The sequence $ (a_n)_{n\ge 1}$ is defined by $ a_1 = 3$, $ a_2 = 11$ and $ a_n = 4a_{n-1}-a_{n-2}$, for $ n \ge 3$. Prove that each term of this sequence is of the form $ a^2 + 2b^2$ for some natural numbers $ a$ and $ b$.
Problem
Source: Yugoslavia National Olympiad 2008
Tags: induction, quadratics, number theory, algebra proposed, algebra
22.04.2008 17:46
I think it's more number theory than algebra . Clearly it suffices to prove that each of the terms has prime divisors of the form $ 8k+1$ or $ 8k+3$ . Then we can easily find by induction that $ x_{n-1}x_{n+1}=x_n^2+2$ so each prime divisor of $ x_{n-1},\ x_{n+1}$ divides $ x_n^2+2$ so $ -2$ is a quadratic residue so this numbers are of the form $ 8k+1$ or $ 8k+3$ and we are done .
27.04.2008 23:44
Why is from here ob the problem trivial? Because i dont see why every number of the form 8k+1 or 8k+3 is ekspresible as a^2+2b^2
28.04.2008 00:20
Off topic: there is no country in the world that's called Yugoslavia for many years, so I doubt there is a National Olympiad of the same as well. I should know! Please change your subtitle...
09.06.2008 00:04
By induction $ a_{2n}=a_n^2+2.(1+a_1+a_2+...+a_{n-1})^2$ and $ a_{2n+1}=a_n^2+2.(1+a_1+a_2+...+a_n)^2$.
17.01.2009 01:42
Colud someone explain why each number with prime divisors of the form$ 8k+1$ and $ 8k+3$(only) can be expressed in the form $ a^2+2b^2$?
18.01.2009 13:17
Is is possible to see all the steps in crazyfehmy's proof by induction? It doesn't seem straightforward to me.
18.01.2009 17:00
Try first for 3 and 4, then assume it's true for 2n-2 and 2n-1, prove that it's true for 2n and 2n+1.
18.01.2009 19:19
That's what I tried and it didn't seem straightforward which is why I asked to see the proof in full.
20.01.2009 19:07
I'm still working on a proof by mathematical induction but haven't cracked it yet. Any ideas?
24.01.2009 20:06
I have tried to locate the source of this question but cannot find any reference to the Yugoslavia National Olympiad 2008 on the internet. Is there any such thing?
27.01.2009 15:31
Ok, let show this proof; First we will prove that $ 2(1 + a_1 + a_2 + ... + a_n) = a_{n + 1} - a_n$ by induction. For $ n = 1$ , $ 2(1 + 3) = 11 - 3$ true. Let it's true for $ n = k$. Because of $ a_{k + 2} = 4a_{k + 1} - a_k \Longrightarrow$ $ 2(a_1 + a_2 + ... + a_k + a_{k + 1}) = a_{k + 1} - a_k + 2a_{k + 1} = a_{k + 2} - a_{k + 1}$. Now we will prove that $ a_{2n} = a_n^2 + 2.(1 + a_1 + a_2 + ... + a_{n - 1})^2$ and $ a_{2n + 1} = a_n^2 + 2.(1 + a_1 + a_2 + ... + a_n)^2$ by induction. For $ n = 1 , 11 = 3^2 + 2(1)^2$ and $ a_3 = 41 = 3^2 + 2(1 + 3)^2$ true.Let it's true for $ n = k \Longrightarrow$ $ a_{2k + 2} = 4a_{2k + 1} - a_{2k} = 3{a_k}^2 + 8(1 + a_1 + a_2 + ... + a_k)^2 - 2(1 + a_1 + a_2 + ... + a_{k - 1})^2 =$ $ [3a_k + 2(1 + a_1 + a_2 + ... + a_{k - 1})]^2 + 2(a_1 + a_2 + ... + a_k)^2 = {a_{k + 1}}^2 + 2(a_1 + a_2 + ... + a_k)^2$ and by same way; $ a_{2k + 3} = {a_{k + 1}}^2 + 2.(1 + a_1 + a_2 + ... + a_{k + 1})^2$
27.01.2009 15:34
AndrewTom wrote: I have tried to locate the source of this question but cannot find any reference to the Yugoslavia National Olympiad 2008 on the internet. Is there any such thing? As milin alrady pointed out, there's no such country in the world anymore. The problem is from Serbian Mathematical Olympiad 2008.
