Student Projects

The following projects are offered to BSc and MSc students at the University of Reading who are reading Cybernetics or Computer Science. Projects could be offered to students reading Mathematics or Philosophy. PhD projects could also be offered.

Anyone, anywhere in the world is free to carry out these projects using their own resources.

Perspex GUI

The perspex machine is a theoretical machine that exceeds the computational power of the Turing machine. It operates by performing matrix transformations in a 4D space. A simulation of the perspex machine and a GUI, both implemented in C⁺⁺, already exist. The simulation could be improved in efficiency by replacing linked lists with a hash table.

Alternatively, the GUI could be linked to efficient simulations of the perspex machine developed in other projects. The GUI could be extended by adding features to improve the visualisation and debugging of perspex programs. This GUI project can be performed on a PC using a C⁺⁺ compiler and OpenGL. It requires the purchase of a LEDA licence at approximately £150. This sum will be paid by the university. The project would suit one student and might lead to results publishable in a scientific conference.

C to Perspex Compiler

A compiler that converts a subset of the C programming language into perspex neural nets already exists. The compiler is implemented in Pop11. The project will re-implement the existing compiler in C and extend it with the eventual aim, outside this project, of achieving recursion when the compiler can be compiled into a perspex neural net. The project can be performed on a PC using compiler tools such as LEX and YACC. The project could be performed by one or two students. In the case of two students each would develop a part of the compiler and compare compilation strategies. Later, once the core of C had been implemented, each student could take on different parts of the C language. The project might lead to results publishable in a scientific conference.

Perspex Emergence from Random Space

There is a theoretical reason to believe that if a perspex space is initialised at random and execution of perspexes is repeatedly started at random places then the execution will not be random, but will organise into large clusters of space, filled with subspaces, that all perform semi-stable computations. The project will use statistical methods and a GUI to test the hypothesis of self organisation. This project can be performed on a PC. It requires the purchase of a LEDA licence at approximately £150. This sum will be paid by the university. The project would suit one student and might lead to results publishable in a scientific conference.

Perspex Genetic Algorithms

An algorithm exists for converting any Turing computable program into a genetic structure where a single parameter controls the whole program and successive parameters control successively finer parts of the program. This raises the possibility of having an apparently directed genetic evolution in which a program finds a local optimum in a search space before performing fine optimisations. This project will implement the algorithm and test it to see if this global-to-fine optimisation does occur. This project can be performed on a PC. It requires the purchase of a LEDA licence at approximately £150. This sum will be paid by the university. The project would suit one or two students. In the case of two students they would work together to implement the algorithm and then each student would explore different search spaces, mutation and cross-over strategies. The project might lead to results publishable in a scientific conference.

Isolinear Perspex Computation

Perspex space can be convolved with a triangular kernel, subject to certain boundary conditions, so that it yields an isolinear space in which every point in space contains a program and nearby points contain nearly the same program. This project will aim to produce an efficient simulation of an isolinear perspex space, implemented in C⁺⁺, and will test isolinear perspex programs to see how resilient they are to damage and whether or not the position of a perspex start can be used as an alternative to a genetic algorithm. This project can be performed on a PC. It requires the purchase of a LEDA licence at approximately £150. This sum will be paid by the university. This project would suit one student. The project might lead to results publishable in a scientific conference.