Browse · MathNet
PrintSaudi Arabian IMO Booklet
Saudi Arabia algebra
Problem
Let the sequence be defined in the following way: and inductively . Prove that the sequence contains infinitely many perfect squares.
Solution
Let us analyze the sequence defined by and .
Suppose at some step is a perfect square, say for some integer . Then:
Now, consider the next terms: But , and is between and . Specifically, for , so . Therefore:
Now, But , so . Since , , and (since for ), so . Thus:
But let's look for the next perfect square. Let us try to see if the sequence can reach . From above, , .
Let us try to generalize. For large , increases slowly compared to , so the sequence grows slowly.
But more importantly, we can show that for any starting , the sequence will eventually reach a perfect square, and then the above process repeats.
Let us prove that for any , there exists such that is a perfect square.
Let be arbitrary. Let . Then .
If , we are done. If not, for .
At each step, . Since , , so .
Let us consider the sequence of modulo :
Let us try to reach .
From , after steps, (since for ).
We want . So:
Since , is positive and less than . So is an integer if and only if divides .
But even if is not integer, after steps, will reach or exceed , at which point .
Therefore, after a bounded number of steps, the sequence will cross from to .
Now, consider the difference . For some , will be exactly if is congruent to modulo .
But even if not, after passing , the process repeats with replaced by .
Therefore, for any starting , the sequence will eventually hit a perfect square, and then the process repeats. Thus, the sequence contains infinitely many perfect squares.
Therefore, the sequence contains infinitely many perfect squares.
Suppose at some step is a perfect square, say for some integer . Then:
Now, consider the next terms: But , and is between and . Specifically, for , so . Therefore:
Now, But , so . Since , , and (since for ), so . Thus:
But let's look for the next perfect square. Let us try to see if the sequence can reach . From above, , .
Let us try to generalize. For large , increases slowly compared to , so the sequence grows slowly.
But more importantly, we can show that for any starting , the sequence will eventually reach a perfect square, and then the above process repeats.
Let us prove that for any , there exists such that is a perfect square.
Let be arbitrary. Let . Then .
If , we are done. If not, for .
At each step, . Since , , so .
Let us consider the sequence of modulo :
Let us try to reach .
From , after steps, (since for ).
We want . So:
Since , is positive and less than . So is an integer if and only if divides .
But even if is not integer, after steps, will reach or exceed , at which point .
Therefore, after a bounded number of steps, the sequence will cross from to .
Now, consider the difference . For some , will be exactly if is congruent to modulo .
But even if not, after passing , the process repeats with replaced by .
Therefore, for any starting , the sequence will eventually hit a perfect square, and then the process repeats. Thus, the sequence contains infinitely many perfect squares.
Therefore, the sequence contains infinitely many perfect squares.
Techniques
Recurrence relationsFloors and ceilings