Sum Product Conjecture
[size=500][center]The Sum Product Conjecture[/center][/size]
Step 1: Pick a number . Step 2: Let be the sum of the digits of . Step 3: Find the solution to such that and is as close to as possible. Now your number is .
Repeat.
[b]The Sum-Product Conjecture[/b] states that you will eventually end up at 0 or 4. [list] [*][b]Claim 1:[/b] Going through all these steps, you will eventually end up at a one-digit integer. First, we prove that we will eventually reach a number with less than four digits, and then we will prove the for a number less than or equal to 4 digits, that it will eventually reach a previous case and onto the 1 digit cases, which is proved below. [color=#f00][b]Lemma.[/b][/color] For all , .
[i]Proof by djmathman:[/i] Proceed by induction. The case can be computed by hand, since . For the inductive step, write \[ \tfrac{81}4(k+1)^2 = \tfrac{81}4k^2\cdot(\tfrac{k+1}k)^2 \stackrel{\textrm{(IH)}}<10^{k-1}\cdot 2^2\leq 10^k, \] where in the second-to-last equality we use the fact that for all positive integers .
To proceed, observe now that, by the AM-GM inequality, \[ ab\leq \left(\frac{a+b}2\right)^2= \frac{S(n)^2}4. \] We now split into cases. [list] [*][b]Case 1.[/b] If has two digits, then , so in accordance with your notation . Using this value as our new value of , we get , so ; then , so ; then , so ; then , so ; then , so .
Unfortunately, we can't press on using this logic anymore, so we need to brute-force this part: [list] [*] For , is at most , so is a single-digit integer. [*] For , , so . Then we're in the previous case. [*] For , , so . Then we're in the previous case. [*] For , , so . Then we're in the very first case. [*] For , , so . Then we're in the second case. [*] For , , so . Then we're in the very first case. [/list] [*][b]Case 2.[/b] Suppose has three digits. Then , so . In turn, using as our new value of , , so . Now we're in Case 1. [*][b]Case 3.[/b] Finally, suppose has at least digits. Choose such that , noting that . Then is at most (when all digits are nines), and so \[ ab\leq \frac{S(n)^2}4 \leq \frac{(9k)^2}4 = \frac{81}4k^2 \stackrel{(*)}< 10^{k-1} \leq n. \][/list] This means that the process strictly decreases the value of . Hence, after a finite number of iterations, our integer will have fewer than digits, which reduces the problem to one of the previous few cases. [*][b]Claim 2:[/b] If your number is a one-digit integer, then you will end up 0 or 4. [list] [*][b]Proof: The sum of the digits of one-digit integers is itself.[/b] [list] [*]Case 1: 1. We have , so that and (symmetry, and is the same), so that . We have ended at 0, and we repeat, finding the and , so that we end up at again. [*]Case 2: 2. We have that , and and , and . We already proved it for 1, so this is done [*]Case 3: 3. We have that , so that , , but we already proved it for 2, and we are done. [*]Case 4: 4. We have that , so that , and we end up at again. [*]Case 5: 5. We have that , and , so that . Now, we have , , so that . Now, we have , , and . , we already proved it for 2, and we are done. This step also proves it for and because we went through 6 at one step. [*]Case 6: 6. We are done because of Case 5. [*]Case 7: 7. We have , , so that . Sum of digits is 3, and we already proved it for 3, so we are done. [*]Case 8: 8. We have that , and , so that . Sum of digits is , but we already proved it for , so we are done. [*]Case 9: 9. We are done because of Case 5. [/list] [/list] [/list] We have proved this claim. Q.E.D.