Fibonacci-Factors Polynomial

Suppose f f is a polynomial with integer coefficients, such that for all positive integers n n the n n -th Fibonacci number u n u_n divides f ( u n + 1 ) f(u_{n+1}) . Find the smallest possible positive value of f ( 4 ) f(4) .

Details and assumptions

The Fibonacci numbers are defined as follows:

u 1 = 1 , u 2 = 1 ; u_1=1,\ u_2=1; u n = u n 1 + u n 2 u_n=u_{n-1}+u_{n-2} for all n 3 n\geq 3 .

0 is not a positive number.


The answer is 255.

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.

2 solutions

Let f ( x ) = k = 0 n a k x k f(x)= \sum \limits_{k=0}^{n} a_k x^k . From definition, we have f ( u n + 1 ) 0 ( m o d u n ) f(u_{n+1}) \equiv 0 \pmod{u_n} f ( u n + u n 1 ) 0 ( m o d u n ) \implies f(u_{n} + u_{n-1}) \equiv 0 \pmod{u_n} k = 0 n a k ( u n + u n 1 ) k 0 ( m o d u n ) \implies \sum \limits_{k=0}^{n} a_k(u_n + u_{n-1})^k \equiv 0 \pmod{u_n} k = 0 n a k ( m = 0 k ( k m ) ( u n ) m ( u n 1 ) k m ) 0 ( m o d u n ) \implies \sum \limits_{k=0}^{n} a_k \left ( \sum \limits_{m=0}^{k} \binom{k}{m}(u_n)^m (u_{n-1})^{k-m} \right ) \equiv 0 \pmod{u_n} Where the last step follows from the binomial expansion of each of the terms. Since we are considering the expression modulo u n u_n , we can ignore all terms which contain a positive power of u n u_n . In general, from the expansion of ( u n + u n 1 ) k (u_n+u_{n-1})^k , the only term that gets selected is ( k 0 ) u n 1 = u n 1 \binom{k}{0} u_{n-1}= u_{n-1} . Continuing, k = 0 n a k ( u n 1 ) k 0 ( m o d u n ) \sum \limits_{k=0}^{n} a_k (u_{n-1})^k \equiv 0 \pmod{u_n} Notice that this is simply f ( u n 1 ) f(u_{n-1}) , so f ( u n 1 ) 0 ( m o d u n ) f(u_{n-1}) \equiv 0 \pmod{u_n} . Substituting n + 1 n+1 in place of n n , we find out that u n + 1 f ( u n ) u_{n+1} \mid f(u_n) .

Now, if n n is even, gcd ( n + 1 , n 1 ) = 1 \gcd (n+1, n-1)= 1 , so gcd ( u n + 1 , u n 1 ) = u gcd ( n + 1 , n 1 ) = u 1 = 1 \gcd (u_{n+1}, u_{n-1})= u_{\gcd (n+1, n-1)}= u_1 = 1 If n n is odd, gcd ( n + 1 , n 1 ) = 2 \gcd (n+1, n-1)= 2 , so gcd ( u n + 1 , u n 1 ) = u gcd ( n + 1 , n 1 ) = u 2 = 1 \gcd (u_{n+1}, u_{n-1}) = u_{\gcd (n+1, n-1)} = u_2= 1 Either way u n + 1 u_{n+1} and u n 1 u_{n-1} are always coprime. Since both of them divide f ( u n ) f(u_n) , it immediately follows that u n + 1 u n 1 f ( u n ) u_{n+1}u_{n-1} \mid f(u_n) .

It is known that for all n N n \in \mathbb{N} , u n + 1 u n 1 = ( u n ) 2 + ( 1 ) n u_{n+1}u_{n-1}= (u_n)^2 + (-1)^n . The proof is trivial by induction.

We then conclude that ( u n ) 2 + ( 1 ) n f ( u n ) (u_n)^2+(-1)^n \mid f(u_n) . If n n is even, ( u n ) 2 + 1 f ( u n ) (u_n)^2+1 \mid f(u_n) . If n n is odd, ( u n ) 2 1 f ( u n ) (u_n)^2-1 \mid f(u_n) .

