C. Michael Overstreet

A specification language to assist in analysis of discrete event simulation models

Effective development environments for discrete event simulation models should reduce development costs and improve model performance. A model specification language used in a model development environment is defined. This approach is intended to reduce modeling costs by interposing an intermediate form between a conceptual model (the model as it exists in the mind of the modeler) and an executable representation of that model. As a model specification is constructed, the incomplete specification can be analyzed to detect some types of errors and to provide some types of model documentation. The primitives used in this specification language, called a condition specification (CS), are carefully defined. A specification for the classical patrolling repairman model is used to illustrate this language. Some possible diagnostics and some untestable model specification properties, based on such a representation, are summarized.

Redundancy in model specifications for discrete event simulation

Although redundancy in model specification generally has negative connotations, we offer arguments for revising those convictions. Defining “representational redundancy” as the inclusion of any symbols not required to fulfill the study objectives, we cite several sources of redundancy, classified as accidental or intentional, that contribute positively to the model development tasks. Comparative benefits and detriments are discussed briefly. Focusing on the most interesting source of redundancy‐that which is intentionally induced by a modeling methodology—we demonstrate that automated elimination of redundancy can actually improve model execution time. Using four models drawn from the literature that are easily understood, but which represent some differences in size and complexity, the direct graphical representations shows improvements over a base case ranging from 27.3 percent to 68.1 percent in execution time. Further, increasing improvement is realized with increasing model size and complexity. These results are encouraging because they suggest that modeling methodologies with automated model diagnosis can significantly reduce both execution and developments time and cost.

Exploring the forms of model diagnosis in a simulation support environment

Our purpose is to explain the categorization of diagnostic assistance, using an example, in order to address the central question for a simulation support environment:In what forms can computer assistance be provided to improve the simulation model development task? This question must be answered in the context of the economic and technical realities of today, for the near future, and the long term. We present several examples of computer assistance which indicate that a significant degree of assistance is possible.

Using graphs to translate between world views

The standard world views for discrete event simulation of event scheduling, activity scanning, and process interaction are each used for model implementation, each supported by one or more simulation programming languages. We present transformations of a model representation in one form into model representations into each of these world views. This transformation process is interesting in that 1) it requires a characterization of each of these world views and 2) it demonstrates the potential of each world view to simplify a model specification.

An Optimal Parallel Algorithm to Reconstruct a Binary Tree from its Traversals

We consider the following problem. For a binary tree T=(V,E) where V={1, 2,...,n}, given its inorder traversal and either its preorder traversal or it postorder traversal, reconstruct the binary tree. We present a new parallel algorithm for this problem. Our algorithm requires O(n) space. The main idea of our algorithm is to reduce the reconstruction process to parallel merging. With the best results for parallel merging, our algorithm can be implemented in O(logn) time using O(n/logn) processors on the EREW PRAM, or in O(loglogn) time using O(n/log logn) processors on the CREW PRAM. Consequently, an ordered tree can be reconstructed from its preorder and postorder traversals. Our results improve the best previous results for this problem in literature either in cost or in the model of computation.

Introduction to Special Issue: Modeling and Simulation in Teaching and Training

Simulation has always been about learning; sometimes by model builders, sometimes by model users. In this special issue, we focus on simulations explicitly designed for use in educational settings. Educational applications are found in many fields and with different educational objectives. We have collected contributions from such diverse fields as teaching and training in psychology, in health science, in computer science, and in the military domain. Use of modeling and simulation in teaching and training can be closely related to Intelligent Tutoring Systems (ITS). In combining these two areas, the ITSs’ knowledge bases can be used for steering simulation executions, providing information for use in simulations, or for giving ’intelligent’ advice and feedback to students. Also, research in ITSs can by supported by introducing and simulating learner models for testing tutoring software, and integrating additional simulated actors in ’role-playing’. Simulated actors are sometimes also called “virtual humans”. From a top-level perspective, the domain of modeling and simulation for teaching and training can be divided in three main branches:

Computer Productivity Initiative

Over the last three decades computer science has evolved into a mature and experimentally oriented discipline with a well defined curriculum. Only recently have we come to realize that as a discipline computer science must reach beyond its own subject area to applications in other disciplines in order to stay relevant. Most computer science curricula teach principles and programming skills in isolation from an application perspective, provide limited laboratory experience, and introduce inadequate integration with non-CS components. The Computer Productivity Initiative, described in this paper, proposes to alleviate these problems by integrating a multi-year project into the curriculum. The project involves courses normally taken in three different years of the curriculum. It includes hardware and software issues and also addresses engineering, business, and other non-CS issues. The initiative uses prototyping and simulations in the development of specifications for an integrated television communication and display computer system. The students apply principles of productivity and make extensive use of leading-edge technologies both in the process of the project development and the product being developed. They hone essential career-oriented skills in the areas of management, formal presentations, and group problem solving. This paper is a report of work in progress. It emphasizes the implementation issues we are facing and the integration of evaluation into our curriculum development. It describes our preparation for the dissemination of a model curriculum when we are able to demonstrate that the approach is adaptable to CS departments across the country.

Reconstructing a binary tree from its traversals in doubly logarithmic CREW time

Abstract We consider the following problem. For a binary tree T = ( V , E ) where V = {1, 2, ..., n }, given its inorder traversal and either its preorder or its postorder traversal, reconstruct the binary tree. We present a new parallel algorithm for this problem. Our algorithm requires O ( n ) space. The main idea of our algorithm is to reduce the reconstruction process to merging two sorted sequences. With the best parallel merging algorithms, our algorithm can be implemented in O (log log n ) time using O ( n /log log n ) processors on the CREW PRAM (or in O (log n ) time using O ( n /log n ) processors on the EREW PRAM). Our result provides one more example of a fundamental problem which can be solved by optimal parallel algorithms in O (log log n )time on the CREW PRAM.

Performance trade-offs for a multimedia distributed application

We have designed and developed an Interactive Remote Instruction (IRI) system to support university-level instruction in a networked environment. The IRI system creates a virtual classroom where all participants have the same experience independent of their spatial location (separated by up to order of 100 km). Participants interact with audio, video, and collaborative tool operation information through a multimedia workstation. In this paper we describe an IRI testbed consisting of Sun Sparcstations 5 running Solaris 2.4 implemented over a local area network. We discuss the performance trade-offs and the scalability of the IRI system by analyzing results obtained during the execution of a test suite on the IRI testbed. The test suite consists of operations such as running a tool on a single workstations, running the tool collaboratively on multiple workstation, and finally running the tool collaboratively along with the display of student and teacher video images and audio. The most important conclusion of this work is that, though some improvements are required, we are able to achieve an acceptable performance within the goals we had established for IRI-required functionalities using moderately price hardware based on current technologies (about \

Enhancing understanding of discrete event simulation models through analysis

7K per student workstation). Feasibility means that response time for user commands is no more than twice in the worst case (Mosaic running on a single workstation by itself versus Mosaic running collaboratively on the testbed with multiple video streams and audio added). The results we obtained from this experimentation are being used in the design of a dynamic resource management scheme which will regulate the demand on the IRI system resources to maximize user perception of the quality of the virtual classroom. We briefly describe these and other planned improvements to the IRI system as well.