I made a quick experiment with other bases.
3-digits numbers, base 8 has fixed point 374_8=252_10. 252 `mod` 7 = 0.
base-9, base-11 and base 13 seem do not have fixed points.
base-12, 3 digits has (at least one) fixed point 5b6_12 = 858_10. 858 `mod` 11 = 0.
I think, it exploits some properties of positional notation of numbers so that fixed points all divisible by (base-1).
Kaprekar's operation always results in a multiple of 9.
Given a > b > c > d. Kaprekar's operation gives:
1000a + 100b + 10c + d - (1000d + 100c + 10b + a) =999(a - d) + 90(b - c) =9(111(a - d) + 10(b - c))
Sum(i=0..n, a[i]*10^i)
A power of 10 (10^n) can be re written as
(1 + 9 * Sum(i=0..n-1, 10^i).
When you subtract two numbers with the same digits, you end up being able to factor a nine out of these sums fairly easily.
I wrote a proof of the number - reverse(number) a while back. It can be found on archive.org
http://web.archive.org/web/20050314023901/http://jimfl.tense...
I made a quick experiment with other bases.
3-digits numbers, base 8 has fixed point 374_8=252_10. 252 `mod` 7 = 0.
base-9, base-11 and base 13 seem do not have fixed points.
base-12, 3 digits has (at least one) fixed point 5b6_12 = 858_10. 858 `mod` 11 = 0.
I think, it exploits some properties of positional notation of numbers so that fixed points all divisible by (base-1).