Abstract:This paper introduces TLAD(temporal logic based animation description)model and an animation description language SDL/A based on this model. SDL/A haspowerful description functions, and is convenient to express hierarchy abstract specification and the step -wise refinement procedure of designing an animation. This language can be used to describe multiactor action concurrency. It is easy to be integrated into a powerful CASE environment-XYZ system. This paper mainly introduces the design philosophy and the implementation techniques of SDL/A.

Abstract:Parametric design is the core of the current CAD technology. The way tomanage the constraints conveniently and efficiently is required by the procedures to modeland solve the geometric constraint system which is the base of parametric design. A newmethod called as MRA(maximal reduction algorithm) is presented in this paper. It owesto the application about analysis of degree of freedom, graph theory and sparse matrix theory. The MRA is very effective for geometric constraint consistency checking, maximaldecomposition of geometric constraint system and constraint management. It is worth thatthe MRA express the serial, parallel and coupling mechanism of constraint propagationwith reducing tree.

Abstract:IMC (SIMD C) is a SIMD(single instruction multiple data) parallel programming language extended to the syntax of usual C language. SIMC can be easy to program SIMD algorithms. Moreover, it can be used to study SIMD algorithms on differentkinds of SIMD architecture which is defined in SIMC. The simulation environment ofSIMC has been implemented on a sequential computer. This environment can be used toevaluate the performance of SIMD s algorithms and architecture.

Abstract:This paper discusses the basic structure of general forms processing system,suggests a new method-forms registration by anchors in the four corners, and gives the rules to implement this method. It is shown by experiment that forms registration by anchors in the four corners is general, rapid and exact.

Abstract:This paper discusses the synchronization of the inference engine with the constraint solver in the implementation of the BPU -CLP (R) system. Synchronization is needed in backtracking. This method can solve the synchronization problem properly even in the presence of storage optimization in the system.

Abstract:This paper discusses about the problem of software pipelining on complicatedloops. It first introduced a software pipelining method called ruminate method, which canoptimize program with complicated loops. Then it outlined the procedure to realize ruminate method and described the hardware support. The performance of ruminate method isanalyzed at the end of this paper by the aid of the preliminary experimental results.

Abstract:In this paper, a fast new neural network learning algorithm based on multiobjective optimization is presented according to the new research results of multiobjectiveoptimization, cognitive science and neurology. The experiment results have shown thatthis algorithm can solve the problems of long learning time, local minima and difficult network optimization.

Abstract:This paper presents a N-R method of acquiring multi-level reasoning production rules automatically based on NN(neural network), so as to resolve the bottle neckproblem in building a knowledge - based system-knowledge acquisition. This methodadopts a characteristic NN structure model and corresponding learning algorithm, so thatN -R method can fix the number of hidden nodes dynamically and form new conceptswhich are not defined in the sample set in the course of learning. Having been trained, NNis transformed into an RN (reasoning network) with the proposed transform algorithm,from which productionrules can be obtained easily.

Abstract:The concept of (n,k) permutation polynomial over GF(q) is first introduced.The properties of (n,k) permutation polynomials and the relation to k^th order correlationimmune functions have been studied. Sufficient and necessary conditions are proved forsome special n-ary functions to be m^{th(m＜n) order correlation immune and all functions with degree no greater than 2 to be (n－1)thorder correlation immune. The results show that over GF(q)(q＞4) are there nonlinear functions of highest possible correlation immunity order. An efficient method is put forward to construct functions of high nonlinearityfrom those of lower nonlinearity with the same correlation immunity order.}