Search

This paper presents a technique for simulation of the stock preparation and Fourdrinier dynamics which is flexible, convenient, and detailed enough to allow the user to study physical process modifications and transient response dur ing upsets and "furnish" (fibers with different character istics, fillers, dyes, etc.) changes and to evaluate simplified mathematical models and control schemes. The technique uses digital simulation in a block diagram form, each block of which corresponds to one type of physical unit in the process (e.g., cleaner or tanks). Simu lation of a particular process then involves specification of the particular parameters of each physical unit together with the way in which they are interconnected. Individual components of the furnish may be traced through the whole process. A complete wet-end configuration, including the white water recovery system, is simulated using the technique described in the paper, and typical results obtained during grade changes and disturbances are presented.

In the early 1950's at M.I.T. the Servomechanisms Laboratory (now the Electronic Systems Laboratory) devised and developed the first automatically controlled milling machine. The controlling information for the machine was introduced in the form of punched paper tape, on which all dimensional information and instructions for the various feeds and cutter speeds was contained. At first the punched paper tape was prepared manually by some human operator who translated, in effect, the detail drawing of the part to be machined into numerical form and then into appropriate patterns of holes in the tape. This was a tedious and entirely mechanical chore, and it was only natural that short cuts in the process began to suggest themselves. The scope of such short cuts began to spread through the fabric of the technique, and it was not long before the computer was involved in implementing them.

This paper presents a tutorial analysis of a hybrid computer model of a missile reentering the atmosphere. An important aspect of the paper is that an example of how to go about analyzing the suitability of a hybrid computer configuration for a particular large- scale simulation is presented from an overall systems point of view. Equations describing a missile (six- degrees of freedom) reentering the atmosphere are given, as is a hybrid computer configuration for solving them. This configuration is then analyzed to establish its suitability for simulation.

This paper presents an improved solution procedure for the warehouse location problem where continuous economies of scale are present. The improvement is achieved by an effective combination of two heuristic methods presented recently by Drysdale and Sandiford, and Khumawala. The solution procedure first focuses on reducing the potential warehouse set using Khumawala’s method. An iterative technique is then used to optimally allocate customers to these warehouses, taking into account their diminishing marginal costs. As demonstrated in this paper, the procedure converges rapidly to a “good” solution. The efficiency of the procedure is compared with another solution method and the results are quite promising.

The ideas presented in this paper have evolved gradually over the past 10 to 15 years. Their genesis may be ob served in the three reports in the bibliography, covering work the author contributed to and directed. The approach has been used in many practical problems, including the simulation of guided missile systems at the large analog computer facility at Wright-Patterson Air Force Base, Ohio. The ideas have further evolved during the teaching of courses in numerical analysis at the University of Miami, a two week session at the EAI Princeton Computation Center in 1964, and extensive use of the Milgo Computa tion Center in Miami. The paper itself was prepared specifically for the meet ing at which it was presented. It was couched in general phase-space terminology, but the examples were all two dimensional. The reviewer kindly suggested that either the discussion be directed toward the phase plane, or higher dimensional examples be included. Following his sug gestion, higher dimensional examples have been solved which will be presented as a sequel, to avoid making this paper too long. In addition, one of the author's graduate students, Mr. Alex Koler, has been producing some inter esting results on the global topology of the phase plane of quadratic systems, which will be presented separately. The field appears fruitful for further investigations.

In these four pages, this paper can be neither a complete introduction nor a complete survey of the role of computer graphics in simulation. However, these examples drawn from the experience of the Naval Post graduate School computer laboratory will illustrate a wide class of applications which may suggest uses in other fields. The interested reader should refer to the January 1971 issue of SIMULATION (which was devoted entirely to computer graphics applications in simulation) and to the supplementary reading list at the end of this paper.

