home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!mcsun!sunic!dkuug!diku!torbenm
- From: torbenm@diku.dk (Torben AEgidius Mogensen)
- Newsgroups: sci.math
- Subject: Re: Looking for a slick proof
- Message-ID: <1992Oct9.155336.3455@odin.diku.dk>
- Date: 9 Oct 92 15:53:36 GMT
- References: <1aterkINNj4f@matt.ksu.ksu.edu> <1992Oct9.92304.1850@ms.uky.edu>
- Sender: torbenm@freke.diku.dk
- Organization: Department of Computer Science, U of Copenhagen
- Lines: 40
-
- cyeomans@ms.uky.edu (Charles Yeomans) writes:
-
- >In article <1aterkINNj4f@matt.ksu.ksu.edu> bubai@matt.ksu.ksu.edu (P.Chatterjee) writes:
- >>Hi,
- >>
- >>How does one prove that 2^n divides n*(n+1)*(n+2)*......*(2n) without
- >>having to resort to the 'odd' and 'even' induction? Any ideas?
-
- >It looks to me like half of those 2n numbers are even.
-
- Yes, but that only proves that 2^(n/2) divides the product. However,
- you can use that 4 divides every fourth number etc. Thus we have
-
- (n+1) div 2 numbers where 2 is a divisor
- (n+1) div 4 numbers where 4 is a divisor
- (n+1) div 8 numbers where 8 is a divisor
- etc.
-
- All divisions round down. The numbers that divide by 2 also divide by
- 4 etc., so we get only one power of two for each number in each list.
- Now,
-
- n <= (n+1) div 2 + (n+1) div 4 + (n+1) div 8 + ... + (n+1) div log2(n)
- <= n+1-log2(n+1)
-
- so 2^n divides the product above.
-
- In order to see that the sum is at least n, consider how we round the
- fractions. In the worst case, we round every fraction down, so we lose
-
- 1/2 + 1/4 + 1/8 + ... 1/log2(n+1) = 1-log2(n+1)
-
- from the best case, where we always have exact divisions (when (n+1)
- is a power of 2), so at worst we get
-
- n+1-log2(n+1) - (1-log2(n+1)) = n
-
- Q.E.D.
-
- Torben Mogensen (torbenm@diku.dk)
-