A fast learning artificial neural network model (FLANN II) is developed to solve problems in the area of pattern classification for continuous input values. The operational aspects of the neural network are illustrate...
详细信息
A fast learning artificial neural network model (FLANN II) is developed to solve problems in the area of pattern classification for continuous input values. The operational aspects of the neural network are illustrated by two process control diagnostic problems.
When we consider the numerical solution of the 2-dimensional linear hyperbolic problem by implicit difference equations we need to solve a set of linear systems Ax = b with many rightband sides b, where A is large, sp...
详细信息
When we consider the numerical solution of the 2-dimensional linear hyperbolic problem by implicit difference equations we need to solve a set of linear systems Ax = b with many rightband sides b, where A is large, sparse and nonsymmetric. The SUR (Successive Underrelaxation) and GCG (Generalised Conjugate Gradient) methods are used for solving the linear systems. We compare the two methods on sequential and parallel computations. Numerical results indicate that the SUR method is nearly twice as fast as the GCG method and the SUR method has an almost linear speedup.
In this paper, the implementation of the AGE iterative algorithm for the solution of 2-point boundary value problems on a transputer network is discussed,
In this paper, the implementation of the AGE iterative algorithm for the solution of 2-point boundary value problems on a transputer network is discussed,
Some essential techniques for the protection of data structures shared amongst multiple processors are described in this paper. In particular, implementation details for stacks, queues and dequeues are given.
Some essential techniques for the protection of data structures shared amongst multiple processors are described in this paper. In particular, implementation details for stacks, queues and dequeues are given.
The numerical solution of elliptic partial differential equations in non-rectangular regions poses difficulties for the finite difference method because of the need to use irregular mesh points. Here we consider a rho...
详细信息
The numerical solution of elliptic partial differential equations in non-rectangular regions poses difficulties for the finite difference method because of the need to use irregular mesh points. Here we consider a rhombic region and solve the Poisson equation using skew rectangular coordinates by the Successive Block Overrelaxation method. A comparison between 1 line, 2 line and peripheral block methods is given.
An asynchronous scheduling strategy for use in conjunction with any sequential merging algorithm to implement an asynchronous parallel merging algorithm on an MIMD multiprocessor is presented. In particular, an asynch...
详细信息
An asynchronous scheduling strategy for use in conjunction with any sequential merging algorithm to implement an asynchronous parallel merging algorithm on an MIMD multiprocessor is presented. In particular, an asynchronous parallel implementation of the 2-way merge algorithm is described and used as a component in three versions of a parallel sorting algorithm. Their performance is compared when executed on the Sequent Balance 8000 multiprocessor.
In this paper the Galerkin finite element method is applied to the 2 point boundary value problem derived from a 2nd order differential equation and the resulting matrix system solved by the CAGE iterative method a co...
详细信息
In this paper the Galerkin finite element method is applied to the 2 point boundary value problem derived from a 2nd order differential equation and the resulting matrix system solved by the CAGE iterative method a coupled form of the AGE method (Evans, 1985).
The Schwarz Alternating procedure is used in a domain decomposition strategy to study the suitability of overlapping domains when a nonlinear elliptic partial differential equation is solved asynchronously on a multip...
详细信息
The Schwarz Alternating procedure is used in a domain decomposition strategy to study the suitability of overlapping domains when a nonlinear elliptic partial differential equation is solved asynchronously on a multiprocessor with p processors.
In this paper the Alternating Group Explicit (AGE) method is presented for the iterative solution of the finite difference equations derived from the discretisation of an elliptic partial differential equation specifi...
详细信息
In this paper the Alternating Group Explicit (AGE) method is presented for the iterative solution of the finite difference equations derived from the discretisation of an elliptic partial differential equation specified within a two-dimensional domain with given boundary conditions and implemented on a MIMD Balance system. The AGE method is shown to be eminently suitable for parallel computers since it possesses separate independent tasks i.e. (2 x 2) sub-blocks which can be expressed explicitly and can be executed simultaneously. Three parallel strategies are developed and implemented and the results compared.
A new parallel sorting algorithm, Parasort, based upon Quicksort and two-way merging and suitable for implementation on shared memory multiprocessors is presented. An asynchronous parallel partitioning algorithm is us...
详细信息
A new parallel sorting algorithm, Parasort, based upon Quicksort and two-way merging and suitable for implementation on shared memory multiprocessors is presented. An asynchronous parallel partitioning algorithm is used to distribute the work evenly amongst the processors during merging to ensure an even load balance which is crucial if high efficiency is to be achieved. The implementation of this parallel sorting algorithm on the Balance 8000 multiprocessor exhibits good speedup rates approaching linearity when compared to sequential quicksort.
暂无评论