home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.ai.neural-nets
- Path: sparky!uunet!usc!rpi!lima
- From: lima@moon.ral.rpi.edu (Pedro Lima)
- Subject: Question:Measuring complexity of CPU-time?
- Message-ID: <6ghyfh_@rpi.edu>
- Nntp-Posting-Host: moon.ral.rpi.edu
- Organization: Rensselaer Polytechnic Institute, Troy NY
- Date: Thu, 20 Aug 1992 19:44:02 GMT
- Lines: 17
-
- The definition of Kolmogorov Complexity of a string s as the smallest program p
- that when presented as input to an Universal Turing Machine U outputs s
- may cause practical problems since U is allowed to run p for any length of
- time as long as it eventually terminates. This time may of course be too
- long to be practical (think of a NP-complete problem). Thus, some authors
- define a time-bounded complexity where essentially U is restricted to
- terminate within T steps leaving s as output.
-
- My question is: is there any problem to use in the definition of K-compl.
- the number of time-consuming operations (say flops) instead of the length
- in bits of the machine code of a program and thus consider computational
- time as a measure of complexity itself?
-
- Thanks
-
- Pedro Lima
- lima@ral.rpi.edu
-