[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference rusure::math

Title:Mathematics at DEC
Moderator:RUSURE::EDP
Created:Mon Feb 03 1986
Last Modified:Fri Jun 06 1997
Last Successful Update:Fri Jun 06 1997
Number of topics:2083
Total number of notes:14613

1907.0. "Lots of 1s only numbers" by EVTSG8::ESANU (Au temps pour moi) Wed Nov 16 1994 08:33

Eric's skeptical remark (1906.1) reminds me of the following
elementary problem:

Let  n  be a natural number prime with 10 (i.e. they have no common
divisor). Then there exists a multiple of  n  whose writing in base 10 
has at most  n  digits, all  = 1 .

Mihai.
T.RTitleUserPersonal
Name
DateLines
1907.1HANNAH::OSMANsee HANNAH::IGLOO$:[OSMAN]ERIC.VT240Thu Nov 17 1994 12:4020
So, for example, 7 being prime to 10 means some number of 1's, at most 7
of them, is a multiple of 7.

	1 nope
	11 nope
	111 um... nope
	1111 ... nope
	11111 ... nope
	111111 ... yes !  7*15873


I wonder which n require us to "go all the way".  7 didn't.  In other words,
we stopped at 6 7's and didn't have to try 7 7's.

3 goes all the way, since we have to go to 111.

How about 9.  It seems to require all 9 1's.

/Eric
1907.2How about 11? :-)EVMS::HALLYBFish have no concept of fireThu Nov 17 1994 21:290
1907.3WRKSYS::BRANDENBERGThu Nov 17 1994 21:384
>How about 9.  It seems to require all 9 1's.

Rule of nines.
1907.4True if n is relatively prime to 30BALZAC::QUENIVETMargins are often too small.Mon Nov 21 1994 05:1644
    Let phi be Euler's function, i.e. for each natural n > 0, the number of
    naturals relatively prime to n. Then for each natural n and each
    natural a relatively prime to n, we have:
    
      phi(n)
    a        = 1   mod n.
    
    This is a corollary of Lagrange's theorem which states that for each
    finite group of order m and each element a of the group:
      m 
    a    = 1.
    
    Apply to the multiplicative group of Z/nZ, the ring of integers modulo
    n, whose order is m = phi(n).
    
    Now, if we take a = 10 and n relatively prime to 10, we get:
    
      phi(n)                                   phi(n)
    10       = 1  mod n.  That is, n divides 10      - 1.
    
    Let m = phi(n). That means that:
    
    1000 ... 000 = 1 mod n (m 0's) or
    999 .... 999 = 0 mod n (m 9's)
    
    If in addition, n is relatively prime with 3, we can divide by 9 and we get:
    
    111 ... 111 = 0 mod n .
    
    The smallest number is given by:
    
    111 ... 111 = 0 mod n, with r 1's, where r is the order of 10 in the
    multiplicative group of Z/nZ. Hence, r is a divisor of m = phi(n).
    
    For example, 10 is of order 3 in (Z/37Z)*, the multiplicative group of
    integers mod 37. So: 37 divides 111.  111 = 3 x 37.
      
    If n is divisible by 3, well, I don't know the answer right now. 
    
    That was one of the questions I was asked when I graduated as a math
    teacher, some 17 years ago...
     
	Herv�
    
1907.5re .1 & re .4EVTSG8::ESANUAu temps pour moiWed Nov 23 1994 04:5767
My proof of .0: for  1 <= k <= n , we have

	 1...1   = q(k) * n + r(k)       with  0 <= r(k) < n
	k digits

Either one of the  r(k)  is  0 , or apply the box principle to the  n - 1
remaining possible values of the  r(k)  and consider that  n  is relatively
prime to  10 .


This proof also suggests the following generalization of .0: .0 does not
depend on the numerical base. More precisely:

	Let  n  be relatively prime to  r . Then there exists a multiple
	of  n  whose writing in base r  has at most  n  digits, all  = 1 .


The result

      phi(n)
    a        = 1   mod n     for  a   and  n  relatively prime

is known as the Euler or the Fermat-Euler Theorem. There are several
classical proofs for it; the first proof was given by Euler.


Eric's question (.1):

> I wonder which n require us to "go all the way".

As Eric puts it, to "go all the way" means that the smallest  k ,
1 <= k <= n , for which  n |  1...1   is  k = n .
                             k digits

Herve has proved (.4) that if  n  is relatively prime to  30 , then
n |     1...1     . As  1 <= phi(n) < n , this means that "we don't go
    phi(n) digits
all the way" for  n  if  n is relatively prime to  3  (or, equivalently,
to  9 ).

Transposing Herve's solution to the case of the base r : "we don't go all
the way" in base r  for numbers relatively prime to  (r - 1) * r .

Now, in base r , we do " go all the way" for  n = r - 1  (rule of nines
in base r ). What about the other  n  not relatively prime to  r - 1 ?

Let  p(i) ,  1 <= i <= s , be the divisors of  r - 1 . Take  n  relatively
prime to  r , but not relatively prime to  r - 1 ; then

	                         s        alpha(i)
	n = m * t ,  where  t =  PI   p(i)         ,
	                        i = 1

m  is relatively prime to  r - 1  and  alpha(i) <> 0  for at least
an  i  (1 <= i <= s).

We have  m |    1...1      and  t |  1...1   , where  1 <= x <= t .
            phi(m) digits           x digits

Consider the number  u =      1...1        . Then  m | u  and  t | u ,
                         x * phi(m) digits
hence   n | u , but  x * phi(m) < x * m = n, so "we don't go all the way"
in base r  for  n  if  n  has a divisor relatively prime to  r - 1 .

What about the  n  whose only divisors are those of  r - 1 ?

Mihai.