Handwriting Recognition Technology has been improving much under the purview of pattern recognition and image processing since a few decades. This paper focuses on the comprehensive survey on on-line handwriting recognition system along with the real application by taking Nepali natural handwriting (a real example of one of the cursive handwritings). The survey mainly includes pre-processing, feature vector and similarity measures in between the non-linear 2D sequences of coordinates, and their effective applications. A very highlighting topic "Dynamic Time Warping Algorithm'' (DTW) is introduced, which has been popular in determining the distance between two non-linear sequences ranging from handwriting to speech recognition. Besides these contemporary research issues/areas, stroke number and order free Nepalese natural handwritten recognition system is presented in the second step. Writing one's own style brings unevenness in writing units, which is the most difficult part to classify. Writing units reveal number, shape, size, order of stroke, and speed in writing. Variation in the number of strokes, their order, shapes and sizes, tilting angles and similarities among characters from one another are the important factors, which are to be considered in classification for Nepali. This paper utilizes structural properties of those alphanumeric characters, which have variable writing units. It uses a string of pen tip's positions and tangent angles of every consecutive point as a feature vector sequence of a stroke. We constructed a prototype recognizer that uses the DTW algorithm to align handwritten strokes with stored strokes' templates and determine their similarity. Separate system is trained for original and preprocessed writing samples and achieved recognition rates of 85.87% and 88.59% respectively. This introduces novel real time handwriting recognition on Nepalese alphanumeric characters, which are independent of number of strokes, as well as their order. Key Words: Handwriting Recognition System; Pre-processing; Feature Vector; Dynamic Time Warping; Agglomerating Hierarchical Clustering; Nepali. DOI: 10.3126/kuset.v5i1.2845 Kathmandu University Journal of Science, Engineering and Technology Vol.5, No.1, January 2009, pp 31-55

An important part of the work of an actuary consists in making and testing graduations of mortality statistics. Much space has been devoted in the Journal to methods of graduation, but for many years, apart from the recent paper by H. L. Seal (J.I.A. Vol. lxxi, p. 5), little attention seems to have been paid to the testing of graduations. Although statistical methods are usually employed in making the tests, consideration has not often been given to ascertaining whether mortality statistics fulfil the conditions necessary for the tests to be strictly applicable. It is the purpose of this paper to set out some investigations into this question. Such matters may be thought to be of only theoretical interest, since, in practice, the tests work satisfactorily as a general rule; but actuaries should not employ their tools in a mechanical fashion without realizing fully their limitations and implications, and the question therefore takes on a practical as well as a theoretical aspect. It is hoped that this paper will show how far the tests of a graduation are satisfactory in theory and practice and indicate where modifications are required.

This paper develops a simulation model which can be used to investigate a wide variety of stock market invest ment strategies. A brief review of the literature of stock market forecasting is given. The paper describes the de tails that any simulation of a stock market investor would have to include if the model is to be realistically com pared to the performance of real investors. An outline of the necessary features of any program which is to be used to investigate may different combinations of invest ment strategies and forecasting devices is also given. The program is described in detail and a few preliminary results are given.

GASP IV is a FORTRAN-based simulation language which provides the framework for modeling systems involving both continuous and discrete phenomena. A general ization of the definition of "event" allows for time to be advanced in a next-event fashion while permit ting a step-wise evaluation of system state variables described by difference or differential equations. Subprograms are included in GASP IV to handle the details of state and event control (including state- variable integration when necessary), information storage and retrieval, collection and analysis of data on system performance, and generation of reports and plots. In this paper, the GASP IV philosophy and modeling approach are described. Descriptions of the subpro grams included in GASP IV and the required user- written subprograms are given. The types of applica tions that have utilized GASP IV are listed. A com panion paper (beginning on p. 71) presents a detailed example of the use of GASP IV for simulating a continuous reaction process involving discrete startups and shutdowns.

