Happy Numbers

Date: 06/21/98 at 08:14:53
From: Catherine Jarman
Subject: Happy numbers

Do you have any information on happy numbers? A happy number is a 
positive integer for which the sum of the squares of the digits 
eventually ends in 1.

For example, 13 is happy since 1^2 + 3^2 = 10 and 1^2 + 0^2 = 1. 

Date: 06/24/98 at 17:39:58
From: Doctor Hauke
Subject: Re: Happy numbers

Hi Catherine,

1. Take a number n.
2. Dissect it into digits.
3. Square them all and add them up
4. You get a new number m.
5. If m = 1, n is happy; otherwise set n = m and repeat at 1.
6. If you run into a loop, n is not a happy number.

   a) Pick a number, as gigantic as you like.

   b) In the first iteration, it will give maximally the number 
      9^2+9^2+9^2+... - even a 10-digit number will give a three-digit 
      number after iteration 1. (Can you check that?)

   c) In the second iteration, all three-digit numbers can turn 
      maximally into 9^2+9^2+9^2 = 243. 

      Nines fatten the result, so the biggest number in the next 
      iteration will be produced by 199, giving 1^2+9^2+9^2 = 163. 

      You have to check these 163 numbers by hand.

      Of course I'm lazy and wrote a FORTRAN program that said:

      - only 52 of these numbers don't get smaller in one step
      - only 22 in two steps
      - and so on. 

      After five iterations, the computer says only 4 and 5 must be 
      checked further, and there is probably exactly one unhappy loop.

I highly recommend trying it on your own :-)

- Doctor Hauke, The Math Forum
  http://mathforum.org/dr.math/   

Date: 11/19/98 at 00:38:51
From: Jeremy
Subject: A previous Dr. Math question...

I don't understand the logic you used here. After the second iteration, 
your number can still be unbounded. And from what I have found, 4 is 
not a happy number, which means 2 is not a happy number either. Nor 
is any number created in the "4-loop," or any number there with zeros, 
at will, placed between or after the digits. I don't understand your 
conclusion.

Date: 11/23/98 at 09:56:31
From: Doctor Rob
Subject: Re: A previous Dr. Math question...

The logic is as follows. You want to find the smallest number on every
loop. If any number leads to a smaller number after n steps, it cannot 
be the smallest number in any loop. (You don't care whether it is on a 
loop or not.)  That way you can eliminate many, many numbers after 
just a few steps.

In particular, numbers n with e digits, that is, with 
10^(e-1) <= n < 10^e, lead in one step to numbers at most e*9^2 = 81*e. 
For e >= 4, this makes them smaller. Thus all numbers with at least 
4 digits cannot be the smallest number on a loop. That means that you 
only have to check numbers smaller than 1000 to see if they are the 
smallest number in a loop. It also implies that there is at most a 
finite number of loops.

After most of the numbers are eliminated, you can test the rest rather 
quickly. Then you find that there is the 4-loop 4, 16, 37, 58, 89, 
145, 42, 20, 4, and that 5 leads into that loop via 5, 25, 29, 85, 89, 
.... That is the loop into which all unhappy numbers lead. There is 
also the 1-loop 1,1, ...,  into which all happy numbers lead.

- Doctor Rob, The Math Forum
  http://mathforum.org/dr.math/