home *** CD-ROM | disk | FTP | other *** search
- Xref: sparky comp.ai.neural-nets:3369 comp.ai:3295 comp.theory:1854
- Newsgroups: comp.ai.neural-nets,comp.ai,comp.theory
- Path: sparky!uunet!secapl!Cookie!frank
- From: frank@Cookie.secapl.com (Frank Adams)
- Subject: Re: measuring complexity (was Re: Kolmogorov Complexity)
- Message-ID: <1992Sep01.221808.40967@Cookie.secapl.com>
- Date: Tue, 01 Sep 1992 22:18:08 GMT
- References: <1992Aug27.181810.7249@cs.sfu.ca> <1992Aug28.164941.2028@fid.morgan.com> <1992Aug31.211753.13335@cs.sfu.ca>
- Organization: Security APL, Inc.
- Lines: 23
-
- In article <1992Aug31.211753.13335@cs.sfu.ca> jamie@cs.sfu.ca (Jamie Andrews) writes:
- >In article <1992Aug28.164941.2028@fid.morgan.com> sethb@fid.morgan.com (Seth Breidbart) writes:
- >>What do you mean by probability? If you mean, for any string S, with
- >>probability > (1-delta) P is within epsilon of the shortest program
- >>for S, then A will lead to a probabilistic solution to the busy beaver
- >>problem.
- >
- For any given problem X, there is a number K and a probability psi>0, such
- that for any length N > K, the probability that a string S of length N is a
- program easily recognizable as having a halting problem equivalent to X, is
- greater than psi. So if X is an unsolvable problem, you will never be able
- to get delta less than psi.
-
- > I'm just throwing out ideas, really, trying to come up with
- >a formulation of a problem that is interesting and non-trivial.
- >You're welcome to your view that the problems I've stated so far
- >are trivial, but that doesn't convince me that the whole area is
- >trivial (though it may convince me to stop talking about it on
- >the net :-)).
-
- In general, this kind of problem does not admit of *any* sort of general
- solution. You may learn something by trying things and figuring out why
- they don't work, but don't expect to find anything.
-