This paper describes a batch-mode FORTRAN-based continuous-system simulation language possessing most of the desirable features specified by the Simulation Council's CSSL committee. Of course, there are already many continuous-system simulation languages, such as DSL/90 and CSMP (block- oriented), MIMIC, SL-1, CSSL III, etc.; however, to the writers' knowledge, only MIMIC and CSSL III are currently available for CDC-6000 series systems. MIMIC, while equation-oriented, is not as powerful as CSSL III, is somewhat rigid in format, and hard to modify. CSSL III usage involves a surcharge to the program originator, and is thus not attractive for educational applications. DARE IIIB, locally developed under NSF sponsorship, is a modern CSSL-type language, almost as powerful as CSSL III, that is available to any potential user for the cost of a magnetic tape. DARE IIIB is structured to permit convenient user modification, e.g., it is relatively easy to add new integration rules, run-time strategies, or special library subprograms. Problem equations are entered in a form close to ordinary mathematical notation; the associated proce dural language is FORTRAN. User-chosen variable names in the problem description are accommodated by a translation phase. Run-time data is entered via standard FORTRAN IV namelist data cards. A variety of output options are provided, including line- printer listings and plots, and CalComp plotter dis plays. Single- or multiple-run problems may be specified, with storage files used for saving output data for subsequent cross-plotting, reentry into future runs, etc. In addition to an internal collection of library functions, special functions may be specified in FORTRAN, and one-or-two variable table lookup function generation is provided. DARE IIIB provides a choice of several internal fixed- and variable-step integra tion rules; the user can insert his own run-time con trol subroutine for special application. The system is presently implemented on a CDC 6400 with 65K of core memory (SCOPE 3.2 operating system), and an off-line CalComp plotter. DARE IIIB can accommodate problems with 100 parameters, 200 state variables, and 350 output variables. This paper describes the major features of the lan guage, and presents basic user information, and several examples. Inquiries regarding this software package should be addressed to Professor John V. Wait.

In the last couple of decades, a logician or a philosopher has run a risk whenever he has put the term “information” into the title of one of his papers. In these days, the term “information” often creates an expectation that the paper has something to do with that impressive body of results in communication theory which was first known as theory of transmission of information but which now is elliptically called information theory (in the United States at least).1 For the purposes of this paper, I shall speak of it as statistical information theory. I want to begin by making it clear that I have nothing to contribute to this statistical information theory as it is usually developed.

This paper presents a technique for selecting an optimum alternative from among a limited set of possible alternatives. The method presented is an algorithmic successive-elimination procedure which uses simulation to measure the "cost" of each alter native. The algorithm requires the calculation of estimates of confidence intervals of the mean "cost" for each alternative after each replication of the simulation. By comparing the confidence limits and "costs" for each alternative after each replication, one or more alternatives may be elim inated from consideration. The procedure ends when only one alternative remains or when the criteria used indicate that further analysis would be uneconomical.

This paper studies the effects of using correlated random-number streams in replications of a simula tion experiment in which output is to be related to input by linear or quadratic response surfaces. For the linear case, it is shown that either positively or negatively correlated streams produce a smaller generalized variance for the least-squares estimates than occur when the streams are independent. Al though the prediction variance is small with nega tively correlated streams, positively correlated streams produce a smaller variance for the estimated slope. In the quadratic case it is shown that of twenty- seven possible input correlation designs only six guarantee a smaller generalized variance. Only one of these produces a smaller prediction variance. However, no plan will guarantee a smaller slope variance. In the quadratic case one often wants the slope variance to be as small as possible since the objective of the study may be to determine where the maximum or minimum response occurs in terms of the input.

SMOOTH is a packaged network simulation language written in FORTRAN. It is based on the GASP TV and GERT IIIZ simulation languages and is capable of representing and analyzing models which include both continuous variables and discrete events. As a net work language, SMOOTH consists of a set of graphical symbols, users' input instructions, and a computer program. The basic symbols are nodes and branches: nodes perform specified functions on system varia bzes, while branches indicate precedence or function aZ relations between nodes. In using SMOOTH, the user models the system he wishes to study as a pair of interrelated networks and then describes the net works to the computer by means of the SMOOTH lan guage's input instructions. The SMOOTH computer pro gram analyzes the netzoork using simulation. No com puter programming or sophisticated mathematical tech nique is required on the part of the user. This paper discusses the concepts and symbols of SMOOTH and includes as an example a SMOOTH model of a lake ecosystem. SMOOTH has been used to study biological systems, environmental pollution by trace metals and insecticides, and problems in systems dy namics. It is the first network simulation language capable of analyzing models of a combined continuous discrete nature.

