NUMERICAL STUDIES OF THE COLD FLOW FIELD IN MODEL COMBUSTION CHAMBERS

Numerical analyses of the cold turbulent flow in model combustion chambers were made by using κ-Ε turbulent model. The hybrid difference scheme and SNIP method were employed. Numerical solutions for retouchment length and velocity distributions were obtained in the recirculating zone of the combustion chambers. The calculation results were in fairly good agreement with the reported experimental data. The work presented in this paper was a basic part of the calculation model of sudden-enlarged combustion chambers.

SYNTHESIS OF THE COPOLYMERS OF p-DIETHYNYLBENZENE WITH OTHER ACETYLENIC DERIVATIVE S INITIATED BY (Ph_3P)_2 PdCl_2

The copolymerization of p-diethynylbenzene (PDEB) with phenylacetylene (PhA), 4, 4'-diethynylbiphenyl (DEBP)or m-diethynylbenzene (MDEB) are studied by varying mole ratios of monomers. When the mole ratios of PDEB/PhA are less or equal to 1/5, the copolymers are soluble and fusible, but the other copolymers are insoluble and infusible. The results show that the good solvent of cross-linked copolymers is benzene and their solubility parameter is 9.15 cal^(0.5).cm(-1.5). And their swellability (θ_p), Huggins parameter (χ), density (d_4^(25)) and the average molecular weights between crosslinks (c) are measured. It is found that θp and c of copolymers are greater but d_4^(25) is less than that of respective homopolymers. IR spectra show that the copolymers have transoid configuration and small number of unreacted ethynyl groups exist in the copolymers. The mechanism about the polymerization or acetylenic derivatives initiated by (Ph_3P)_2PdCl_2 is discussed.

MECHANICAL PROPERTIES AND INTERFACIAL STRUCTURES OF ALUMINIUM BORATE WHISKER REINFORCED ALUMINIUM COMPOSITES

Quality 9Al_2O_3-2B_2O_3 whisker reinforced 6061Al or Al composites have been prepared by technique of squeeze casting.The mechanical properties of the 9Al_2O_3·2B_2O_3(w)/6061Al composites could be hardly improved by T6 treatment,owing to cause the remarkable interracial reaction.An observation under high resolution trans- mission electron microscope on interfaces shows that the serious chemical reaction occurs and the product is Al_2MgO_4.But no such reaction is found at the interfaces of 9Al_2O_3. 2B_2O_3(w)/Al composite.EDS analysis on the interfaces shows that above mentioned interracial reaction may be resulted from the interracial segregation of Mg atoms in 6061Al matrix during sequeeze casting.

VISUAUZATION OF THE MATHEMATICAL OPERATION OF CONVOLUTION

E. O. Brigham mentioned in the book The Fast Fourier Transform that it is extremely difficnlt to visualize the mathematical [operation of convolutional integration. He stated a general rule for determining the convolutioual integration limits, but it is yet difficult to master and use. In this letter a pattern convenient to visualizing the mathematical operation is proposed. Let f1（t） and f2（t） be functions, which can bo written as single analytic expressions in their durations [l2,r1] and [l2, r2] respectively, l1≠∞, l2≠∞,

A New Integrated System of Logic Programming and Relational Database

Based on the study of the current two methods—interpretation and compilation—for the integration of logic programming and relational database,a new precompilation-based interpretive approach is proposed.It inherits the advantages of both methods,but overcomes the drawbacks of theirs.A new integrated system based on this approach is presented,which has been implemented on Micro VAX Ⅱ and applied to practise as the kernel of the GKBMS knowledge base management system.Also discussed are the key implementation techniques,including the coupling of logic and relational database systems,the compound of logic and relational database languages,the partial evaluation and static optimization of user's programs,fact scheduling and version management in problem-solving.

PSOF: A PROCESS MODEL BASED ON OR-FOREST DESCRIPTION

This paper presents a process model, based on the OR-forest description, for the parallel execution of logic programs. In the PSOF model, the parallel execution of a program is realized by the parallel search of the unique OR-forest for the program: an OR-parallel execution of a goal is carried out by successor processes simultaneously searching multiple branches, and all the processes searching the same tree are logically independent and need not communicating; AND-parallel execution of a goal is practised by slave processes simultaneously searching multiple trees. A slave process sends a message to its master process only when it reaches a leaf node. The subject of the automatic partition of sub-goals in the framework of OR-forest is also discussed in this paper.

THE SHARING OF ENVIRONMENT IN AND- & OR-PARALLEL EXECUTIONS OF LOGIC PROGRAMS

An environment-sharing scheme for both AND- and OR-parallel executions o?logic programs is presented in this paper. In this scheme, an environment consists of binding records for storing binding values o?variables. A binding record contains only the values of instantiated variables. Unbound variables never occur in binding records. Therefore, binding records of a process can be safely shared by its successor/slave processes, overcoming the drawback of checking and copying "uncommitted context" in OR-parallel environment-sharing scheme.A 2-level storage structure for environment storage and procedures for creating, accessing and updating the environment are defined in this paper. This scheme, including all the procedures, has been implemented in PROLOG and tested by running a number of benchmarks.

DESIGNING THE ARCHITECTURE OF A PROLOG MACHINE

In order to implement Prolog efficiently, Prolog-oriented processors and VLSI chips have been proposed and designed. However, most of the Prolog machines and processors emerged recently suffer from two weaknesses: they cannot implement Prolog database operations or cannot perform numerical evaluations at high speed. In this psper, the architecture of a Prolog machine GKD-PLM is presented. Based on the WAM-PLUS which is an execution model for sequential Prolog, the GKD-PLM can not only execute Prolog code rapidly, but also support database operations and numerical evaluations. First, the characteristics of the machine are described, followed by the description of the instruction set, the data representations, and the hardware structures of the execution unit and the microprogram control unit of the Prolog processor SPP. The performance of the machine is estimated.