27.01.2009 21:02
Wonderful crazyfehmy. Would you mind posting the steps for the case a(2k+3)?
28.01.2009 13:52
By using $ 2(1+a_1+a_2+...+a_k)=a_{k+1}-a_k$ $ 4[{a_{k+1}}^2+2(1+a_1+a_2+...+a_k)^2]-a_k^2-2(1+a_1+a_2+...+a_k)^2=$ $ {a_{k+1}}^2+2(1+a_1+a_2+...+a_{k+1})^2 \Longrightarrow$ $ 4a_{2k+2}-a_{2k+1}={a_{k+1}}^2+2(1+a_1+a_2+...+a_{k+1})^2=a_{2k+3}$
28.01.2009 14:06
Thanks. I've just done it but yours is shorter.
29.01.2009 09:55
Having proved that a(n-2)a(n) = (a(n-1))^2 + 2, I am now looking at silouan's post and would like to see much more detail as I am presuming that this means -2 is a quadratic residue of a(n) but don't know why this means that the prime factors of a(n) are 1 or 3 (mod 8), although I've noticed that the first few terms, 3, 11, 41, 153, 571, 2131, 7953 and 29681 are, nor do I know why this means that a(n) is of the form a^2 + 2b^2.
30.01.2009 03:00
A generalization one . . Let $ a_n$ be a sequence in a algebraically closed field defined by $ \ a_0 = 1, \ a_1 = B - 1,\ a_{n + 2} = Ba_{n + 1} - a_n$ . Prove that : $ a_{2n} = a_n^2 + (B - 2)(a_0 + ... + a_{n - 1})^2,\ a_{2n + 1} = a_n^2 + (B - 2)(a_0 + ... + a_n)^2$
01.02.2011 08:20
I calculated some values and got the results $a_1=(1,1)$ $a_2=(3,1)$ $a_3=(3,4)$ $a_4=(11,4)$ $a_5=(11,15)$ $a_6=(41,15)$ here $(x,y)$ denotes $x^2+2y^2$. From these I guessed that the sequence works in the following way: if $a_n=a^2+2b^2$ then $a_{n+1}$ is either $a^2+2(a+b)^2$ or $(a+2b)^2+2b^2$ and these two happen alternately. This fact is easy to prove by induction.
23.07.2016 22:32
Let $ (b_n)_{n\ge 1}$ the sequence defined by $b_1=1$, $b_2=4$ and $b_n=4b_{n-1}-b_{n-2}$, for $n\geq 3$ and we assumed and we assume for convenience that $a_1=1,a_2=3$ and $a_n=4a_{n-1}-a_{n-2}$, for $n\geq 3$ We prove by induction on $n$ that: $a_{n+1}-a_{n}$ $=$ $2b_n...(\star)$. For $n=1,2$ we get $3-1=2.1$ and $11-3=2.4$ which satisfies $(\star)$ For $n\geq 2$, assuming the result is true for $1,2, . . . , n$, we have: $a_{n}-a_{n-1}$ $=$ $2b_{n-1}$ and $a_{n+1}$ $-$ $a_{n}$ $=$ $2b_{n}$. Since $b_{n+1}$ $=$ $4b_{n}-b_{n-1}$ $\Longrightarrow$ $2b_{n+1}$ $=$ $8b_{n}-2b_{n-1}$ $\Longrightarrow$ $2b_{n+1}$ $=$ $(4a_{n+1}-4a_{n})-(a_n-a_{n-1})$ $=$ $4a_{n+1}$ $-$ $5a_n$ $+$ $4a_{n}$ $-$ $a_{n+1}$ $=$ $a_{n+2}$ $-$ $a_{n+1}$ which satisfies $(\star)$. We prove by induction on $n$ that: $(a_{2n-1}$ $=$ $a_{n}^2$ $+$ $2b_{n}^2$ and $a_{2n}$ $=$ $a_{n+1}^{2}$ $+$ $2b_{n}^2)$ $...