We claim that x 2 + 1 f ( x ) x^2+1 \mid f(x) . To prove this, we note that when f ( x ) f(x) is divided by x 2 + 1 x^2+1 , we shall get a polynomial P ( x ) P(x) which is either linear or constant. If it is constant, it must be 0 0 , which immediately follows by plugging x = u n x= u_n for any even n n . If it is linear, note that P ( u n ) 0 ( m o d ( ( u n ) 2 + 1 ) ) P(u_n) \equiv 0 \pmod{((u_n)^2+1)} for all even n n . However, since P ( u n ) P(u_n) is a linear and ( u n ) 2 + 1 (u_n)^2+1 is a quadratic in u n u_n , for some sufficiently large n n , ( u n ) 2 + 1 > P ( u n ) (u_n)^2+1>P(u_n) , a contradiction.

Similarly, we find out x 2 1 f ( x ) x^2 - 1 \mid f(x) . Noting that gcd ( x 2 + 1 , x 2 1 ) = 1 \gcd (x^2 + 1, x^2-1)= 1 , we combine them into x 4 1 f ( x ) x^4-1 \mid f(x) . The value of f ( x ) f(x) which returns the smallest possible positive integer value of x x is given by x 4 1 x^4-1 . Our desired answer is 4 4 1 = 255 4^4-1= \boxed{255} .

A few clarifications

  • We used the identity gcd ( u a , u b ) = u gcd ( a , b ) ( a , b ) N 2 \gcd (u_a, u_b)= u_{\gcd (a,b)} \ \forall \ (a, b) \in \mathbb{N}^2 when we showed that u n + 1 u_{n+1} and u n 1 u_{n-1} are always coprime. The proof can be found here .

  • In order to make the proof short, I skipped the proof of the identity u n + 1 u n 1 = ( u n ) 2 + ( 1 ) n u_{n+1}u_{n-1}= (u_n)^2 + (-1)^n . This can be proved by induction. For the case case, we note that u 3 u 1 = 2 × 1 = ( u 2 ) 2 + ( 1 ) 2 u_3u_1= 2 \times 1 = (u_2)^2 + (-1)^2 , so our proposition is true for n = 2 n=2 . Assume it is true for n = k n=k , so that u k + 1 u k 1 = ( u k ) 2 + ( 1 ) k u_{k+1}u_{k-1}= (u_k)^2+(-1)^k . Then, note that u k u k + 2 = u k ( u k + u k + 1 ) = ( u k ) 2 + u k u k + 1 = u k + 1 u k + 1 + u k u k + 1 ( 1 ) n u_k u_{k+2} = u_k (u_k + u_{k+1})= (u_k)^2 + u_k u_{k+1} = u_{k+1}u_{k+1} + u_ku_{k+1} -(-1)^n = u k + 1 ( u k + u k + 1 ) + ( 1 ) k + 1 = u k + 1 u k + 3 + ( 1 ) k + 1 = u_{k+1}(u_k + u_{k+1}) + (-1)^{k+1} = u_{k+1}u_{k+3} + (-1)^{k+1} This completes our induction step, and hence our assumption is true for all n N n \in \mathbb{N}

  • Here's why we can choose a n n such that ( u n ) 2 + 1 (u_n)^2+1 is greater than P ( x ) P(x) for all linear polynomials P ( x ) P(x) . Let P ( x ) = m x + c P(x)= mx+c . We can simply solve the inequality x 2 + 1 > m x + c x^2 + 1 > mx+c x > 1 2 ( 4 c + m 2 4 + m ) \implies x > \frac{1}{2} \left ( \sqrt{4c+m^2-4} + m \right ) We can simply choose any n n such that u n u_n exceeds this value. In general, any higher degree polynomial with positive coefficients grows faster than any lower degree polynomial with positive coefficients.

