Two sorting algorithms are described, general_cycle_sort, and special_cycle_sort, based upon the decomposability of a permutationinto a product of cyclic permutations. Under the conditions in which these algorithms ar...
详细信息
Two sorting algorithms are described, general_cycle_sort, and special_cycle_sort, based upon the decomposability of a permutationinto a product of cyclic permutations. Under the conditions in which these algorithms are applicable, the sorting can be accomplishedin linear *** May 1987. revised April 1990.
The goal of Program Verification is to obtain objective evidence that a given program does exactly what it is specified to do. The primary method for several years has been testing, i. e. , executing the program for s...
详细信息
The goal of Program Verification is to obtain objective evidence that a given program does exactly what it is specified to do. The primary method for several years has been testing, i. e. , executing the program for specific input values. However, this method suffers from inconclusiveness about its results; hence current research deals with good criteria for selecting test data.
A multiprocessing model called software pipelining is introduced for which it is easy to guarantee determinancy of results. Design tools for general applications can be used to determine whether a design can cause int...
详细信息
A multiprocessing model called software pipelining is introduced for which it is easy to guarantee determinancy of results. Design tools for general applications can be used to determine whether a design can cause interference and hence, cause loss of data integrity and failure of the determinancy of results. The model is called automatic because the implementor does not need to supply the synchronization mechanism nor identify the points of synchronization. The linker can supply the synchronization mechanisms to prevent interference and yield determinancy results independent of the implementation language. In fact, only minor changes need to be made to existing compilers. No special synchronization statements need to be used and no changes to the implementation language need to be made, independent of which language is used.
This paper deals with the following mechanism for controlling the multiprogramming set in a demand paging system: processes are dynamically divided into several categories according to the number of page faults genera...
详细信息
This paper deals with the following mechanism for controlling the multiprogramming set in a demand paging system: processes are dynamically divided into several categories according to the number of page faults generated during their residence in main memory. A process is admitted into the multiprogramming set only if there is enough space free in the main memory to contain the number of pages corresponding to the current category of the process. Using a queueing network model of an interactive system with such a control mechanism the authors study the effectiveness of the control considered, and, more particularly, its ability to partition the memory space according to the locality of processes.
This paper describes a language in which synchronization behavior for abstract data objects can be specified. Specifications written in this language express synchronization properties independently of other kinds of ...
详细信息
This paper describes a language in which synchronization behavior for abstract data objects can be specified. Specifications written in this language express synchronization properties independently of other kinds of behavior. The modularity of such specifications can be an important aid in constructing reliable software. In addition to a description of the specification language itself, examples of specifications in the language are presented, and the use of these specifications for a variety of purposes is illustrated.
We present a hybrid model predictive control (MPC) approach for optimal operation of an electrical heating system. The system comprises two electrical heat generators, thermal and electrical energy storage systems and...
详细信息
Recently, D. Bertsekas proposed a new method to solve the linear assignment problem that appears to be well suited for concurrent computation. The method, called the auction algorithm is a primal-dual iterative scheme...
详细信息
Recently, D. Bertsekas proposed a new method to solve the linear assignment problem that appears to be well suited for concurrent computation. The method, called the auction algorithm is a primal-dual iterative scheme that updates the dual solution by performing coordinate steps similar to those of the Jacobi and Gauss-Seidel (GS) methods for unconstrained optimization. Bertsekas noted that the GS version tends to converge a little faster on uniprocessor computers but is generally much less parallelizable. Consequently, several questions arise regarding the relative performance of these two versions: How will their performances be affected by changes in problem size, cost range, and problem data? How should one design an efficient implemenation for each version that takes full advantage of novel computer hardware capabilities such as concurrency and vectorization? How does the concurrency of computations affect the sensitivity of the solution times to changes in the cost range for each version? Will Jacobi outperform Gauss-Seidel in a vector-concurrent computational environment? In this paper we address these questions and present insightful algorithmic and empirical comparisons between the two versions. Both versions have been tested on 224 randomly generated dense assignment problems with the number of variables ranging from 10,000 to 16 million and with cost ranges of [0, 10], [0, 100], [0, 1,000], and [0, 10,000]. The vector-concurrent computational environment used in this study is the one available on the Alliant FX/8 at Argonne National Laboratory. In our computational experiments the Jacobi version has exhibited higher speedups, less sensitivity to changes in cost ranges, and better solution times in vector-concurrent mode for cost ranges [0, 1,000] and [0, 10,000]. However, for cost range [0, 100] the GS version can be up to five times faster and solved the one million variable problem in less than one second and the 16 million variable problem, in less tha
Data organization in major-minor loop magnetic bubble memory is described, including an analysis of sequential and random accessing. A typical sorting application is described and analyzed, including information on op...
详细信息
Data organization in major-minor loop magnetic bubble memory is described, including an analysis of sequential and random accessing. A typical sorting application is described and analyzed, including information on optimal algorithms, buffering criteria, and run time. Deviations from this example are presented and analyzed, including highly parallel systems, highly independent systems, and a range of floating buffering schemes.
A technique is presented for the succinct, precise description of solutions to data sharing problems in concurrent software systems. The technique is of use in modelling a software system during the architectural phas...
详细信息
A technique is presented for the succinct, precise description of solutions to data sharing problems in concurrent software systems. The technique is of use in modelling a software system during the architectural phase of its development, when components, component interfaces and component interactions are being defined. The precision of descriptions using the technique offers the opportunity to perform analysis during design and several approaches to analysis afforded by the technique are discussed.
The paper is devoted to testing the reliability of software embedded in military systems. Proposed reliability test criteria are discussed of large-scale software subsystems and approaches are recommended to software ...
详细信息
The paper is devoted to testing the reliability of software embedded in military systems. Proposed reliability test criteria are discussed of large-scale software subsystems and approaches are recommended to software reliability testing.
暂无评论