I can't see how a computer, without some analysis, would help here - the numbers seem too big - but maybe I'm missing something...
Assume that M = 2m where m is an integer (*Justification later).
Then: 1997M - 1 can be written as 19972^m - 1 and factorised as follows:
(1997-1)(1997+1)(19972+1)(19974+1)(19978+1)......(19972^(m-1)+1) (1)
[replacing the first two brackets by a single bracket and then repeating the
process reduces the complete product (1) to the required 19972^m - 1]
Now, the first bracket in (1) contains the factor 22 (1996 = 22 * 449), while
the remaining m brackets each contain the factor 2 to power 1 only. This is true
since each such bracket is of the form (4n + 1)2^r + 1, which can be reduced
using the binomial theorem to the form 4n + 2 = 2(2n + 1).
Thus (1) contains 2m+2, and so to be divisible by 21998, we need m + 2 = 1998,
giving m = 1996, and M = 21996, as the smallest possible value of M.
_________________________
* If M is not a power of 2, but lies between 2m and 2m+1, then it can be written
in the form M = 2a + 2b + 2c +...+ 2m where a < b < c ...< m.
Now, writing A = 19972^a - 1, B = 19972^b - 1 ..., it follows
that 1997M - 1 = (19972^a 19972^b 19972^c.......) - 1
= [(A + 1)(B + 1)(C + 1)....] - 1 (2)
From the initial theory, A contains a factor of 2a+2, B contains 2b+2 etc, so when
(2) is expanded (and the '1's drop out), the term containing the lowest
power of 2 is A. Thus 2a+2 is the highest power of 2 that divides (2), and this
shows that 1997M - 1 would contain a smaller power of 2 than 19972^m - 1,
and so justifies the assumption that M = 2m gives the minimum value for M.
Edited on January 5, 2011, 6:55 pm
|
Posted by Harry
on 2011-01-04 10:50:21 |