home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: sci.crypt
- Path: sparky!uunet!cs.utexas.edu!qt.cs.utexas.edu!yale.edu!cs.yale.edu!ginkgo.theory.cs.yale.edu!fischer
- From: fischer@ginkgo.theory.cs.yale.edu (Michael Fischer)
- Subject: Re: Math Induction (was Re: tabulation of primes)
- Message-ID: <1992Aug12.220058.15036@cs.yale.edu>
- Sender: news@cs.yale.edu (Usenet News)
- Nntp-Posting-Host: ginkgo.theory.cs.yale.edu
- Organization: Yale University Computer Science Dept., New Haven, CT 06520-2158
- X-Newsreader: Tin 1.1az (for az) PL4
- References: <1992Aug12.213221.8240@wdl.loral.com>
- Date: Wed, 12 Aug 1992 22:00:58 GMT
- Lines: 35
-
- mab@wdl39.wdl.loral.com (Mark A Biggar) writes:
- : I have seen two different definitions of induction:
- :
- : method 1
- : prove p(0)
- : prove (p(0) & p(1) & ... & p(n)) -> p(n+1)
- : therefore p(x) holds for all positive integers
- :
- : method 2
- : prove p(0)
- : prove p(n) -> p(n+1)
- : therefore p(X) holds etc.
- :
- : Are these of equal strength, in other words can I prove some things with 1
- : that I can't prove with just 2?
-
- They're equivalent. Obviously anything you can prove with 2 can also
- be proved with 1. The other way around, define the predicate
- q(n) = p(0) & p(1) & ... & p(n)
- Take your proof using 1, and where you prove
- (p(0) & p(1) & ... & p(n)) -> p(n+1)
- replace it instead by the following:
- Assume q(n).
- Hence, p(0) & p(1) & ... & p(n) holds.
- <include old proof of (p(0) & p(1) & ... & p(n)) -> p(n+1)>
- By modus ponens, conclude p(n+1).
- Hence, (p(0) & p(1) & ... & p(n)) & p(n+1) = q(n+1) holds.
- By deduction theorem,
- q(n) -> q(n+1)
- By method 2, q(x) holds for all positive integers x.
- Since q(x) -> p(x), conclude that p(x) holds for all positive integers x.
- --
- ==================================================
- | Michael Fischer <fischer-michael@cs.yale.edu> |
- ==================================================
-