05.01.2013 Views

COMPUTATIONAL PROBLEMS IN ABSTRACT ALGEBRA.

COMPUTATIONAL PROBLEMS IN ABSTRACT ALGEBRA.

COMPUTATIONAL PROBLEMS IN ABSTRACT ALGEBRA.

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

198 A. L. Tritter<br />

is found to be “yes”, in case it is found to be “no”, in which of two modes<br />

a continuously monitored visual display is to show progress, and finally<br />

something slightly peculiar. Suppose that an input discrepancy is found<br />

on, say, a B-C pass. A decision is taken to “back off” and run A -+ B again.<br />

Now, if this decision is implemented immediately, no new problem arises.<br />

But, suppose as well that we interrupt at this point; then, when we go back<br />

onto the machine to re-run A-B, there is no record anywhere except in<br />

our own minds that it is not possible to back off from this pass (as C is<br />

not the predecessor of A here-it has been altered on the abortive B-C<br />

pass). The last switch is used to inform the program that thejirst magnetic<br />

tape pass of the present machine run is one from which no back-off is<br />

possible; it is, of course, used on the very first run, as well as in the more<br />

complex situation described above.<br />

One more word about the triangularization. It has been necessary, for<br />

efficiency’s sake, to economize as much as possible on the total time for<br />

which the program occupies Atlas, and this has meant developing a new<br />

triangularization algorithm. This algorithm will be discussed elsewhere,<br />

rather than here. It is a natural modification of existing methods, simply<br />

taking maximal advantage of the fact that the matrix is over a finite field<br />

while, at the same time, using the space outside the upper right triangle<br />

of the matrix, as it is progressively vacated by the triangularization, as the<br />

place to record all the “pedigree” data (there is exactly the necessary<br />

amount of space).<br />

8. Present status. The matrix-generating program is written, debugged,<br />

and run. The triangularization algorithm has been tested by hand, and is<br />

correct. The triangularization program is written and undergoing debugging.<br />

REFERENCES<br />

1. C. R. B. W RIGHT: On the nilpotency class of a group of exponent four. Pac. J.<br />

Maths. 11 (1961), 387-394.<br />

2. Unpublished private communication.<br />

3. S. M. JOHNSON: Generation of permutations by adjacent transposition. Maths.<br />

Comp. 17 (1963), 282-285.<br />

4. J. ARMIGER TROLLOPE: Grandsire: The Jasper Snowdon Change-ringing Series,<br />

pp. 51, 122 (Whitehead & Miller, 1948).<br />

5. DOROTHY L. SAYERS: The Nine Tailors (Victor Gollancz, 1934, many reprints).<br />

Some combinatorial and symbol manipulation<br />

programs in group theory<br />

JOHN J. CANNON<br />

Introduction. Over the past two years computers have been used to<br />

carry out a number of large calculations, of both a numerical and nonnumerical<br />

nature, arising out of research in group theory at Sydney.<br />

These problems include :<br />

(i) construction of subgroup lattices;<br />

(ii) investigation of positive quadratic forms;<br />

(iii) determination of the groups of order p6, p > 2;<br />

(iv) construction of counter-examples to Hughes’ conjecture in group<br />

theory.<br />

Only (i), (iii) and (iv) will be discussed here.<br />

All programs described here have been written in the English Electric<br />

KDF9 Assembly Language, USERCODE.<br />

Subgroup lattices. A program has been written which determines the<br />

generators and relations of all the subgroups of a finite soluble group. The<br />

program finds the subgroups using the same method as Neubiiser [l] with<br />

the difference that as a subgroup of order d is found the generators and<br />

relations of all possible groups of order d are checked through to find a set<br />

of generators and relations for the subgroup. As the program is restricted<br />

to groups of order less than 400 by machine considerations, one only needs<br />

to know all the possible groups of order less than 200, and most of these<br />

are known.<br />

The program reads the generators and relations of the given group and,<br />

using coset enumeration, finds a faithful permutation representation<br />

(possibly the regular representation). The permutation representation is<br />

used to determine the elements of the group and to find its Cayley table,<br />

and is then discarded. At the same time as the group elements are being<br />

found as permutations they are also found as words in the original abstract<br />

generators. These n words are stored in an n-word stack so that instead<br />

of using either the abstract word or its permutation representation,<br />

one uses the number indicating the position of the group element in this<br />

CPA 14<br />

199

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!