4
$\begingroup$

I got this problem which I encountered during a limit of sequence calculation:

Show that $\forall n\in\Bbb{N}, (3+\sqrt 7)^n+(3-\sqrt 7)^n\in\Bbb{Z}$ And that $\forall n\in\Bbb{N}, (2+\sqrt 2)^n+(2-\sqrt 2)^n\in\Bbb{Z}$

I tried induction but I failed to show it. Thanks on any help.

Note: I don't know advanced number theory.

$\endgroup$
  • $\begingroup$ Do you know about binomial expansion? $\endgroup$ – abiessu Sep 21 '14 at 18:15
  • $\begingroup$ The sequence is therefore twice Sloane's A146963. $\endgroup$ – Robert Soupe Sep 21 '14 at 19:49
8
$\begingroup$

Way 1: Imagine expanding using the Binomial Theorem, and adding. There is nice cancellation of the terms that involve odd powers of $\sqrt{7}$.

Way 2: Our sum is invariant under the mapping that sends numbers of the form $a+b\sqrt{7}$, where $a$ and $b$ are integers, to $a-b\sqrt{7}$. But it is not hard to see that only integers are invariant under that mapping.

Way 3: We can also use a recurrence. Note that $3+\sqrt{7}$ and $3-\sqrt{7}$ are roots of the equation $x^2-6x+2=0$. Let $a_n=(3+\sqrt{7})^n$ and $b_n=(3-\sqrt{7})^n$. It is easy to verify that $a_{n+2}=6a_{n+1}-2a_n$ and $b_{n+2}=6b_{n+1}-2b_n$.

Let $c_n=(3+\sqrt{7})^n+(3-\sqrt{7})^n$. Then by linearity we have $$c_{n+2}=6c_{n+1}-2c_n.\tag{1}$$ Note that $c_0=2$ and $c_1=6$, both integers. It follows by induction using (1) that $c_n$ is an integer for all $n$.

Way 4: For brevity write our sum as $\alpha^n +\beta^n$. Note that $$\alpha^{n+1}+\beta^{n+1}=(\alpha^n +\beta^n)(\alpha +\beta)-\alpha\beta(\alpha^{n-1}+\beta^{n-1}).\tag{1}$$ Since $\alpha+\beta$ and $\alpha\beta$ are integers, using (1) we can show that if $\alpha^n+\beta^n$ and $\alpha^{n-1}+\beta^{n-1}$ are integers, then so is $\alpha^{n+1}+\beta^{n+1}$. This does the induction step.

| cite | improve this answer | |
$\endgroup$
  • $\begingroup$ I really like Way 2! No computation is needed. $\endgroup$ – Martin Brandenburg Sep 21 '14 at 18:38
7
$\begingroup$

If you know a bit of abstract algebra, there is nothing to compute. (And this is one of those examples which show that abstraction doesn't make things complicated, but in fact easier.)

Consider the field extension $\mathbb{Q}(\sqrt{7})$ of $\mathbb{Q}$. Its automorphism group is generated by $\sqrt{7} \mapsto -\sqrt{7}$. Since $(3+\sqrt{7})^n + (3-\sqrt{7})^n \in \mathbb{Q}(\sqrt{7})$ is obviously invariant under this transformation, it lies in $\mathbb{Q}$. It is also integral over $\mathbb{Z}$ (since $3,\sqrt{7}$ are integral). Since $\mathbb{Z}$ is integrally closed, the element actually lies in $\mathbb{Z}$.

The same argument works for $(2+\sqrt{2})^n+(2-\sqrt{2})^n$ and tons of other examples. For example, $(1+i)^n + (1-i)^n \in \mathbb{Z}$ by considering the field of Gaussian rationals $\mathbb{Q}(i)$.

Edit: As explained by André Nicolas in this answer, the argument gets even simpler by working directly with the ring $\mathbb{Z}[\sqrt{7}]$.

| cite | improve this answer | |
$\endgroup$
5
$\begingroup$

We have $$\begin{align}(3+\sqrt 7)^n+(3-\sqrt 7)^n&=\sum_{k=0}^{n}\binom{n}{k}3^{n-k}(\sqrt 7)^k+\sum_{k=0}^{n}\binom{n}{k}3^{n-k}(-\sqrt 7)^k\\&=\sum_{k=0}^{n}\binom{n}{k}3^{n-k}(\sqrt 7)^k(1+(-1)^k).\end{align}$$Here, note that if $k$ is odd, then $$(\sqrt 7)^k(1+(-1)^k)=0$$ and that if $k$ is even, then $$(\sqrt 7)^k(1+(-1)^k)\in\mathbb Z.$$

| cite | improve this answer | |
$\endgroup$
2
$\begingroup$

If $a$ and $b$ are any integers (positive or negative), then $(a+\sqrt b)^n + (a-\sqrt b)^n$ is an integer. This is because if you expand it out using the binomial expansion, the terms $(\sqrt b)^m$ from $(a+\sqrt b)^n$ cancel out the terms $(-\sqrt b)^m$ from $(a-\sqrt b)^n$ for odd $m$.

| cite | improve this answer | |
$\endgroup$
1
$\begingroup$

Hint:

In the binomial expansion of

$$(3+\sqrt 7)^n+(3-\sqrt 7)^n$$

we have terms such as

$$3^n+3^{n-1}\sqrt 7{n\choose 1}+3^{n-2}\sqrt 7^2{n\choose 2}+\dots$$

What happens when two such binomials are expanded, particularly when one of them involves a negative?

Alternative hint:

What happens in the following multiplication:

$$((2+\sqrt2)^n+(2-\sqrt2)^n)((2+\sqrt2)+(2-\sqrt2))$$

(This one may lend itself to induction...)

| cite | improve this answer | |
$\endgroup$
1
$\begingroup$

It's a great idea to look at $\mathbb{Z}[\sqrt{7}]$.

But if you haven't gotten to algebraic number theory in your studies just yet, let's try basic algebra. I shall use $n$ here to refer to 0 or a positive integer.

What's the formula for $(3 - \sqrt{7})^n$? Define the recurrence relations $a_n = 6a_{n - 1} - 2a_{n - 2}$, with $a_0 = 1, a_1 = 3$ and $b_n = 6b_{n - 1} - 2b_{n - 2}$ with $b_1 = 1, b_2 = 6$. Then, $(3 - \sqrt{7})^n = a_n - b_n \sqrt{7}$ for $n > 0$. Verify this before going further. Verify also that $a_n$ and $b_n$ are integers.

Now, what's the formula for $(3 + \sqrt{7})^n$? It turns out to be very similar, and it even employs the same recurrence relations defined earlier. Thus, $(3 + \sqrt{7})^n = a_n + b_n \sqrt{7}$.

Putting it all together: $(3 - \sqrt{7})^n + (3 + \sqrt{7})^n = a_n - b_n \sqrt{7} + a_n + b_n \sqrt{7} = 2a_n$.

| cite | improve this answer | |
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.