IMO Shortlist 2013 A4

Algebra Level 4

Let n n be a positive integer, and consider a sequence a 1 , a 2 , , a n a_1, a_2, \ldots , a_n of positive integers. Extend it periodically to an infinite sequence a 1 , a 2 , a_1, a_2, \ldots by defining a n + i a i a_{n+i} \equiv a_i for all i 1 i\geq1 . If a 1 a 2 a n a 1 + b a_1 \leq a_2 \leq \cdots \leq a_n \leq a_1 + b and a a i n + i 1 a_{a_i} \leq n+i-1 for i = 1 , 2 , . , n i = 1, 2,\ldots., n then for n = 2013 n = 2013 what is the strongest upperbound on the sum of the first n n numbers in the sequence (which will always work).

Note: Translated title problem into one with a numerical answer.


The answer is 4052169.

This section requires Javascript.
You are seeing this because something didn't load right. We suggest you, (a) try refreshing the page, (b) enabling javascript if it is disabled on your browser and, finally, (c) loading the non-javascript version of this page . We're sorry about the hassle.

1 solution

Razzi Masroor
Mar 11, 2021

I will show the general the answer to be n 2 n^2 . When n = 2013 n=2013 this gives 4052169.

We will prove that a 1 + a 2 + . . . + a n n 2 a_1+a_2+...+a_n\leq n^2 by contradiction. First of all, if a 1 + a 2 + . . . + a n > n 2 a_1+a_2+...+a_n> n^2 then one of the n n terms must be greater than n n . Let a k a_k be the first term which satisfies this property. We will show that k > 1 k>1 . Using the fact that a 1 a 2 . . . a n a_1\leq a_2\leq ... \leq a_n , n = n + 1 1 a a 1 a 1 n = n+1-1 \geq a_{a_1} \geq a_1 . If a 1 = n a_1 = n then a n = a a 1 n a_n = a_{a_1} \leq n and because a n a 1 a_n \geq a_1 , a n = n a_n=n . Since a 1 a i a n a_1\leq a_i \leq a_n for all i i , we must have that a i = n a_i = n . Thus a 1 + a 2 + . . . + a n = n 2 a_1+a_2+...+a_n = n^2 , a contradiction. This means that a 1 < n a_1 < n so k > 1 k > 1 .

Now by the definition of k k , a 1 , a 2 , . . . , a k 1 n a_1, a_2, ..., a_{k-1} \leq n and a k , a k + 1 , . . . , a n a k n + 1 a_k, a_{k+1}, ..., a_n \geq a_k \geq n+1 . Since a a 1 n + 1 1 = n a_{a_1} \leq n+1-1 = n , we know that a 1 a_1 must have a remainder of at most k 1 k-1 when divided by n n . a 1 n a_1 \leq n so we can go further and state that a 1 a_1 is at most k 1 k-1 . Now let a i a_i be the first term so that a i k a_i \geq k . Now a 1 , a 2 , . . . , a i 1 k 1 a_1, a_2, ..., a_{i-1} \leq k-1 , k a i , a i + 1 , . . . , a k 1 n k \leq a_i, a_{i+1}, ..., a_{k-1} \leq n , and n + k 1 n + a 1 a n , a n 1 , . . . , a k n + 1 n+k-1 \geq n+a_1\geq a_n, a_{n-1}, ..., a_k \geq n+1 .

Now i k i\leq k since a k > n k a_k>n\geq k . If i = k i=k then our first k 1 k-1 terms are at most k 1 k-1 and the remaining n k + 1 n-k+1 terms are at most a 1 + n n + k 1 a_1+n \leq n+k-1 , giving an upper bound of ( n ( k 1 ) ) ( n + ( k 1 ) ) + ( k 1 ) 2 = n 2 (n-(k-1))(n+(k-1))+(k-1)^2 = n^2 on the sum of the terms. This is impossible since we assumed the sum to be greater than n 2 n^2 . Thus i < k i<k and i > 1 i>1 since a 1 k 1 a_1\leq k-1 .

First of all the first i 1 i-1 terms are at most k 1 k-1 and the i i to k 1 k-1 -th terms are just equal to themselves. Now the k k through a i a_i -th terms are all at most a a i n + i 1 a_{a_i} \leq n+i-1 (the a i a_i -th term is exactly a a i a_{a_i} ). Next the a i + 1 a_i+1 through a i + 1 a_{i+1} -th terms are at most a a i + 1 n + i a_{a_{i+1}} \leq n+i . This repeats until we get that the a k 2 + 1 a_{k-2}+1 through a k 1 a_{k-1} -th terms are at most a a k 1 n + k 2 a_{a_{k-1}} \leq n+k-2 . Finally, the a k 1 + 1 a_{k-1}+1 through the n n -th terms are at most a 1 + n n + k 1 a_1+n \leq n+k-1 . Thus the sum of all n n terms is at most ( i 1 ) ( k 1 ) + l = i l = k 1 a l + ( a i ( k 1 ) ) ( n + ( i 1 ) ) + l = i l = k 2 ( a l + 1 a l ) ( n + l ) + ( n a k 1 ) ( n + ( k 1 ) ) (i-1)(k-1)+\sum_{l=i}^{l=k-1}{a_l}+(a_i-(k-1))(n+(i-1))+\sum_{l = i}^{l=k-2}{(a_{l+1}-a_l)(n+l)}+(n-a_{k-1})(n+(k-1)) Now ( a i ( k 1 ) ) ( n + ( i 1 ) ) + l = i l = k 2 ( a l + 1 a l ) ( n + l ) + ( n a k 1 ) ( n + ( k 1 ) ) (a_i-(k-1))(n+(i-1))+\sum_{l = i}^{l=k-2}{(a_{l+1}-a_l)(n+l)}+(n-a_{k-1})(n+(k-1)) telescopes into n ( n + ( k 1 ) ) l = i l = k 1 a l ( k 1 ) ( n + ( i 1 ) ) n(n+(k-1))-\sum_{l=i}^{l=k-1}{a_l}-(k-1)(n+(i-1)) so we can use this substitution to reduce our original value down to ( i 1 ) ( k 1 ) + l = i l = k 1 a l + n ( n + ( k 1 ) ) l = i l = k 1 a l ( k 1 ) ( n + ( i 1 ) ) (i-1)(k-1)+\sum_{l=i}^{l=k-1}{a_l}+n(n+(k-1))-\sum_{l=i}^{l=k-1}{a_l}-(k-1)(n+(i-1)) ( i 1 ) ( k 1 ) + n ( n + ( k 1 ) ) ( k 1 ) ( n + ( i 1 ) ) (i-1)(k-1)+n(n+(k-1))-(k-1)(n+(i-1)) With some expanding this can be compressed into ( i 1 ) ( k 1 ) + n ( n + ( k 1 ) ) ( k 1 ) ( n + ( i 1 ) ) = ( i 1 ) ( k 1 ) + n 2 + n ( k 1 ) ( k 1 ) n ( k 1 ) ( i 1 ) = n 2 (i-1)(k-1)+n(n+(k-1))-(k-1)(n+(i-1))= (i-1)(k-1)+n^2+n(k-1)-(k-1)n-(k-1)(i-1)=n^2 Thus the sum of all of the terms is at most n 2 n^2 , a contradiction. This completes our proof so a 1 + a 2 + . . . + a n n 2 a_1+a_2+...+a_n\leq n^2 . Q.E.D.

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...