1

Let $n \in \mathbb{N}$. Find all possible values of $\gcd(n^2+3, (n+1)^2+3)$.

I began this problem giving some values for $n$ and I found that $\gcd(n^2+3, (n+1)^2+3)=1$ for most of $n$ I tried, but if $n=6$, then $\gcd=13$.

Then I tried to prove that only for $n=6$, $\gcd \neq 1$ but I couldn't.

Can someone help me with this problem?

metamorphy
  • 39,111
rowcol
  • 887

5 Answers5

5

I think that I just solve it:

Let $d=gcd(n^2+3, (n+1)^2+3).$ By the property $\gcd(a,b)=\gcd(a,b-a)$, we obtain: $$d=\gcd(n^2+3,2n+1).$$ Since $d \mid n^2+3,$ we have $n^2+3=ds$ for some integer $s$.

Similarly, $2n+1=dr$ for some integer $r$, and by this relation we obtain: $4n^2=d^2 r^2 -2dr +1 \iff 4ds=4(n^2+3)=d^2 r^2 -2dr +13$.

Hence, it must be clear that $13=dq$ for some integer $q$. Then $d \mid 13$.

So $d=13$ or $d=1$ since $13$ is prime.

It is correct?

Thomas Andrews
  • 177,126
rowcol
  • 887
  • 2
    Alternatively, $\gcd(n^2+3,2n+1)=\gcd(n^2+3-3(2n+1),2n+1)=\gcd(n^2-6n,2n+1)$. Then $\gcd(n,2n+1)=1$ so $d=\gcd(n-6,2n+1)=\gcd(n-6,2n+1-2(n-6))=\gcd(n-6,13).$ So the $d=13$ when $n\equiv 6\pmod {13}$ and $d=1$ otherwise. – Thomas Andrews Jul 30 '19 at 19:43
  • 3
    The only incomplete part is you haven't shown it is possible for $d=1$ or $d=13.$ It could possible be always $13,$ or always $1.$ You have shown that it only can be $13$ or $1,$ but you haven't shown that the value is ever $1$, and likewise if the value is ever $13.$ To show that it can be $1$, use $n=0.$ To show it can be $13,$ use $n=6.$ – Thomas Andrews Jul 30 '19 at 19:50
2

We assume the Lemma that if $\gcd(a,b)=1$ and $c$ is an integer, then $\gcd(a,bc)=\gcd(a,c).$

Also, that in general: $\gcd(a,b)=\gcd(a,b-ak)$ for any integer $k.$

Then $$\begin{align} \gcd(n^2+3,(n+1)^2+3)&=\gcd(n^2+3,(n+1)^2+3-(n^2+3))\\ &=\gcd(n^2+3,2n+1)\\ &=\gcd(n^2+3-3(2n+1),2n+1)\\ &=\gcd(n(n-6),2n+1)\\ &=\gcd(n-6,2n+1)\quad \quad\text{ since }\gcd(n,2n+1)=\gcd(n,2n+1-2n)=1\\ &=\gcd(n-6,2n+1-2(n-6))\\ &=\gcd(n-6,13) \end{align}$$

So the GCD is $13$ when $n-6$ is divisible by $13$ and $1$ otherwise.

Thomas Andrews
  • 177,126
2

Lemma $\ \ (f(n\!+\!1),f(n))\, =\, (n-2a,4a+1)\,\ $ for $\ f(n) = n^2+a\ \ $ [OP is $\,a = 3]$

$\begin{align}{\bf Proof}\ \ \ \ (f(n\!+\!1),f(n)) &\,=\, (2n+1,\,f(n))\ \ \ \ \ \ \:\! {\rm by\ \ } 2n\!+\!1 = f(n\!+\!1)-f(n)\ \ \rm and\ Euclid\\[.2em] &\,=\, (\color{#c00}2n+1,\,4a+1) \ \ \ {\rm by}\ \ f(\color{#0a0}n) \equiv 4f(\color{#0a0}{\tfrac{\!-\!1}2})\equiv 4a\!+\!1\!\!\!\pmod{\color{#0a0}{n\equiv \tfrac{\!\!-1}2}}\\[.2em] &\,=\, (n-2a,\, 4a+1)\ \ \ {\rm via\ scale\ by}\ \ \color{#c00}{2^{-1}}\equiv -2a\!\!\!\!\underbrace{({\rm mod}\,\ {4a\!+\!1})}_{\textstyle 1\equiv -4a\equiv \color{#c00}2(-2a)} \end{align}$

where the $\rm\color{#0a0}{fractional}$ evaluation follows by this Theorem.

Remark $ $ See here for a more general way to do what we did above by evaluation at fractions - by instead scaling by leading coef's coprime to the gcd (which preserves the gcd), e.g. above by $(2,2n\!+\!1)=1$ we can scale $\,n^2+a\,$ by $\,2\,$ to get $\,(\color{#c00}{2n})n+2a\equiv (\color{#c00}{-1})n+2a\pmod{\!\color{#c00}{2n+1}}\,$ so $\,(2n+1,n^2+a) = (2\color{#0a0}n+1,\color{#0a0}{n-2a})= (2(\color{#0a0}{2a})+1,n-2a)\,$ by $\,\color{#0a0}{n\equiv 2a\pmod{n-2a}}\,$ using gcd mod reduction (congruence form of reduction step of Euclidean algorithm)

Bill Dubuque
  • 272,048
1

$$\gcd(n^2+3,(n+1)^2+3)=\gcd(n^2+3,2n+1)=\gcd(4n^2+12,2n+1)=$$ $$=\gcd(4n^2+4n+1-4n+11,2n+1)=\gcd(4n-11,2n+1)=$$ $$\gcd(4n+2-13,2n+1)=\gcd(13,2n+1).$$ Can you end it now?

Michael Rozenberg
  • 194,933
1

The Euclidean algorithm in $\mathbb Q[n]$ gives $$ 13 =(2 n + 3)(n^2 + 3)+(1 - 2 n)((n + 1)^2 + 3) $$ Therefore, the gcd is either $1$ or $13$.

Both values occur: $1$ occurs for $n=0$ and $13$ occurs for $n=6$.

In fact, $13$ occurs exactly when $n \equiv 6 \bmod 13$.

lhf
  • 216,483