Ewing L. Lusk

Logic Machine Architecture: Kernel Funtions

In this paper we present an attempt to abstract from the great diversity of approaches to automated deduction a core collection of operations which are common to all of them. Implementation of this kernel of functions provides a software platform upon which a variety of theorem-proving systems can be built, We outline the architecture for a layered family of software tools to support the development of theorem-proving systems and present in some detail the functions which comprise the two lowest layers. These are the layer implementing primitive abstract data types not supported by the host language and the layer providing primitives for the manipulation of logical formulas. This layer includes the implementation of efficient unification and substitution application algorithms, structure sharing within the formula database, and efficient access to formulas via arbitrary user-defined properties. The tools are provided in a highly portable form (implemented in Pascal) in order that a diverse community of users may build on them.

Logic Machine Architecture: Inference Mechanisms

Logic Machine Architecture (LMA) is a layered implementation of theorem-proving tools. The kernel of formula storage, retrieval, and manipulation primitives (layers 0 and 1) is described elsewhere[2]. The layer described here (layer 2) contains resolution- and equality-based inference rules, subsumption, and demodulation. It is designed to provide all of the tools required to create a theorem-prover with minimal effort. Although LMA is currently implemented in Pascal, an interface to LISP will be included in the original release. This paper includes the design principles and techniques used in layer 2, as well as two simple theorem provers which illustrate the services of layer 2 — one written in LISP and the other in Pascal.

Experiments with resolution-based theorem-proving algorithms☆

Abstract This paper describes the study of some specific test problems in automated theorem proving. It is shown how experimentation on relatively simple problems can lead to the development of techniques which prove successful on much more difficult problems. A number of enhancements to a resolution-based system obtained in this way are described.

An automated reasoning system

This paper is an introduction to an automated reasoning program developed at Northern Illinois University and Argonne National Laboratory over the past nine years. Recently the program has reached the stage where it can be considered a useful research tool in a variety of disciplines. It has solved open problems in mathematics and participated in the design of new electronic circuits. Here we describe the general types of capabilities provided to the user by the program and give examples of how they are currently being used in diverse areas of investigation.

A practical design methodology for the implementation of IMS databases, using the entity-relationship model

The design of IMS databases is a current topic of major industrial significance. Compared to many other database management systems, IMS offers a wide variety of physical implementation options. One effort to help users choose among these options was 161. Unfortunately, it dealt mainly with descriptions of the options and offered minimal guidance on how to maka detailed design d8CiSiOnS. In order to keep from getting lost in the detailed terminology of I& one needs recourse to basic data structure design principles. On8 of our purposes here is to describe IMS in tarms of fundamental data structures and show how IMS implementation deoisions follow from an understanding of those basic principles.

Data Structures and Control Architectures for Implementation of Theorem-Proving Programs

This paper presents the major design features of a new theorem-proving system currently being implemented. In it the authors describe the data structures of an existing program with which much experience has been obtained and discuss their significance for major theorem-proving algorithms such as subsumption, demodulation, resolution, and paramodulation. A new architecture for the large-scale design of theorem proving programs, which provides flexible tools for experimentation, is also presented.

An advanced undergraduate course in applied computer science

The purpose of this paper is to describe a senior level course in the applied computer science curriculum at Northern Illinois University. The course, Database and Data Communications, has been taught for four semesters, and enrollment has steadily grown.n The course has a number of purposes. The primary one is to acquaint students with the principles of modern database management and teleprocessing applications and to instill proficiency in the writing of application programs for at least two of the major software systems supporting such applications, IBMs Information Management System (IMS) and Customer Information Control System (CICS). A secondary purpose is to consolidate and extend knowledge and techniques learned in previous courses. Course content which fulfills this purpose includes structured programming in both assembler language and COBOL, applied data structures, and access methods not previously covered (BDAM and VSAM).