home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!spool.mu.edu!agate!usenet.ins.cwru.edu!alpha.ces.cwru.edu!somos
- From: somos@ces.cwru.edu (Michael Somos)
- Newsgroups: sci.math
- Subject: Re: need proof: (1 + 1/n)^n ==> e
- Date: 25 Dec 1992 21:40:44 GMT
- Organization: Computer Engineering and Science, Case Western Reserve University
- Lines: 56
- Message-ID: <1hfv4sINNdlk@usenet.INS.CWRU.Edu>
- References: <1glr2qINN2vg@usenet.INS.CWRU.Edu>
- NNTP-Posting-Host: lychee.ces.cwru.edu
- Summary: truly elementary proof given
-
- The tone of the question suggests that a very elementary proof of
- the limit (1+1/n)^n is requested. It is no trick to come up with
- fancy proofs that obscure rather than clarify. An elementary proof
- may use a few more steps than a slick fancy proof but lead to more
- insight into the problem. While the previously posted proofs all
- have some merit, so far no one has mentioned the following approach.
-
- We consider lists of real numbers like x = [.34,.51,.12]. Define the
- operator P as the product of all the numbers incremented by one. So
-
- P [a,b,c] = (1+a)(1+b)(1+c) , for example.
-
- We now require all the numbers of the list to have the same sign.
- The reason is the following inequality
-
- P [a+b] = 1+a+b < 1+a+b+ab = (1+a)(1+b) = P [a,b] if ab>0 .
-
- We now define a partial ordering of lists by requiring that any
- splitting of a number is considered as being greater, for example,
-
- [a,b+c,d+e,f] < [a,b,c,d+e,f] < [a,b,c,d,e,f] .
-
- Note that two lists that have the same sum also have a common upper
- bound since we can keep splitting numbers of both to get a list of
- numbers which can be rejoined to get either of the original lists.
-
- This definition is designed to ensure that P is monotonic. Since
- (1+a)(1-a) = 1-a^2 < 1 , we now require that all the numbers of the
- list be less than one in absolute value. This ensures the following
-
- P(x) P(-x) = P(-x^2) < 1 , P(x) = P(-x^2)/P(-x) < 1/P(-x) .
-
- where -x is the list of all negatives of the numbers in x. But now
-
- if x < y , then -x < -y and P(x) < P(y) < 1/P(-y) < 1/P(-x) .
-
- All we need now is to show that an increasing sequence of lists,
- where the maximum absolute value of the numbers goes to zero, will
- converge to a limit when operated by P. A simple estimate is needed.
- Define the S operator as the sum of all the numbers of a list. To
- estimate how close we are approaching a limit use the simple
-
- [S(x)] < x , and 1 + S(x) = P [S(x)] < P(x) .
-
- But S(-x^2) approaches 0 by our last hypothesis, so the P(x) and
- the 1/P(-x) approach each other and are equal in the limit.
-
- Finally, consider the sequence of lists [1] , [1/2,1/2] , ... ,
- [1/n,1/n,...,1/n] , ... operated by P giving (1+1) , (1+1/2)^2 ,
- ... , (1+1/n)^n , ... . The limit exists and is defined to be e .
-
- Note that this approach also defines the exponential function at the
- same time. It uses nothing beyond the concept of limits and simple
- algebra. It is truly elementary although it uses more steps.
-
- Shalom, Michael Somos <somos@alpha.ces.cwru.edu> (My opinions alone)
-