This report describes and illustrates a new nonlinear single-step implicit method for the numerical integra tion of general differential systems. The method is A-stable, operates with fixed or variable step size, is computationally fast, and has accuracy comparable to or better than that of other widely used methods. This paper reports computational experiments using this new method and compares the results with those obtained using other integration algorithms, includ ing high-order methods. The comparisons are generally favorable to the new method. The method is applicable to both "stiff" and "nonstiff" systems. On stiff systems the new method was much more accurate than any of the other algorithms tested.

This paper analyzes the course of French society in terms of the time path of a set of variables: exoge nous politico-military pressure, economic growth, unemployment, price rise, population growth, ethnic tension, public unrest, administrative effectiveness, expectations of socii, and an overall measure of societal viability termed Government Stability (Z or GS). The variables reflect the performance of interacting societal institutions that together constitute a complex adaptive total-society system. Government operates as the regulatory subsystem of society. The values of the variables during a given period represent the state of the system. They are ob tained through a simulation algorithm which simu- Zates a societal flow graph representing the pattern of internal system dynamics. The extension of simulation to past and future periods permits retrodictive confirmation and predictive inference. The correspondence and consistency of the generated results with the reported and available data makes possible the validation of the analysis. The simu lation algorithm, which is applicable to all socie tal systems, is here applied to chart the course of French society for the two decades from 1955 to 1973. Values of economic growth, price rise, unemployment, and change of government generated by simulation are directly confirmable from known empirical facts. The confirmation of qualitative variables like ethnic tension, administrative effectiveness, and expectations of socii as gener ated by the model is, however, based on inference. They conform to the broad empirical pattern as outlined in the secondary sources of data.

When queueing systems can be modeled in the form of a system of simultaneous first-order difference- differential equations and when the transient-state of such a system is likely to defy analysis, it may be wiser, less expensive, less troublesome, and faster to use some numerical method. The outstanding advantage of numerical methods is that it is rela tively easy to estimate the error (due to truncation of higher-order terms) in the results. The argument for seeking a numerical analysis method for solving a wide class of these models has a strong practical appeal. This paper explores the transient solutions to nonhomogeneous Markov processes (those in which the parameters vary with time), using Runge-Kutta and Hamming methods. For illustration, a simple but rather instructive, single queueing model is solved manually. The solutions obtained by these methods are compared with the corresponding analytical solu tions. Several commonly-used queueing models are solved to demonstrate the wide potential applicabil ity of the numerical methods.

In this paper a review of the approaches to modelling adopted by programming languages currently available for discrete-system simulation leads to a close look at some of the advantages of languages which employ an "activity scan." These advantages include faster run-times for highly interrelated systems, simpler event routines, and increased security for the model. Decision tables provide a clear and concise format for specifying a complex set of conditions and the various consequent courses of action. They are therefore ideal for describing the conditions for interaction between component parts of a model as specified in the user-defined event routines. A preprocessor to make decision tables computer- readable would greatly enhance the process of modeZ ling and would allow a considerable extension of the range of conditioned statements to be used in the condition stubs of the table. A decision-table facility would form a useful extension to the many event-oriented simulation programming languages.

This paper outlines the software for an open-water version of CLEAN, an ecosystem model for lakes. The modular program structure is described, examples of driving variables are given, and the output (including values of several parameters) is shown. The program, with extensive annotations to facilitate adaptation by potential users, is available from author Richard A. Park or from SCS.

This paper presents an optimal algorithm for scheduling a project network when the tasks are operating in a resource pool which has a finite upper bound on each resource type. A state space description of task processing time is used in the dynamic programming formulation. A finite time processing interval and a penalty cost rate ia assigned to each task to cost the system in the optimal search.

Computer simulation of cognitive processes requires con structing an explicit model of human thinking behavior in the form of a computer program. Such a program cannot avoid being explicit and unambiguous. Thus it serves as a hypothesis to be tested against the corresponding human behavior. Models that don't "work" suggest questions to ask. Recent work in simulating cognition suggests that the basic elements in cognitive processes are attention, conscious ness, and memory. In this paper, a hypothesis is presented describing probable interactions among these elements for a simple cognitive process. To facilitate simulation a paradigm is presented which characterizes experimenter subject interactions in such a way as to allow an investiga tor to more easily attempt a simulation. A sample simu lation of concept identification behavior is included to illustrate the use. of this paradigm.

