18
$\begingroup$

We call an integer $k\geq 1$ good if for all $q\in\mathbb{Q}$ there are $a_1,\ldots, a_k\in \mathbb{Q}$ such that $$q = \prod_{i=1}^k a_i \cdot\big(\sum_{i=1}^k a_i\big).$$ Euler showed that $k=3$ is good.

Is the set of good positive integers infinite?

$\endgroup$
28
$\begingroup$

I suspect that $k = 4$ is good, but am not sure how to prove it. However, every positive integer $k \geq 5$ is good. This follows from the fact (see the proof of Theorem 1 from this preprint) that for any rational number $x$, there are rational numbers $a$, $b$, $c$, $d$ so that $a+b+c+d = 0$ and $abcd = x$. In particular, one can take $$ a(x) = \frac{2(1-4x)^{2}}{3(1+8x)}, b(x) = \frac{-(1+8x)}{6}, c(x) = \frac{-(1+8x)}{2(1-4x)}, d(x) = \frac{18x}{(1-4x)(1+8x)}, $$ as long as $x \not\in \{1/4, -1/8\}$. (For $x = 1/4$ one can take $(a,b,c,d) = (-1/2,1/2,-1,1)$ and for $x = -1/8$ one can take $(a,b,c,d) = (-2/3,25/12,-1/15,-27/20)$.)

Now, fix $k \geq 5$, let $q \in \mathbb{Q}$ and take $a_{1} = a(q/(k-4))$, $a_{2} = b(q/(k-4))$, $a_{3} = c(q/(k-4))$, $a_{4} = d(q/(k-4))$ and $a_{5} = a_{6} = \cdots = a_{k} = 1$. We have that $$ a_{1} + a_{2} + a_{3} + a_{4} + \cdots + a_{k} = 0 + a_{5} + \cdots + a_{k} = k-4 $$ and $a_{1} a_{2} a_{3} a_{4} \cdots = \frac{q}{k-4} \cdot 1 \cdot 1 \cdots 1 = \frac{q}{k-4}$. Thus $$ \left(\prod_{i=1}^{k} a_{i}\right) \left(\sum_{i=1}^{k} a_{i}\right) = \frac{q}{k-4} \cdot (k-4) = q. $$

$\endgroup$
  • 4
    $\begingroup$ For any rational number $x$, by Euler's result we can write $-x$ as $abc(a+b+c)$ with $a,b,c\in\mathbb Q$, thus $x=abcd$ with $d=-a-b-c$, and $a+b+c+d=0$. So Euler's result for $k=3$ essentially implies that $k\ge5$ is good. $\endgroup$ – Zhi-Wei Sun Jun 22 '18 at 2:31
  • 3
    $\begingroup$ Indeed, we were unaware of Euler's result. $\endgroup$ – Anton Klyachko Jul 20 '18 at 0:06
21
$\begingroup$

Here I present the joint solution for $k=4$ by Guang-Liang Zhou and me. Actually, we not only show that $k=4$ is good, but also obtain a stroger result: Each rational number $q$ can be written as $abcd$ with $a,b,c,d\in\mathbb Q$ such that $a+b+c+d=1$. (We may also replace the number $1$ by any nonzero rational integer, but that's equivalent to the current version.)

Let $x=-81q/32$. If $q\not\in\{4/81,-8/81\}$, then $x\not\in\{1/4,-1/8\}$, hence $$a(x)+b(x)+c(x)+d(x)=1$$ and $$a(x)b(x)c(x)d(x)=-\frac{32}{81}x=q,$$ where \begin{gather*}a(x)=-\frac{(1+8x)^2}{9(1-4x)},\ \ b(x)=\frac49(1-4x), \\ c(x)=\frac{2(1-4x)}{3(1+8x)},\ \ d(x)=\frac{12x}{(1-4x)(1+8x)}.\end{gather*} For $q=4/81$, clearly $$\frac4{81}=\frac{4}3\times\frac13\times\left(-\frac13\right)\times\left(-\frac13\right)\ \ \text{with}\ \frac 43+\frac13+\left(-\frac13\right)+\left(-\frac13\right)=1.$$ Similarly, for $q=-8/81$ we have $$-\frac{8}{81}=\frac{1}3\times\frac23\times\frac23\times\left(-\frac23\right)\ \ \text{with}\ \frac 13+\frac23+\frac23+\left(-\frac23\right)=1.$$

$\endgroup$
  • $\begingroup$ Amazing, thank you! This answer would also deserve acceptance like the one above $\endgroup$ – Dominic van der Zypen Jun 23 '18 at 15:45

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.