home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!usc!zaphod.mps.ohio-state.edu!saimiri.primate.wisc.edu!caen!destroyer!cs.ubc.ca!unixg.ubc.ca!unixg.ubc.ca!israel
- From: israel@unixg.ubc.ca (Robert B. Israel)
- Newsgroups: sci.math
- Subject: Re: 1+1/2+1/3+...+1/n > x (SOLUTION)
- Date: 20 Nov 92 22:39:36 GMT
- Organization: The University of British Columbia
- Lines: 55
- Message-ID: <israel.722299176@unixg.ubc.ca>
- References: <1992Nov20.101643.17628@husc15.harvard.edu>
- NNTP-Posting-Host: unixg.ubc.ca
-
- In <1992Nov20.101643.17628@husc15.harvard.edu> blom@husc15.harvard.edu writes:
-
- > n 1 1 inf BERNOULLI (k)
- > SUM --- = LN (n) + euler_gamma + --- - SUM ---------------
- > k=1 k 2 n k=2 k
- > k (n)
-
-
-
- >By expanding the above sequence, we can find the sum of the first n
- >reciprocals quickly and easily. My computer gives anomalous results
- >for n in the range in which the sum > 100, but maybe someone else could
- >use this formula to verify the exact value for which we are looking.
- >The greatest integer less than
-
- > 100 - euler_gamma
- > e
-
- >is a good estimate, but it might be a little off.
-
- Actually, you don't need to be quite so sophisticated.
-
- The answer is N=15092688622113788323693563264538101449859497 which is the
- integer part of exp(100 - gamma).
-
- Let h_N = sum_{n=1}^N 1/n.
- Note that h_N_ - ln(N+1) = sum_{n=1}^N (1/n - ln((n+1)/n)).
- The limit as N -> infinity is Euler's constant gamma. So
- h_N - ln(N+1) - gamma = - sum_{n=N+1}^infinity (1/n - ln((n+1)/n)).
- We can expand f(t) = 1/t - ln((t+1)/t) = 1/(2 t^2) - 1/(3 t^3) + ...,
- an alternating series (for t > 1). So 1/(2 t^2) - 1/(3 t^3) < f(t) < 1/(2 t^2),
- and h_N - ln(N+1) - gamma < - int_N^infinity (1/(2 t^2)) dt = - 1/(2 N)
- while h_N - ln(N+1) - gamma > - int_(N+1)^infinity (1/(2 t^2) - 1/(3 t^3) dt
- = - 1/(2(N+1)) + 1/(6(N+1)^2).
- So, if u(N) = ln(N+1) + gamma - 1/(2N)
- and v(N) = ln(N+1) + gamma - 1/(2(N+1)) + 1/(6(N+1)^2)
- we want u(N-1) < 100 (to ensure h_(N-1) < 100) and v(N) > 100
- (to ensure h_N > 100). Since the difference between u(N) and v(N) is
- of order 1/N^2, we have an excellent chance that this will be true for
- some N.
-
- Now, I checked using both Maple and Mathematica:
-
- exp(100 - gamma) = 15092688622113788323693563264538101449859497.30...
- Taking N as the integer part of this,
- u(N-1) = 99.999999999999999999999999999999999999999999942747
- v(N) = 100.00000000000000000000000000000000000000000000900
-
-
-
- --
- Robert Israel israel@math.ubc.ca
- Department of Mathematics or israel@unixg.ubc.ca
- University of British Columbia
- Vancouver, BC, Canada V6T 1Y4
-