This paper introduces the concept of a general inter face between simulation models and simulation pro gramming languages. The interface is a program generator. The simulation model is represented in a symbolic form through an entity-cycle diagram. Three program generators are cited, and examples of two are given. The advantages of the technique are discussed.

This paper describes and compares the application of two alternative approaches to the identification of linear systems from input-output data. The first approach is based on fitting input and output data by sums of exponentials and deriving both the system structure and the parameter values from the resulting functions. The second approach is based on assuming a differential-equation structure and finding only the parameter values from the given data. The relative advantages, disadvantages, and computational aspects of both approaches are discussed with reference to a specific biological problem. The results are applicable to a variety of engineering and scientific problems described by systems of first-order linear differential equations.

A simulation of human speech is normally performed to represent the acoustic signal in a form from which meaningful features of interest in the time and frequency domains may be identified and used in attempting to solve research and applications prob- Zems. Frequency-domain simulations related to vocal tract modeling have been the most popular techniques used to date. However, approaches have been investi gated which, for the limited case of voiced sounds, simulate human speech in the time domain by time- ordered time-based wavefunction sets. This paper discusses the use of a new technique for an accurate, complete simulation of any human speech by the family of Hanning Cosine Modulated (HCM) wavefunctions. The previous research on wavefunction simulations of speech are first discussed and then, through illus trative examples, it is demonstrated that HCM wave functions may accurately simulate the behavior of speech in either the time or frequency domain. The actual parameters of the wavefunctions which re present filtered speech may be measured by an auto matic digital signal analysis technique. These parameters then form a data base which, in conjunc tion with the HCM model, provide a closed-form solution in the time or frequency domain for analyzing and synthesizing speech, and they permit a convenient determination of a wide variety of speech features of interest.

This paper is a brief report on an investigation into the performance of integration routines used in the simula tion of dynamic systems by general-purpose digital simu lation programs. The performance is compared with respect to overall speed, accuracy, and convenience. Several routines are considered, including rectangular, Runge-Kutta, Runge-Kutta-Blum, Runge-Kutta-Mer son, and Adams-Moulton methods. The last two may use either fixed- or variable-step-size methods. The Tus tin method and the State-Transition method are also included. By examining the performance of the routines through tests on representative systems, it is shown that for the simulation of linear systems the state transition method works best. For the general simulation of linear and non linear systems the variable-step-size Runge-Kutta-Mer son method proves to be most accurate and most efficient. The variable-step-size method may be designed to gen erate outputs at uniform intervals, thus greatly enhanc ing its value when synchronous interaction with other parts of a simulation is required.

This paper discusses the present atzd futtsre of hybrid computation: Has it reached its zenith, or will it continue to develop as a powerful but specialized technique? The minicomputer has made current hybrid systems practical by providing a low-cost digital component. Unless the cost of the analog computer is reduced, the increasing power and decreasing cost of the minicomputer may severely limit future hybrid computer usage by providing an economic all-digital alternative. However, problems with alternative solution techniques indicate that for some important applications the hybrid computer will be the most cost- effective approach.

ISL-8 is the first single precision, parallel interactive simulation language available for a minicomputer. This paper illustrates ISL-8 programming procedures and opera tional conveniences for solving a second-order differential equation and three nonlinear problems from science and engineering. The relationship to integration step size and computational accuracy is also shown. The article con cludes that ISL-8 as implemented on any of the 4K PDP-8 family of computers can accommodate up to 25 nonlinear differential equations at computational speeds 10 to 50 times faster than languages using floating-point methods. (8K systems can accommodate up to 100 nonlinear equa tions.)

