Dr Peter Fenwick

Current Position

After about 20 years in the Department of Computer Science (and 12 years before that in the Computer Centre) at the University of Auckland, I took early retirement from Jan 2005.

My University email and other addresses (but not telephone) should continue to be available, but will be handled less often ....

Important notice — I am NOT available for supervising student projects or similar activities.
This most definitely includes internships for foreign students, so don't even bother asking.

Enquiries about past research may be viewed sympathetically, but by now I am probably out of touch.

17 Dec 2015

Research Interests since retirement

I have been conducting occasional research since retirement. In chronological order the results are --

1. A general overview and cogitations on Burrows-Wheeler compression.
   See Burrows–Wheeler compression : Principles and reflections” Theoretical Computer Science Vol 387(2007) pp 200–219.
   Abstract After a general description of the Burrows Wheeler Transform and a brief survey of recent work on processing its output, the paper examines the coding of the zero-runs from the MTF recoding stage, an aspect with little prior treatment. It is concluded that the original scheme proposed by Wheeler is extremely efficient and unlikely to be much improved.
   The paper then proposes some new interpretations and uses of the Burrows Wheeler transform, with new insights and approaches to lossless compression, perhaps including techniques from error correction.

2. “PPM Compression without Escapes”, Software – Practice and Experience, online : 25 APR 2011, DOI: 10.1002/spe.1070 ; printed Vol 42, no 2, pp 255–260, February 2012)
   Abstract A significant cost in PPM data compression (and often the major cost) is the provision and efficient coding of escapes while building contexts.
   This paper presents some recent work on eliminating escapes in PPM compression, using bit-wise compression with binary contexts. It shows that PPM without escapes can achieve averages of 2.5 bits per character on the Calgary Corpus and 2.2 bpc on the Canterbury Corpus, both values comparing well with accepted good compressors.

3. A book “Introduction to Computer Data Representation” Bentham Publications 2014. It provides an introduction, from first principles, to the internal (bit-wise) representation of integers, real numbers, characters and text, checksums and universal coding of integers. It has about 270 pages and is available as both an eBook(eISBN 978-1-60805-882-2, 2014) and printed copy (“Print-On-Demand” ISBN 978-1-60805-883-9).
   While not a book for beginners, it is intended for reference by advanced students and professionals.
   You may download a 1-page flier, or the primary link includes ordering information, the ability to download individual chapters and a free “sample” which is the first 3-4 pages of each chapter or other section. The first 3 sections (which are all free) give a good overview of the book.

4. “A Note on Variable-Length Codes with Constant Hamming Weights”
   Journal of Universal Computer Science, vol. 21, no. 9 (2015), 1136-1142 is available at http://www.jucs.org/jucs_21_9/a_note_on_variable.
   Abstract A recent paper described a variable-length integer code based on the Goldbach conjecture where every codeword had exactly 2 1-bits but with an extremely irregular structure. A later, unpublished, work produced a much more regular code, again with a Hamming weight of 2. This paper extends that later work to weight-3 and weight-4 codes, which are shown to be competitive with more-usual codes over a useful range of values.

Most of my papers and reports are available from here

You may also read ―

A list of my publications

Some ideas on Burrows-Wheeler compression, and compression in general

— F I N I S —