, this algorithm is improved to be an O(n log n) algorithm. In this note, we present another O(n log n) algorithm with simplifications on data structure and implementation details.展开更多
摘要Both flexible jobshop scheduling and parallel batch processing machine scheduling have been extensively considered;however,the flexible jobshop and parallel batch processing machine scheduling problem(FJPBPMSP)is prevalent in real-life manufacturing processes and is seldom investigated.In this study,FJPBPMSP is examined,where flexible processing and batch processing are performed sequentially.An adaptive imperialist competitive algorithm with cooperation(CAICA)is proposed to minimize makespan and total energy consumption simultaneously.In CAICA,a four-string representation is adopted,and initial empires with novel structures are formed by uniformly dividing the population.An adaptive assimilation and revolution are designed.An adaptive assimilation and revolution are designed.An adaptive imperialist competition with cooperation is provided.Search strategies,imperialists,and colonies are also renewed by new procedures.Computational experiments are conducted on 50 instances.The computational results show that the new strategies of CAICA are effective,and CAICA can provide better results than its comparative algorithms in solving FJPBPMSP.
基金Supported by the National Natural Science Foundation of China(11271338, 11201121, 71201049) Supported by the National Natural Science Foundation of Henan Province(112300410078) Supported by the Natural Science Foundation of the Education Department of Henan Province(2011B110008)
摘要In the rescheduling on a single machine, a set of the original jobs has already been scheduled, in order to make a given objective function is optimal. The decision maker needs to insert the new jobs into the existing schedule without excessively disrupting it. A batching machine is a machine that can handle up to some jobs simultaneously. In this paper,we consider the total completion time under a limit on the sequence disruptions for parallel batching based on rescheduling. For the parallel batching problem based on rescheduling, we research the properties of feasible schedules and optimal schedules on the total completion time under a limit on the maximum time disruptions or total time disruptions, in which the jobs are sequenced in SPT order, and give out the pseudo-polynomial time algorithms on the number of jobs and the processing time of jobs by applying the dynamic programming method.
基金This research was supported by the National Natural Science Foundation of China(Nos.11701148,11871213 and 11571321)Henan University of Engineering(No.D2016017).
摘要The online scheduling on an unbounded parallel batch machine with delivery times and limited restarts is studied in this paper.Here,online means that jobs arrive over time and the characteristics of a job become known until it arrives.Limited restarts mean that once a running batch contains at least one restarted job,it cannot be restarted again.The goal is to minimize the time by which all jobs have been delivered.We consider a restricted model that the delivery time of each job is no more than its processing time.We present a best possible online algorithm with a competitive ratio of 3/2 for the problem.
基金This research was supported in part by the National Natural Science Foundation of China(Nos.11771406,11571321 and U1504103).
摘要This paper considers competitive project scheduling on two unbounded parallel batch machines.There are two competing firms,and each firm has an unbounded parallel batch machine.All projects must be performed in batches by Firms 1 and 2 on their machines,respectively.The profit that each firm obtains from each project depends on whether the firm finishes the job before or after its competitor.In the first problem,given a feasible schedule for Firm 1,the objective is to find an optimal schedule to maximize the total reward for Firm 2 under the given schedule for Firm 1.The corresponding total reward for Firm 1 is called the worst-case total reward of the given schedule for Firm 1.In the second problem,the objective is to find an optimal schedule for Firm 1 to maximize the worst-case total reward.We provide optimal algorithms for the two problems,respectively.
摘要Training deep neural networks(DNNs)requires a significant amount of time and resources to obtain acceptable results,which severely limits its deployment in resource-limited platforms.This paper proposes DarkFPGA,a novel customizable framework to efficiently accelerate the entire DNN training on a single FPGA platform.First,we explore batch-level parallelism to enable efficient FPGA-based DNN training.Second,we devise a novel hardware architecture optimised by a batch-oriented data pattern and tiling techniques to effectively exploit parallelism.Moreover,an analytical model is developed to determine the optimal design parameters for the DarkFPGA accelerator with respect to a specific network specification and FPGA resource constraints.Our results show that the accelerator is able to perform about 10 times faster than CPU training and about a third of the energy consumption than GPU training using 8-bit integers for training VGG-like networks on the CIFAR dataset for the Maxeler MAX5 platform.
摘要We sttidy the problem of scheduling n jobs on m parallel bounded batch machines to minimize the sum of squared machine loads. Each batch contains at most B jobs, and the processing time of a batch is equal to the longest processing time of the jobs in this batch. We prove this problem to be NP-hard. Furthermore, we present a polynomial time approximation scheme (PTAS) and a fully polynomial time approximation scheme (FPTAS) for this problem.
基金Supported by NSFC(11571323 11201121)+1 种基金NSFSTDOHN(162300410221)NSFEDOHN(2013GGJS-079)
摘要In parallel-batching machine scheduling, all jobs in a batch start and complete at the same time, and the processing time of the batch is the maximum processing time of any job in it. For the unbounded parallel-batching machine scheduling problem of minimizing the maximum lateness, denoted 1|p-batch|Lmax, a dynamic programming algorithm with time complexity O(n^2) is well known in the literature.Later, this algorithm is improved to be an O(n log n) algorithm. In this note, we present another O(n log n) algorithm with simplifications on data structure and implementation details.