John Craig Comfort

The simulation of a master-slave event set processor

Event set manipulation may consume a considerable amount of the computation time spent in performing a discrete-event simulation. One way of minimizing this time is to allow event set processing to proceed in parallel with the remainder of the simulation computation. This paper describes a multiprocessor simulation computer, in which all non-event set processing is performed by the principle processor (called the host). Event set processing is coordinated by a front end processor (the master) and actually performed by several other functionally identical processors (the slaves). A trace-driven simulation pro gram modeling this sytem was constructed, and was run with trace output taken from two different simulation programs. Out put from this simulation suggests that a significant reduction in run time may be realized by this approach. Sensitivity analysis was performed on the significant parameters to the system (number of slave processors, relative processor speeds, and in terprocessor communication times). A comparison between ac tual and simulated run times for a one-processor system was used to assist in the validation of the simulation.

Numerical ellipsometry: enhancement of new algorithm for real-time, in situ film growth monitoring

Abstract The equations associated with the popular models of film deposition of a material on a substrate tend not to be invertible in practical situations. Thus, development of numerical techniques for obtaining approximate solutions to these equations has become necessary. The variably damped least squares (VDLS) algorithm of Levenberg and Marquardt is commonly used, although it requires accurate initial estimates and is typically to slow for real-time applications. Another algorithm combines the speed of an artificial neural network (ANN) for accurate initial estimates, and the real-time refinement capability of VDLS for real-time performance; however, substantial off-line processing (from hours to days) is necessary to train ANNs. A third algorithm (called 2C) has been developed, which is fast enough for reliable real-time solutions without requiring either accurate initial estimates or any off-line processing. The algorithm 2C employs the techniques of dynamic step adjustment, overrelaxation and reduction in dimensions, as well as curve following and curve intersection location techniques developed by the authors. The performance of the new 2C algorithm in terms of speed and convergence is comparable with the ANN-VDLS algorithms for Ni and BK7 glass.

Numerical techniques useful in the practice of ellipsometry

While well-established methods for sample preparation and measurement make ellipsometry attractive for surface and film investigations, development of reliable methods for computing unknown physical attributes of reflecting surfaces is just beginning. This situation has arisen because the equations of ellipsometry are transcendental, have not been inverted and thus require numerical techniques for solution in all but the simplest of reflecting surface configurations. The problem of data analysis is by no means solved and remains one of the key issues in ellipsometry. This paper presents a survey of numerical techniques and principles for approximating solutions to problems, and illustrates an application of these techniques in the development of a new algorithm to address the problem.

Numerical ellipsometry: Real‐time solutions using mapping onto the complex index plane

While in situ ellipsometry measurements can be made on growing films, real‐time solutions for thickness and optical properties remain challenging. Numerical techniques are necessary because for practical material systems, the equations are not generally invertible. In the absorbing film on a known substrate problem, the typical real number unknowns (film n1, k1, and d) outnumber the real parameters (Ψ and Δ) obtained in a single measurement. Although it is known that two intersecting sets of n1, k1, and d values are established from two measurements, in the past these sets have not been well understood. The work here is a thorough investigation of such curves and explores all theoretically possible intersections for metals depositing onto an oxide substrate. Mathematical and material considerations identify the solution intersection in a few milliseconds for real‐time monitoring and control.While in situ ellipsometry measurements can be made on growing films, real‐time solutions for thickness and optical properties remain challenging. Numerical techniques are necessary because for practical material systems, the equations are not generally invertible. In the absorbing film on a known substrate problem, the typical real number unknowns (film n1, k1, and d) outnumber the real parameters (Ψ and Δ) obtained in a single measurement. Although it is known that two intersecting sets of n1, k1, and d values are established from two measurements, in the past these sets have not been well understood. The work here is a thorough investigation of such curves and explores all theoretically possible intersections for metals depositing onto an oxide substrate. Mathematical and material considerations identify the solution intersection in a few milliseconds for real‐time monitoring and control.

Intelligent OCR processing

Optical Character Recognition (OCR) has become a highly demanded information transfer technology in recent years. This demand has been driven by the increasing needs for information sharing and office automation, and by the increasing accessibility to large‐scale, fast, and powerful computer resources. A problem of current OCR technology is that texts produced by the state‐of‐the‐art OCR software contain an unacceptable frequency of errors. This prevents the OCR technology from being efficiently used for vast‐volume information transfer or daily office operation applications. To correct these errors in a conventional way requires a significant amount of costly human‐machine interaction. In this article, we identify and classify the types and distributions of optical recognition errors. We propose a novel post‐processing strategy, based on machine learning techniques, to correct errors resulted from unrecognized or misrecognized characters during the recognition process. By applying this strategy, the accuracy of recognition can be significantly improved, and the human interaction required can be dramatically reduced. Experimental results indicate that, in a typical environment, about 46% of total errors can be corrected automatically (i.e., without human interference), with an accuracy of 91%.