PERFORMANCE ANALYSIS OF STATIC DATA FLOW COMPUTERS

First, a model of static data flow computer and a model of data flow graph are pro-posed; then a model of system is presented to calculate practical parallelism degree withoverhead of instruction execution on data flow computers as its parameter. From the compu-tation, the maximum practical parallelism degree of a program running on a static dataflow computer is determined with MP/OH (MP is the mean parallelism degree of a program,OH is the overhead of instruction execution on the computer). Therefore the overhead hasgreat influence on the performance of a data flow computer.

Parallel Execution of Prolog on Shared-Memory Multiprocessors

Logic programs offer many opportunities for the exploitation of parallelism.But the parallel execution of a task incurs various overheads This paper focuses on the issues relevant to parallelizing Prolog on shared-memory multiprocessors efficiently.

Detection of And-Parallelism in Logic Programs

In this paper, we present a detection technique of and-parallelism in logic programs. The detection consists of three phases: analysis of entry modes, derivation of exit modes and determination of execution graph expressions. Compared with other techniques, our approach, with the compile-time program-level data-dependence analysis of logic programs, can efficiently exploit and-parallelism in logic programs. Two precompilers, based on our technique and DeGroot' s approach respectively, have been implemented in SES-PIM system. Through compiling and running some typical benchmarks in SES-PIM, we conclude that our technique can, in most cases, exploit as much and-parallelism as the dynamic approach does under 'producer-consumer' scheme, and needs less dynamic overhead while exploiting more and- parallelism than DeGroot's approach does.

A Goal-Type Driven Method of Solving Horn Logic with Equality

A new method of solving Horn logic with equality,the goal-type driven method,is presented, which considers explicitly the unification operator as a goal and merged it into the resolution process. The method has the following advantages.The resolution and the unification have been integrated in a uniform way.The architectures of the inference engines based on Horn logic with equality are simplified. Any techniques of exploiting AND/ OR parallelism to solve goals can also be applied to unification at the same time.The method can be used to integrate the styles of functional language and logic lan- guage by a uniform framework.It can also deal with infinite data structures.

DISCRETE HIERARCHY AND ORIGINALSHAPE VECTORIZATION

In Ref. [1] it has been pointed out that in the course of tracing an A0-type loop in accordance with serial arithmetic semantics, the occurrences of elements constitute a rank. For any given A0-type loop, Theorem 1 proves that the occurrence rank of its elements is to be the only determined, and Theorem 2

SCKE:Combining Logic- with Object-Oriented Paradigm

A new implementation(SCKE—Stractured Communication Knowledge Entity)has been proposed towards combining the logic-with the object-oriented paradigm of computing.It is intended to explore the advantages in these two paradigms in a structured,natural and efficient manner for large-scale know- ledge processing.The SCKE model supports modularity and protection for the structured development of knowledge systems.It also introduces the concepts that are typical for the object-oriented systems in the logic-oriented paradigm,without losing its advantages as a declarative language.Various inheritance hier- archies are supported in the SCKE model.They provide the semantics basis for various knowledge in AI systems.The M-entity/K-entity/Instance inheritance captures the relationship among the control, procedural and factual knowledge in AI systems,And,the super-entity/entity/instance inheritance shows the concepts of data abstraction in the knowledge of a particular domain.In addition,the SCKE model is not simply supported on top of Prolog like other attempts to integrate the object-into the log- ic-oriented paradigm.The SCKE model is a tighltly coupled model of the logic-and the object-oriented paradigm and its interpreter uniformly interprets the logic semantics and the object-oriented semantics.

Optimizing FORTRAN Programs for Hierarchical Memory Parallel Processing Systems

Parallel loops account for the greatest amount of parallelism in numerical programs.Executing nested loops in parallel with low run-time overhead is thus very important for achieving high perform- ance in parallel processing systems.However,in parallel processing systems with caches or local memo- ries in memory hierarchies,“thrashing problem”may arise whenever data move back and forth between the caches or local memories in different processors. Previous techniques can only deal with the rather simple cases with one linear function in the perfect- ly nested loop.In this paper,we present a parallel program optimizing technique called hybrid loop inter- change(HLI)for the cases with multiple linear functions and loop carried data dependences in the nested loop.With HLI we can easily eliminate or reduce the thrashing phenomena without reducing the program parallelism.

Loop Staggering,Loop Compacting:Restructuring Techniques for Thrashing Problem

Parallel loops account for the greatest amount of parallelism in numerical programs.Executing nested loops in parallel with low run-time overhead is thus very important for achieving high performance in parallel processing systems.However,in parallel processing systems with caches or local memories in memory hierarchies,“thrashing problem”may arise when data move back and forth frequently between the caches or local memories in different processors.The techniques associated with parallel compiler to solve the problem are not completely developed.In this paper,we present two restructuring techniques called loop staggering,loop staggering and compacting,with which we can not only eliminate the cache or local memory thrashing phenomena significantly,but also exploit the potential parallelism existing in outer serial loop.Loop staggering benefits the dynamic loop scheduling strategies,whereas loop staggering and compacting is good for static loop scheduling strategies,Our method especially benefits parallel programs,in which a parallel loop is enclosed by a serial loop and array elements are repeatedly used in the different iterations of the parallel loop.