Information Technology Reference
In-Depth Information
redundancy due to the intrinsic structure of the language which every native speaker knows.
This redundancy can be used to correct mistakes in transmission. The Huffmann algorithm
for compression
16
is based on a knowledge of the relative probabilities of different
symbols, measured over a particular text. Since Shannon, the analogy between DNA and
protein sequences and natural languages has been pervasive.
Information theory was developed in dialogue with the construction and use of
computers which have made both the examination of the arrangement of atoms and the
operation of data-bases possible. “Cyberspace” was invented and colonised the literary
world
17
.
Donald Booth at Birkbeck, recruited by Bernal to make a computer for
crystallography, invented the floppy disc
18
, using a primitive speech recorder with a
magnetic disc, but he discarded it, and toyed with the machine translation of natural
languages, an idea, which emerged in discussions with Warren Weaver. The Cambridge
Crystal Structure Database was begun in an attic at Birkbeck College, originally on cards,
before being established in Cambridge. Its creation was due to Olga Kennard and J. D.
Bernal (who had far earlier been concerned with the development of Structure Reports
(originally Strukturbericht) collecting all data on the arrangement of atoms in crystals.
Gregory Chaitin proposed that the amount of information in a structure could be
defined in terms of the shortest computer programme necessary to generate it. The number
of operations necessary to sort a sequence of N numbers into an arbitrary order is N log N
("operation" needs more careful definition).
5. Cellular automata
Robert May alerted us to the fact that there were many “simple mathematical models with
very complicated dynamics”, although the immensely creative J. B. S. Haldane had noted
this more graphically in 1932
19
. In particular, finite difference equations, for example x
t+1
=
f[x
t
], have results which cannot be predicted far ahead better than by simply iterating the
process. It also emerges that eventually the finite accuracy of all computing processes,
including those in nature, will render the outcome indefinite and unpredictable. This kind of
equation can be extended to two or three (or more) dimensions, the equations may be
coupled or non-linear, so that the complexity increases. Stephen Wolfram
20
has developed
certain classes of “cellular automata” in such detail that classification is possible. Intriguing
and unpredictable patterns may emerge
21
. It is immediately clear that patterns in nature,
particularly those in biological systems produced by the switching on and off of genes
which synthesise proteins, must be physically analogous to such mathematical phenomena.
Now even the classical mechanical problems of Newton, the pendulum and the solar system
are seen to be weakly chaotic.
6. Structural molecular biology. Proteins and nucleic acids
Desmond Bernal had the good fortune to be the right man in the right place at the right
time. In February 1945, before returning to Birkbeck after the war, Bernal produced a plan
"to set up a research centre for the study of the structure and properties of large molecules
by all available physical and chemical methods".
This was based directly on the thinking of
the Cambridge club and was effectively the charter for the Birkbeck Laboratory, set up in
21-22 Torrington Square, which Bernal headed from 1938 to about 1964. In the 1950s