Environment partitioned distributed simulation of queueing systems

An extension of object-oriented design is used to design computer systems to simulate specific queuing networks, and the performance of these systems is estimated through simulation. Substantial speedups were obtained for all three queuing networks. In addition, sensitivity tests were performed on communication and interface attributes of the system. It is noted that environment partitioned simulation appears to provide an effective means of substantially reducing the run time of queuing simulation programs. The simulated results presented for the CR5 and MMc systems are impressive, and should carry over into any queuing simulation composed of the three standard objects used.<<ETX>>

An algorithm for analyzing ellipsometric data taken with multiple angles of incidence

Abstract The equations of ellipsometry are intractable. They have not been inverted, and thus require numerical techniques for solution in all but the simplest of reflecting surface configurations. In prior work, the authors have presented algorithms for creating approximate solutions for these equations in the case of a sequence of measurements being taken at successive times. In this paper, we extend one of these algorithms to the case of readings taken at one time at multiple angles of incidence. The algorithm is described in some detail, and sample results are presented. In addition, statistical means are used to assess the reliability of the values computed.

Parallel simulation: practice: contrasting distributed simulation with parallel replication: a case study of a queuing simulation on a network of transputers

As discrete event simulation programs become larger and more complex, the amount of computing power required for their execution is rapidly increasing. One way to achieve this power is by a employing a multiple processor network to run the simulation programs.Two approaches to the problem of assigning tasks to processors are described-environment partitioning distributed simulation, in which the tasks required to perform a simulation are assigned to processors in the network; and parallel replication, in which copies of the simulation program are assigned to processors, and the results of their execution aggregated. A simulation of an M/M/c queuing system has been implemented on networks of two and three transputers, using each approach. Heidelbergers statistical efficiency and the stabilization time of the system are used as metrics. The parallel replications tended to stabilize faster, but the statistical efficiencies were not significantly different.

Numerical ellipsometry: Applications of a new algorithm for real‐time, insitu film growth monitoring

The equations associated with the popular models of film deposition of a material on a substrate tend not to be invertible in practical situations. Thus development of numerical techniques for obtaining approximate solutions to these equations has become necessary. At present, real‐time, in situ thin film growth ellipsometry data are limited to a single angle of incidence due to the deposition system windows. The 2C algorithm, developed by the authors, represents a new approach to solve this problem. This algorithm is fast enough for reliable real‐time solutions without requiring either accurate initial estimates or any off‐line processing. Algorithm 2C employs the techniques of dynamic step adjustment, overrelaxation, and reduction in dimension, as well as curve following and curve intersection location techniques developed by the authors. The algorithm was originally developed for nickel deposited on BK7 glass, and in this article it is applied to silver, silicon, and gold depositions on BK7 glass. Gold films could be solved readily with no changes, while problems with convergence were encountered with some silver measurements and silicon data required changes in the ‘‘standard’’ initial estimate.The equations associated with the popular models of film deposition of a material on a substrate tend not to be invertible in practical situations. Thus development of numerical techniques for obtaining approximate solutions to these equations has become necessary. At present, real‐time, in situ thin film growth ellipsometry data are limited to a single angle of incidence due to the deposition system windows. The 2C algorithm, developed by the authors, represents a new approach to solve this problem. This algorithm is fast enough for reliable real‐time solutions without requiring either accurate initial estimates or any off‐line processing. Algorithm 2C employs the techniques of dynamic step adjustment, overrelaxation, and reduction in dimension, as well as curve following and curve intersection location techniques developed by the authors. The algorithm was originally developed for nickel deposited on BK7 glass, and in this article it is applied to silver, silicon, and gold depositions on BK7 glass. Gold...

Contrasting Distributed Simulation With Parallel Replication: A Case Study Of A Queuing Simulation On A Network Of Transputers

As discrete event simulation programs become larger and more complex, the amount of computing power required for their execution is rapidly increasing. One way to achieve this power is by a employing a multiple processor network to run the simulation programs. Two approaches to the problem of assigning tasks to processors are described--environment partitioning distributed simulation, in which the tasks required to perform a simulation are assigned to processors in the network; and parallel replication, in which copies of the simulation program are assigned to processors and the results of their execution aggregated. A simulation of an M/M/c queuing system has been implemented on networks of two and three transputers, using each approach. Heidelbergers statistical efficiency and the stabilization time of the system are used as metrics. The parallel replications tended to stabilize faster, but the statistical efficiencies were not significantly different.