Conrad Mueller

A fundamentals-based curriculum for first year computer science

At Wits we are concerned about offering a good computer science degree but at the same time making our degree programme accessible to all students who have the potential or ability to cope with the material. This paper discusses a new first year curriculum which has been developed to address some of the problems which the course that we offered from 1990 to 1998, with minimal changes, has begun to encounter. The most important of these problems is that of student perceptions of our old course. The new course stresses fundamentals of computer science and is structured around teaching basic principles and competencies.

Agent-based modelling: a case study in HIV epidemic

This research presents an agent-based, bottom-up modelling approach to develop a simulation tool for estimating and predicting the spread of the human immunodeficiency virus (HIV) in a given population. HIV is mainly a sexually transmitted disease (STD) causing a serious problem to human health. The virus is transmitted from an infected person to another who was previously healthy through different biological, social and environmental factors. The research develops the simulation tool by modelling these factors by agents. Although research has and is being conducted to estimate and predict the spread of the HIV epidemic, the proposed research seeks to investigate the spread using a different approach. The previous models used a top-down modelling approach. They are built from the general characteristics and behaviours of the population. They have not explored the potential use of agent technology. This research attempts to investigate the flexibility that the multi-agent system offers. Agent-based models are close to the situations that exist in a given real system that consists of autonomous components interacting with each other. The modelling approach has the advantage of observing the interaction made between agents, which is a difficult task in the top-down modelling approach. The research investigates the performance of the tool and presents the first results obtained.

An improved first year course taking into account third world students

Theincreasing number ofeducationally disadvantaged students, typically from third world environments, entering our first year course has forced us to re-evaluate the course. The class composition is extremely non-homogeneous creating major problems in the teaching of a first year course concentrating on programming. To address these problems we have developed a new curriculum based on the ACM task force report and focusing on fundamental principles.

Development of a decentralised virtual service redirector for Internet applications

An ongoing project of the Highly Dependable Systems Research Programme (PHDS) at the University of the Witwatersrand is the development of a dependable distributed system and its application in supporting Internet services. The current research related to this project includes the design, specification, verification and implementation of a virtual service redirector for critical internet applications. This paper summarises this system and related projects.

Axiomatic computational model

A new model of computation is explored based on axiomatic logic. The significant difference from other models is how reasoning about variables is expressed. The computation is described only by defining variables in terms of other variables. The benefit is the simplicity of the semantics of the axiomatic logic. The model is described in the form of a tutorial using a simple language to illustrate how a program can be expressed in this model.

Addressing: the root of all programming evils

Virtually from the inception of computers, the von Neumann model of computation has gone unchallenged as the accepted general purpose mechanism of computation. A Web-search on computational models does not reveal any alternatives to compete with it. Ideas like data-flow architectures and the biological computer have had little impact. Few have done any critical assessment of the von Neumann model. Backus stirred considerable interest in his paper questioning its merits. However twenty five years later on little has changed. A different angle is taken here in trying to make a critical assessment by considering addressing: a fundamental underpinning of the model. Firstly we need to establish if addressing is inherent to computation. Having established it is not, a brief study is critical analysis is made of addressing. Concluding that there can be significant benefits if alternative to addressing can be found.

Axiom based architecture

The paper proposes an axiom based architecture as an alternative to the von Neumann model. The model has many desirable properties: fine-grained parallelism, simple semantics, better security and easy of programming. The empirical research gives some indication of its performance potential. A description is given as to how algebraic arithmetic expressions of relations can be broken up into primitive expressions consisting of a single operation. These primitive relations are shown to be sufficient to describe a Turing machine. Eight inference rules are given that define how the primitive relations can be evaluated. An outline is given of an architecture based on these inference rules. Finally a brief description is given of an experimental emulation and empirical evaluation of the architecture. Instead of manipulating data or values by applying instructions or functions, computation is applying existing elements to relations to create new elements. The elements identifier determines which relations the element applies to. The relation determines the identifier of the new element and the operation that needs to be applied to create the value of the new element. The conceptually indices are different in this model. Instead of seeing an index as an offset into an array, an index is seen as part of the element identifier. This enables infinitely many relations to be defined between unique sets using universal quantifiers. Thus every element, or value, computed has a unique description.

Retaining the Semantics of Data with the Data

A number of inherent weaknesses are identified with using addresses to reference values. These weaknesses negatively impact on the development of architectures using addressing. It is proposed that values should be referenced using semantic information retained with the value. A computational model based on simple arithmetic and natural deduction is presented that illustrates how such a referencing model can be implemented. This model overcomes the weaknesses identified with the von Neumann address-based architecture and compares well to what we ideally hope for in an architecture. The model goes a long way towards making it easy to write programs that execute efficiently on multi- or many-core systems.

Can Multicore Processing Learn from Arithmetic Concepts

Many see multicore processors as the way to solve the limitations of the uniprocessor. These architectures are not without problems. We argue that to address these problems, architectures needs to move away from an instruction-based model. The paper looks at how arithmetic and natural deduction can form the basis of a computational model that can be used to implement multicore architectures.