This paper is concerned with attempts that have been made to incorporate measures of risk in the selection of financial investments. The methodology that is developed here has been primarily aimed at the evaluation of capital investment projects, but the concepts and ideas are also relevant in the financial investment field. Implicit in all this discussion is the fact that when information is obtained in order to calculate an internal rate of return or a net present value, the figures put into the calculation are estimates rather than precise and exact quantities. In some instances the estimates may be felt to be very good, whilst in other cases it may be felt that there is a wide range of possible deviations. Hence the evaluation of any project or projects should, correctly, be described not by a single criterion but by a range of possible values, some of which are judged to be more likely than others. Having said this, there now comes the need to decide upon the method by which choice is to be exercised. Should it be by the choice of the highest average value of the criterion, or by choosing some more conservative rule?

Most models of economic life to replacement of equipment liable to deterioration with age but not catastrophic failure may be reduced to a basic model. Serious errors could result from use of this model if economic life is sensitive to errors in estimates of its parameters or mistaken assumptions as to the functional relationships between costs and age. Results of an investigation into the sensitivity of economic life to such errors are presented in this paper. The extent to which the model adequately maps the real problem is also considered briefly

Of current interest in the field of automatic control is the problem of system identification in the presence of measurement noise. Generally this problem has been dis cussed in the literature for the case of linear time-invar iant systems where the parameters to be identified are constant or slowly varying. This paper describes the ap plication of continuous stochastic approximation meth ods for the identification of a class of simple nonlinear systems. The two algorithms described are easily imple mented with analog equipment, although one of them requires some logic capability.

This paper presents a method of interactive or real-time predictive control which can be implemented without a stored program computer. By restricting the process dy namic model to a second-order-plus-dead-time format, a configuration which is satisfactory for many chemical or petroleum processing systems, a time-optimal controller can be built entirely on a small analog/hybrid computer. For set-point changes, the controller utilizes an analog model of the process to search out the predicted optimum input switching sequence and continuously maintains the process on the time-optimal trajectory. Because of the predictive nature of the method there is no necessity to include the time-delay term in the high-speed model; the predicted switches in the process input are simply ad vanced in time by the amount of the time-delay.

Iterative techniques for function optimization have been considered extensivezy for use in all-digital computation. Relatively little has been done to take advantage of the much higher integration speed of hybrid computation systems. This paper demonstrates application of one simple procedure in a hybrid envi ronment and compares the results to those obtained by an efficient digital procedure. Even though a much more efficient procedure was used on the digital, time-saving factors between 8 and 2 were obtained via the simpler hybrid implementation. Since the dollar cost of the hybrid is much less than that of the digital, the hybrid has a large advantage per solution.

An essential part of almost all air-defense computer simulation models is the visibility subroutine. This subroutine performs the necessary computations and provides the data required to determine whether enemy aircraft or missiles may be detected by stationary ground-based air-defense system sensors- that is, whether they are within sensor range and whether their visibility is obstructed by terrain irregularities and/or the curvature of the earth. This paper presents a new approach to the visibility subroutine, entitled MAVD (Minimum Altitude Visibi Zi ty Diagram). MAVD provides a visibility subroutine which could be incorporated into existing models relatively easily. The authors believe that it would result in a more realistic simulation and provide a significant reduction in computer time, in comparison with other simulation methods with which the authors are familiar.

The validation of simulation models is not often treated to any great depth in reports on their use. This paper considers the problems involved in the validation and calibration of several classes of simulation models. Particular stress is laid on the checking of the internal consistency of the model against the problem as posed by the model designer. Distinctions are drawn between fitting, calibration, and validation of a model, and some of the implica tions for forecasting using such a model are discuss ed. Systematic errors may be introduced in any of these stages, and it is argued that the estimation of their magnitude is an intrinsic part of the validation process. Distinctions are drawn between validation procedures appropriate to discrete-event and continuous-simulation models, and to dynamic and cross-sectional structures in the systems modelled.