Lars McGee
May 20, 2014

Lemma 1: u n 1 u n + 1 = u n 2 + ( 1 ) n u_{n-1}u_{n+1} = u_n^2 + (-1)^n for all integers n > 1 n > 1 .

Proof: Proceed by induction. As the base case, let n = 2 n = 2 , so that u n 1 u n + 1 = u 1 u 3 = 1 2 = 2 = 1 2 + ( 1 ) 2 = u 2 2 + ( 1 ) 2 = u n 2 + ( 1 ) n u_{n-1}u_{n+1} = u_1u_3 = 1\cdot 2 = 2 = 1^2 + (-1)^2 = u_2^2 + (-1)^2 = u_n^2 + (-1)^n .

Assume for the inductive step that u n 1 u n + 1 = u n 2 + ( 1 ) n u_{n-1}u_{n+1} = u_n^2 + (-1)^n . Then u n u n + 2 = u n ( u n + 1 + u n ) u_nu_{n+2} = u_n(u_{n+1}+u_n) = u n u n + 1 + u n 2 = u_nu_{n+1} + u_n^2 = u n u n + 1 + u n 1 u n + 1 ( 1 ) n = u_nu_{n+1} + u_{n-1}u_{n+1} - (-1)^n = u n + 1 ( u n + u n 1 ) + ( 1 ) n + 1 = u_{n+1}(u_n + u_{n-1}) + (-1)^{n+1} = u n + 1 2 + ( 1 ) n + 1 = u_{n+1}^2 + (-1)^{n+1} . \blacksquare

Lemma 2: u n 1 u_{n-1} and u n + 1 u_{n+1} are relatively prime.

Proof: Suppose that a natural number divides u n 1 u_{n-1} and u n + 1 u_{n+1} . Then it divides u n + 1 u n 1 = u n u_{n+1} - u_{n-1} = u_n , and as a consequence of the recursion divides all Fibonacci numbers including u 1 = 1 u_1 = 1 . \blacksquare

We know that u n f ( u n + 1 ) u_n | f(u_{n+1}) for all n, so f ( u n 1 ) f ( u n + u n 1 ) = f ( u n + 1 ) 0 m o d u n f(u_{n-1}) \equiv f(u_n + u_{n-1}) = f(u_{n+1}) \equiv 0 \mod u_n thus u n f ( u n 1 ) u_n | f(u_{n-1}) for all n, so u n 1 f ( u n ) u_{n-1} | f(u_n) and u n + 1 f ( u n ) u_{n+1} | f(u_n) therefore u n 1 u n + 1 f ( u n ) u_{n-1}u_{n+1} | f(u_n) by Lemma 2. and u n 2 + 1 f ( u n ) u_n^2 + 1 | f(u_n) for even n u n 2 1 f ( u n ) u_n^2 - 1 | f(u_n) for odd n by Lemma 1. If we divide f ( x ) f(x) by x 2 + 1 x^2 + 1 , we get a remainder r ( x ) r(x) of degree at most 1. Then 0 f ( u n ) r ( u n ) m o d u n 2 + 1 0 \equiv f(u_n) \equiv r(u_n) \mod u_n^2 + 1 for even n, but u n 2 + 1 u_n^2 + 1 is quadratic in u n u_n and therefore outgrows r ( u n ) r(u_n) which is linear at most. Thus, r ( x ) = 0 r(x) = 0 and so x 2 + 1 x^2 + 1 divides f ( x ) f(x) . Similarly, x 2 1 x^2 - 1 divides f ( x ) f(x) . Hence, 4 2 + 1 = 17 4^2 + 1 = 17 and 4 2 1 = 15 4^2 - 1 = 15 divide f ( 4 ) f(4) , so 255 f ( 4 ) 255 | f(4) . The least positive multiple of 255 255 is 255 \boxed{255} , which is achieved by letting f ( x ) = x 4 1 f(x) = x^4 - 1 .

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...