Refine
Year of publication
- 2019 (13) (remove)
Document Type
- Article (9)
- In Proceedings (2)
- ZIB-Report (2)
SCIP-JACK is a customized, branch-and-cut based solver for Steiner tree and related problems. ug [SCIP-JACK, MPI] extends SCIP-JACK to a massively parallel solver by using the Ubiquity Generator (UG) framework. ug [SCIP-JACK, MPI] was the only solver that could run on a distributed environment at the (latest) 11th DIMACS Challenge in 2014. Furthermore, it could solve three well-known open instances and updated 14 best-known solutions to instances from the benchmark libary STEINLIB. After the DIMACS Challenge, SCIP-JACK has been considerably improved. However, the improvements were not reflected on ug [SCIP- JACK, MPI]. This paper describes an updated version of ug [SCIP-JACK, MPI], especially branching on constrains and a customized racing ramp-up. Furthermore, the different stages of the solution process on a supercomputer are described in detail. We also show the latest results on open instances from the STEINLIB.
Branch-and-bound (B&B) is an algorithmic framework for solving NP-hard combinatorial optimization problems. Although several well-designed software frameworks for parallel B&B have been developed over the last two decades, there is very few literature about successfully solving previously intractable combinatorial optimization problem instances to optimality by using such frameworks.The main reason for this limited impact of parallel solvers is that the algorithmic improvements for specific problem types are significantly greater than performance gains obtained by parallelization in general. Therefore, in order to solve hard problem instances for the first time, one needs to accelerate state-of-the-art algorithm implementations. In this paper, we present a computational study for solving Steiner tree problems and mixed integer semidefinite programs in parallel. These state-of-the-art algorithm implementations are based on SCIP and were parallelized via the ug[SCIP-*,*]-libraries---by adding less than 200 lines of glue code. Despite the ease of their parallelization, these solvers have the potential to solve previously intractable instances. In this paper, we demonstrate the convenience of such a parallelization and present results for previously unsolvable instances from the well-known PUC benchmark set, widely regarded as the most difficult Steiner tree test set in the literature.
PIPS-SBB is a distributed-memory parallel solver with a scalable data distribution paradigm. It is designed to solve MIPs with a dual-block angular structure, which is characteristic of deterministic-equivalent Stochastic Mixed-Integer Programs (SMIPs). In this paper, we present two different parallelizations of Branch & Bound (B&B), implementing both as extensions of PIPS-SBB, thus adding an additional layer of parallelism. In the first of the proposed frameworks, PIPS-PSBB, the coordination and load-balancing of the different optimization workers is done in a decentralized fashion. This new framework is designed to ensure all available cores are processing the most promising parts of the B&B tree. The second, ug[PIPS-SBB,MPI], is a parallel implementation using the Ubiquity Generator (UG), a universal framework for parallelizing B&B tree search that has been successfully applied to other MIP solvers. We show the effects of leveraging multiple levels of parallelism in potentially improving scaling performance beyond thousands of cores.
Branch-and-bound (B&B) is an algorithmic framework for solving NP-hard combinatorial optimization problems. Although several well-designed software frameworks for parallel B&B have been developed over the last two decades, there is very few literature about successfully solving previously intractable combinatorial optimization problem instances to optimality by using such frameworks.The main reason for this limited impact of parallel solvers is that the algorithmic improvements for specific problem types are significantly greater than performance gains obtained by parallelization in general. Therefore, in order to solve hard problem instances for the first time, one needs to accelerate state-of-the-art algorithm implementations. In this paper, we present a computational study for solving Steiner tree problems and mixed integer semidefinite programs in parallel. These state-of-the-art algorithm implementations are based on SCIP and were parallelized via the ug[SCIP-*,*]-libraries---by adding less than 200 lines of glue code. Despite the ease of their parallelization, these solvers have the potential to solve previously intractable instances. In this paper, we demonstrate the convenience of such a parallelization and present results for previously unsolvable instances from the well-known PUC benchmark set, widely regarded as the most difficult Steiner tree test set in the literature.
Empirical studies are fundamental in assessing the effectiveness of implementations of branch-and-bound algorithms. The complexity of such implementations makes empirical study difficult for a wide variety of reasons. Various attempts have been made to develop and codify a set of standard techniques for the assessment of optimization algorithms and their software implementations; however, most previous work has been focused on classical sequential algorithms. Since parallel computation has become increasingly mainstream, it is necessary to re-examine and modernize these practices. In this paper, we propose a framework for assessment based on the notion that resource consumption is at the heart of what we generally refer to as the “effectiveness” of an implementation. The proposed framework carefully distinguishes between an implementation’s baseline efficiency, the efficacy with which it utilizes a fixed allocation of resources, and its scalability, a measure of how the efficiency changes as resources (typically additional computing cores) are added or removed. Efficiency is typically applied to sequential implementations, whereas scalability is applied to parallel implementations. Efficiency and scalability are both important contributors in determining the overall effectiveness of a given parallel implementation, but the goal of improved efficiency is often at odds with the goal of improved scalability. Within the proposed framework, we review the challenges to effective evaluation and discuss the strengths and weaknesses of existing methods of assessment.