(\star\star)$. For $n=1$, we get $3=1^2+2.1^2$ and $11$ $=$ $3^2+2.1^2$ and which satisfies $(\star\star)$ For $n\geq 1$, assuming the result is true for $1, . . . , n$, we have: $a_{2n-1}$ $=$ $a_{n}^2$ $+$ $2b_{n}^2$ and $a_{2n}$ $=$ $a_{n+1}^{2}$ $+$ $2b_{n}^2$. Since $a_{2n+1}$ $=$ $4a_{2n}$ $-$ $a_{2n-1}$ $\Longrightarrow$ $a_{2n+1}$ $=$ $(4a_{n+1}^2$ $+$ $8b_{n}^2)$ $-$ $(a_{n}^{2}$ $+$ $2b_{n}^2)$ $\Longrightarrow$ $2a_{2n+1}$ $=$ $8a_{n+1}^2$ $+$ $12b_{n}^2$ $-$ $2a_n^2$, replacing $12b_n^2$ $=$ $3$ $.$ $(a_{n+1}-a_{n})^2$ we get $2a_{2n+1}$ $=$ $11a_{n+1}^2$ $+$ $a_n^2$ $-$ $6a_n$ $.$ $a_n$ $=$ $2a_{n+1}^2$ $+$ $(3a_{n+1}$ $-$ $a_n)^2=$ $2a_{n+1}^2$ $+$ $(a_{n+2}$ $-$ $a_{n+1})^2$ $=$ $2a_{n+1}^2$ $+$ $4b_{n+1}^2$ $\Longrightarrow$ $a_{2n+1}$ $=$ $a_{n+1}^2$ $+$ $2b_{n+1}^2$. Since $a_{2n+2}$ $=$ $4a_{2n+1}$ $-$ $a_{2n}$ $\Longrightarrow$ $a_{2n+2}$ $=$ $(4a_{n+1}^2$ $+$ $8b_{n+1}^2)$ $-$ $(a_{n+1}^{2}$ $+$ $2b_{n}^2)$ $\Longrightarrow$ $2a_{2n+2}$ $=$ $6a_{n+1}^2$ $+$ $16b_{n+1}^2$ $-$ $4b_{n}^2$, replacing $12b_{n+1}^2$ $=$ $3$ $.$ $(a_{n+2}$ $-$ $a_{n+1})^2$ and $4b_n^2$ $=$ $(a_{n+1}$ $-$ $a_{n})^2$ and $a_{n+2}$ $=$ $4a_{n+1}$ $-$ $a_n$ we get $2a_{2n+2}$ $=$ $2a_{n+2}^2$ $+$ $4b_{n+1}^2$ $\Longrightarrow$ $a_{2n+2}$ $=$ $a_{n+2}^{2}$ $+$ $2b_{n+1}^2$ hence $(a_{2n+1}$ $=$ $a_{n+1}^2$ $+$ $2b_{n+1}^2$ and $a_{2n+2}$ $=$ $a_{n+2}^{2}$ $+$ $2b_{n+1}^2)$ therefore induction is complete $\Longrightarrow$ each term of this sequence is of the form $ a^2 + 2b^2$ for some natural numbers $ a$ and $ b$.
03.10.2022 14:53
Lemma :A number $K$ can be expressed $a^2+2b^2$ if and only if for primes $p \equiv 1,3 \pmod8$$v_p(K) \equiv 0 \pmod 2$ Proof: if part is trivial .For the only if part we should prove for a prime $p \equiv 1,3 \pmod8$ there exist positive integers $a ,b$ such that $p=a^2+2b^2$ (indeed these $a$ and $b$ are unique) .Becasue of $p \equiv 1,3 \pmod8$ there is a integer $x^2 \equiv -2 \pmod p$ By thue's lemma there exist integers $a,b$ such that $0<a^2,b^2<p$ and $a \equiv xb \pmod p $ so $p|a^2+2b^2$ and $0<a^2+2b^2<3p$ then $p=a^2+2b^2$ or $2p=a^2+2b^2$ for first case we are done for second case $a$ is even $a=2a'$ then we are done again .$2=0^2+2.1^2$ for prime $p \equiv 1,3 \pmod8$ $p^2=p^2+2.0^2$ By Brahmagupta identity , $(x^2+2y^2)(a^2+2b^2)=(ax+by)^2+2(xb-ay)^2$ lemma is proven . $\square$ Beyond this theory problem is trivial .By induction $a_n^2-4a_na_{n+1}+a_{n+1}^2=-2$ if $p|a_n$ then $p|a_{n+1}^2+2$ then $p \not\equiv 5 ,7 \pmod8$. $\blacksquare$