home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: sci.math
- Path: sparky!uunet!haven.umd.edu!darwin.sura.net!wupost!uwm.edu!psuvax1!news
- From: plu@math.psu.edu (Todd Andrew Simpson)
- Subject: Re: Catalan NUmbers
- In-Reply-To: thanasis@cs.columbia.edu's message of Wed, 7 Oct 1992 16:42:59 GMT
- Message-ID: <BvrvJJ.HqG@cs.psu.edu>
- Sender: news@cs.psu.edu (Usenet)
- Nntp-Posting-Host: hilbert.math.psu.edu
- Organization: Department of Mathematics, Pennsylvania State University
- References: <BvrFrn.53A@cs.columbia.edu>
- Date: Wed, 7 Oct 1992 22:23:42 GMT
- Lines: 39
-
- In article <BvrFrn.53A@cs.columbia.edu> thanasis@cs.columbia.edu (Thanasis Tsantilas) writes:
-
- Is there a combinatorial ("bijective") way to prove that the
- Catalan numbers are equal to C(2n,n)/(n+1)?
- The answer looks simple, so I thought there must be something.
- I'm only aware of the generating function solution of the recurrence.
- Any pointers to books/articles are appreciated.
-
- There is a very elegant proof due to D. Andr\'e sometime in the 19th
- century: First, C(2n,n)/(n+1) = C(2n,n)-C(2n,n-1). Then:
-
- The number of paths from (0,0) to (n,n) in the plane, with unit steps
- (1,0) and (0,1), is C(2n,n).
-
- The Catalan number C_n is the number of these paths that stay in the
- region {(x,y): y<=x}.
-
- The number of paths from (0,0) to (n+1,n-1) with unit steps (1,0) and
- (0,1) is C(2n,n-1).
-
- There is a bijection between paths (with unit steps (1,0) and (0,1))
- from (0,0) to (n+1,n-1) and those from (0,0) to (n,n) that do not stay in
- the region {(x,y): y<=x}: given the latter type of path, with the k-th
- step being the first to go into the region y>x, for i from 1 to k if the
- i-th step is (1,0) make it (0,1); if it is (0,1) make it (1,0).
-
- This is described in Knuth's book, The Art of Computer Programming,
- Volume I (which is where I found it). Another approach might be to
- associate each of a set of C_n objects (there are lots of sets
- enumerated by the C_n) with paths from (0,0) to (n,n) in such a way
- that there are n+1 paths for each of the objects. I haven't seen such
- a proof, though it'd be interesting since one would not need to know
- _a priori_ that C(2n,n)/(n+1) was equal to C(2n,n)-C(2n,n-1).
-
- Thanasis Tsantilas
- Columbia University
- thanasis@cs.columbia.edu
-
- - Todd Andrew Simpson
-