The purpose of this paper is to report on a study which involves a simulation of a hypothetical job shop with several machines. The investigation employs GASP-II as a computer language. This simulation study is concerned with: (1) testing a new method of assign ing job due-dates and (2) comparing and evaluating the effect of different processing-time distributions on the performance of a number of scheduling rules. The most significant results of this study are: (1) The shortest-imminent-operation rule is superior to others in reducing job lateness and shop flow time (2) The procedure in which the due-date allowance is proportional to the number of operations and work content of the jobs has proved to be beneficial in the case of the non-due-date rules (3) The operation of a job shop using the shortest-imminent-operation and slack-per-remaining-number-of-operations rules is degraded when the processing-time distribution having Erlang parameter K equal to 4 or 8. However, per formance is better when K = 8 than when K = 4 (4) The GASP-II package works efficiently for large-size shop problems.

Most of the prior research work concerned with the analysis and synthesis of speech has employed Fourier series and transforms as the mathematical framework for the characterization of live speech. Current studies at UCSB have taken a new approach, namely the time domain analysis of speech waveforms. This work, per formed under the direction of Dr. Glen J. Culler, is based on a Gaussian wave-function approach to the analysis, synthesis and decomposition of speech elements. This paper is concerned with the hybrid simulation of speech waveforms utilizing the wave-function representation. A speech simulator consisting of a hybrid interconnec tion of four TR-20 analog computers is described. Re sults of the simulation of two vowel sounds, the "o" in "on" and the "a" in "at," are presented and compared with the original speech data. The results form an im pressive check on the wave-function approach and have provided the impetus for the realization of a hybrid speech-synthesizer based on this technique.

Digital data processing necessarily involves quantization (roundoff) of input data. The statistical theory of amplitude quantization indicates that the effects of quantization on statistics are often negligible or can be approximately predicted and corrected, even with surprisingly coarse quantization. This tutorial paper reviews contributions to the theory made in England, the Netherlands, and Russia, as well as B. Widrow's original work in this country. Applications to remarkably inexpensive hybrid analog-digital averaging computers and correlators are also discussed. The theory can pay very handsome practical dividends: in many applications, 2- to 4-bit analog-to-digital converters and data-transmission channels can yield averages, mean squares, and correlation functions with 10- to 20-bit accuracy, and one-bit (polarity-coincidence) correlators are often practical.

This paper presents analog, hybrid, and digital simulation techniques for one-dimensional flow, diffusion, and wave- type systems. These techniques also provide the basis for the simulation of more complex distributed-parameter systems. Many of the techniques considered are well known; however some of them are just recently devel oped. The various techniques result from either discretiza tion and transformation of the original partial differential equations or from application of the method of charac teristics. From this point of view, it is possible to fit the techniques into a general framework. It will be indicated by means of the properties of each technique which type of computer will best implement it and which system can be treated by it most advantageously.

This paper reviews the software and hardware requirements for on-line computing. It shows these facilities can be used to conduct on-line simulation experiments of models of agricultural systems. The general problems associated with modelling are indicated and the advantages of the on-line approach are discussed. Key words and phrases: on-line computing, modelling agricultural systems, systems analysis in agricul ture.

Some of the material in this paper was first presented at an International Symposium on Sensitivity Analysis in Dubrovnik, Yugoslavia, in September 1964

This paper describes a method for obtaining an engineering solution for integrodifferential equations of physical systems using an electronic system. The components consist of standard plug-in feed-back amplifier units. As the interconnections are wires, resistors, and capacitors, no complicated mechanical layout problem is involved and a generally flexible analyzer need not be set up, for it is a simple matter to assemble the particular circuit for any system of equations for which solutions are desired. The system should, therefore, be of interest to those involved in a study of the dynamics of physical systems.

No mortality curve appears yet to have been devised which both fits the national statistics at all ages, and also has a simple philosophical explanation. The object of this paper may be stated as being the building up of a new mortality curve from its component parts, as many as possible of these parts having a philosophical background, and the whole being applicable to the national data.

This paper describes briefly a digital simulation program developed at the University of Waterloo, Canada. The program accepts the description of a dynamic three-dimensional mechanical system as input (presently limited to particle masses) and then formulates and integrates the equations of motion of the system. The resulting dynamic response is displayed on a CRT screen for observa tion. The theoretical basis for the program is the vector-network model, a recently-developed formula tion process based on graph theory. The computer graphics subroutines include a novel package for previewing and photographing the dynamic sequences on 16-mm motion-picture film. With some further development, the program could become a useful engineering tool for the design of dynamic mechanical systems.

