Humanist Discussion Group, Vol. 17, No. 143.
Centre for Computing in the Humanities, King's College London
www.kcl.ac.uk/humanities/cch/humanist/
Submit to: humanist@princeton.edu
Date: Tue, 08 Jul 2003 06:37:39 +0100
From: Jan Christoph Meister <jan-c-meister@uni-hamburg.de>
Subject: Categorizing loops
13:09
07.07.2003
My question concerns a possible typology of infinite loops occuring in
combinatorial computer programs. The distinction that I have in mind
is that between
(a) infinite loops where recursion results in the continuous
stacking of a processing instruction WITHOUT ever being able to
output results, and
(b) infinite loops where recursion is a consequence of continuously
having to reconsult dynamically expanding data. This happens
when a combinatorial program manages to instantiate its variables,
then writes its newly computed result into its own memory as an
additional 'fact'-clause, and then has to embark on a re-run
because its combinatorial algorithm will now encounter an unflagged
(hitherto unprocessed) fact.
In PROLOG-practice the first results in (a) stack overflow, the second
in (b)heap overflow. To my hermeneutically afflicted mind the two
would seem to be logically related (i.e., both are cases of infinite
nesting), but epistemologically significantly different : (a) is
epistemologically completely redundant, (b) is epistemologically
exponential. Of course, philosophically speaking this might very well
amount to the same thing in that knowing nothing about the world (a)
is about as bad as realizing that one will never be able to know how
much it is that one doesn't know, and what proportionate value the
knowledge produced thus far actually has - this because (b) in
principle subverts the idea of approximation in knowledge.
Anyway - does this type of distinction matter to CS and if so, what is the
appropriate
terminology?
Many thanks,
Chris
*******************************
Jan Christoph Meister
Forschergruppe Narratologie
Universitt Hamburg
ACP - Computer Philology Working Group at Hamburg University
www.c-phil.uni-hamburg.de
NarrNet - the Information hub for Narratologists:
www.narratology.net
My site: www.rrz.uni-hamburg.de/JC.Meister
Mail: jan-c-meister@uni-hamburg.de
Office: +49 - 40 - 42838 4994
Cell: +49 - 0172 40 865 41
This archive was generated by hypermail 2b30 : Tue Jul 08 2003 - 01:44:24 EDT