To realize the best performance in energy supply, it is an important task to determine types, capacities, and numbers of energy conversion equipment appropriately. It is also necessary to take account of seasonal and hourly variations in energy demands and corresponding operational strategies of equipment. The mixed-integer linear programming (MILP) method has been utilized widely to derive the optimal solution for such a design problem. From the design viewpoint, it is important to generate not only the optimal solution but also suboptimal ones which follow the optimal one without omission, what are called K-best solutions. In this paper, an MILP method utilizing the hierarchical relationship between design and operation variables is applied along with some strategies to efficiently drive K-best solutions of a multiobjective optimal design problem of a gas turbine cogeneration system for district energy supply. The annual total cost and primary energy consumption are adopted as the objective functions to be minimized. A case study is conducted, and K-best solutions with an allowable increase in the value of the combined objective function are derived for each value of the weight for the annual total cost. Through the study, it is clarified how the weight affects the computation time, the number of the K-best solutions, and the design specifications and the values of the objective functions for the K-best solutions.
For the purpose of making a decision in the optimal design of an energy supply system, it is important to investigate not only the optimal solution but also suboptimal ones which follow the optimal one without any omissions, what are called K-best solutions. In this paper, a mixed-integer linear programming method utilizing the hierarchical relationship between design and operation variables proposed previ- ously is extended to search the K-best solutions very efficiently. In addition, methods for updating the incumbents are incorporated into the extended method for three options for the criterion set newly in deriving the K-best solutions. This extended method is implemented into open and commercial MILP solvers, and is applied to illustrative and practical case studies, respectively, on the optimal design of cogeneration systems. Through the studies, it turns out that the proposed method is much superior in terms of solution optimality and computation efficiency to a conventional method, and that the computation efficiency to derive one of the K-best solutions by the proposed method increases with the number of K-best solutions. In addition, features of the K-best solutions in the value of objective function are clarified.
Mixed-integer linear programming (MILP) methods have been applied widely to optimal design of energy supply systems in consideration of multi-period operation. A hierarchical MILP method has been proposed to solve such optimal design problems efficiently. An original problem has been solved by dividing it into a relaxed optimal design problem at the upper level and optimal operation problems which are independent of one another at the lower level. In addition, some strategies have been proposed to enhance the computation efficiency furthermore. In this paper, a method of reducing model by time aggregation is proposed as a novel strategy to search design candidates efficiently in the relaxed optimal design problem at the upper level. In addition, the previous strategies are modified in accordance with the novel strategy. This method is realized only by clustering periods and averaging energy demands for clustered periods, while it guarantees to derive the optimal solution. Thus, it may decrease the computation time at the upper level. Through a case study on the optimal design of a gas turbine cogeneration system, it is clarified how the model reduction is effective to enhance the computation efficiency in comparison and combination with the modified previous strategies.
Mixed-integer linear programming (MILP) methods have been applied widely to optimal design of energy supply systems. A hierarchical MILP method has been proposed to solve such optimal design problems efficiently. In addition, some strategies have been proposed to enhance the computation efficiency furthermore. As one of the strategies, a method of reducing model by time aggregation has been proposed to search design candidates efficiently in the relaxed optimal design problem at the upper level. In this paper, a method of clustering periods has been proposed based on the optimal operational strategies of the systems to avoid a large decrease in the lower bound for the optimal value of the objective function by model reduction. This method has been realized only by solving the relaxed optimal design problem at the upper level in advance. The method can decrease the number of operation variables and constraints at the upper level, and thus can decrease the computation time at the upper level. Through a case study on the optimal design of a gas turbine cogeneration system, it is clarified how the proposed clustering method is effective to enhance the computation efficiency in comparison with the conventional one which clusters periods regularly in time series.
Mixed-integer linear programming (MILP) methods have been applied widely to optimal design of en- ergy supply systems in consideration of multi-period operation. A hierarchical MILP method has been proposed to solve such optimal design problems efficiently. An original problem has been solved by dividing it into a relaxed optimal design problem at the upper level and optimal operation problems which are independent of one another at the lower level. In addition, some strategies have been pro- posed to enhance the computation efficiency furthermore. In this paper, a method of reducing model by time aggregation is proposed as a novel strategy to search design candidates efficiently in the relaxed optimal design problem at the upper level. In addition, the previous strategies are modified in accor- dance with the novel strategy. This method is realized only by clustering periods and averaging energy demands for clustered periods, while it guarantees to derive the optimal solution. Thus, it may decrease the computation time at the upper level. Through a case study on the optimal design of a gas turbine cogeneration system, it is clarified how the model reduction is effective to enhance the computation efficiency in comparison and combination with the modified previous strategies.