Is the sum of these irreducible fractions an integer?

176 Views Asked by At

I was reading about irreducible fractions a few days ago, I had a question that I couldn't answer on my own. What if we are trying to find the sum of more than two irreducible fractions? For example, what if we are trying to add $9$ irreducible fractions, is it even possible for their sum to be an integer? I thought of an example that I would like to post here. The question is to answer whether the sum of all these $9$ irreducible fractions is an integer or not. Is it even possible to solve this question?

$$\frac{1}{p} + \frac{q}{p} + \frac{q^2}{p} + \frac{1}{p^2} + \frac{q}{p^2} + \frac{q^2}{p^2} + \frac{1}{p^3} + \frac{q}{p^3} + \frac{q^2}{p^3}$$

where $(q,p)=1$, $q,p = 2n+1$. That is, $q$ and $p$ are coprime and they are both odd numbers. Also assume that $p$ is a prime number.

1

There are 1 best solutions below

6
On

So I think I have found a complete answer for the case even when $p$ is not prime. There are infinite solutions for $p$ if every prime that divides $p$ is of the form $6n+1$. To show that this is true, set $q$ to be

$$q=\frac{1}{2} \left(\sqrt{4 a p^3-3}-1\right)$$

where $a$ is a natural number such that the above expression is integral. To show that such an $a$ must exist, first note that the expression inside the square root is odd and thus it suffices to show that there exists $a$ such that

$$4ap^3-3=s^2$$

for some natural number $s$. Note for later that this $s$ (if it exists) will always be odd. But this is a simple arithmetic progression which has a solution if and only

$$s^2\equiv -3\ (\text{mod }4p^3)$$

(in fact, it will have infinite solutions). To show that this is always the case, first note that

$$1\equiv -3\ (\text{mod }4)$$

Second, we must show that $-3$ is a quadratic residue of every prime in the prime factorization of $p$. But by our assumption these primes have the form $r=6n+1$. Then by Euler's criterion the equation

$$s^2\equiv -3\ (\text{mod }6n+1)$$

has a solution if and only if

$$\left(\dfrac{-3}{6n+1}\right)=1$$

(where $\left(\dfrac{a}{p}\right)$ is the Legendre symbol). Now, if $n=2m$ is even we have

$$6n+1=12m+1$$

and therefore

$$\left(\dfrac{-3}{12m+1}\right)=\left(\dfrac{-1}{12m+1}\right)\left(\dfrac{3}{12m+1}\right)=1\cdot 1=1$$

If $n=2m+1$ then

$$6n+1=12m+7$$

and therefore

$$\left(\dfrac{-3}{12m+7}\right)=\left(\dfrac{-1}{12m+7}\right)\left(\dfrac{3}{12m+7}\right)=(-1)\cdot (-1)=1$$

Either way, we conclude there are infinite integers $a$ such that $q$ is integral.

Next, we must show that $\text{gcd}(q,p)=1$. To do this, we will show there are integers $n,m$ (different from $n$ and $m$ above) such that

$$nq+mp-1=0$$

From above, we know that there exists odd $s$ such that

$$4ap^3-3=s^2$$

Thus, define

$$n=-\frac{s+1}{2}$$

$$m=ap^2$$

Then we have

$$nq+mp-1=-\frac{s+1}{4} \left(\sqrt{4 a p^3-3}-1\right)+ap^3-1$$

Substituting $s^2=4ap^3-3$ then gives us

$$=-\frac{s+1}{4} \left(s-1\right)+\frac{s^2-1}{4}=0$$

Having shown $p$ and $q$ are relatively prime, the last thing to do is show the original fraction in question is actually an integer. Plugging in our $p$ and $q$ and simplifying gives us a final result of $a(1+p+p^2)$ as desired.

To begin showing that these are the only solutions, note that if $p=3r$ for some $r$ then there are no solutions since

$$1+q+q^2\not\equiv 0\ (\text{mod }27)$$

for any $q\in\{0,1,...,26\}$ which implies $p^3\not|(1+q+q^2)$

Finally, if $p$ has a prime factor of the form $6n-1$, then $p^3\not |(1+q+q^2)$ (see answer to this question for reasoning).