In the recent literature, a paper1 and a contract report2 indicate that automatic analog computer patching sys tems have become economically and technologically feasible. In anticipation of the development of a variety of proposed patching systems, a specific decision pro cedure is developed for the rational evaluation of com peting automatic analog patching systems. The design procedure described in this paper facilitates selection of the patching system best suited for a partic ular computer and a particular problem class. Sufficient information is abstracted from a typical problem set to determine the number of relays required for (and thus to assign a cost figure to) each possible patching system without requiring the actual design of each system. Nom inally, the best system is that with the lowest cost figure. It is recognized, however, that there are many subjective decisions (e.g., captive vs. free pots) involved in the de sign process that are not quantified in this overall cost figure. This problem is met in part by providing inter mediate information relating to the optimal module size, combinations of proposed systems, etc. to help estab lish some of the sensitivity properties of the optimal system. The procedure is illustrated by applying it to a particu lar case and evaluating a number of alternative switching systems.

In earlier work we have reported on CELIA, a program for simulating the behavior of linear iterative arrays of cells. The action of a cell is influenced by both its neighbors. Our main application area was biology, where the program can be used to test hypotheses about the developmental rules for organisms. More recently we have been applying our program to test proposed solutions to some fairly complicated biozogically-based problems. We have investigated whether one can achieve regulative global polarity in organisms without polarity in individual cells, by solving the French flag problem of Wolpert, using only symmetric elements. In our attempted simulation of pigmentation patterns on the shells of sea-snails, we have found it necessary to find a solution to a generalized version of the firing squad synchronization problem, in which the firing squad is growing while it is trying to synchronize itself. We found that our original program CELIA was some what awkward for simulating such complicated situa tions. A new program has been devised in which the state of the individual ceZZs is described by a set of attributes and each attribute itself may be a list structure. The present paper reports on these changes and ex plains how they turn out to be useful in the appli cations mentioned above.

A new computer technique is described which yields the partial derivatives of problem variables with re spect to pertinent system parameters simultaneously with the solution of the original system differential equations. These derivatives, known as parameter influence coefficients, are valuable to the analyst in enhancing his understanding of system characteris tics. If the problem solution x(t, λ) and the parameter influence coefficient ∂x/∂λ (t, λ) is known for a par ticular operating point where λ= λ0, then it is pos sible to make a first-order prediction of system behavior at a neighboring point having the new parameter value λ1 = λ0 + Δλ. Similar predictions can be made if not one but several parameters are to be varied. Thus, the knowledge of parameter in fluences often helps to reduce the total number of computer runs required in a parametric system study. Typical applications of the technique are: linear ex trapolation in the neighborhood of a known solution, determination of design tolerances of a system, and prediction of critical parameter values and stability boundaries. The most useful application pertains to systems disturbed by random noise where normally a very large number of computer runs would be re quired to analyze the system on a statistical basis in a variety of operating conditions. Several illustrative examples are presented in the paper.

CHEMCSMP is a convenient problem-oriented language based on CSMP/360 and is designed specifically to solve problems in kinetic chemical equations. Users' statements, which resemble conventional chemical nota tion, describe slow and fast reactions in multicompartment systems and flows between compartments. CSMP statements may be inter mixed with chemical equations to provide for a variety of computations and convenient output. CHEMCSMP is implemented via a pre compiler which translates the user's pro gram into pure CSMP. The paper describes the use and implementation of CHEMCSMP and provides an illustrative example.

This paper presents an analog/hybrid simulation method for linear and nonlinear distributed para meter systems. The modal method is based on a functional approximation for the computer solution of initial value partial differential equation prob lems using eigenfunctions. First, the mathematical foundations are reviewed; then the questions concerning the technical realiza tion on an analog/hybrid computer are treated. The simulation results for two examples of chemical processes are discussed.

Most of the activity of actuaries with computers to date has been concerned with putting actuarial work on to computers, or—at one stage further back—putting life office data on to a computer data-processing system. This paper attempts to contemplate the next stage, the effect of computer knowledge and familiarity on the profession and its techniques.