5
$\begingroup$

How to list all integer solutions of

$m^2-n^5 = m - n$

Here $m$ and $n$ are some positive integers.

Also, I want to know the name of this type equations (if name exist).

Regards Rosy

$\endgroup$
  • $\begingroup$ A general name of this type of equation is a diophantine equation - an equation which asks for integer (or rational) solutions to polynomial equations. $\endgroup$ – Peter Woolfitt Apr 20 '15 at 5:16
  • $\begingroup$ Hmm. Well, $m$ must end in $0$, $1$, $5$, or $6$. (Because $n^5$ and $n$ end in the same digit, so $m^2$ and $m$ must also end in the same digit.) $\endgroup$ – Brian Tung Apr 20 '15 at 5:36
  • $\begingroup$ @BrianTung: That's just about $40\%$ of all numbers. $\endgroup$ – barak manos Apr 20 '15 at 5:44
  • $\begingroup$ Yes, I know that. :-) It wasn't meant to be a complete solution, just an observation off the top of my head. $\endgroup$ – Brian Tung Apr 20 '15 at 5:53
11
$\begingroup$

Write your equation as $m^2 - m = n^5 - n$.
You want $m = (1 + \sqrt{4 n^5 - 4 n + 1})/2$ to be an integer. Trying the first $10^6$ values of $n$, we find that $4n^5 - 4n + 1$ is a square for $n = 1, 2, 3, 30$, corresponding to $m = 1, 6, 16, 4930$.

The curve $x^2 - x - y^5 + y$ has genus $2$ (according to Maple), so by Faltings's theorem there are only finitely many rational points, and in particular only finitely many integer solutions. I suspect that the solutions I listed are all the positive integer solutions, but I don't know if it's possible to prove that.

$\endgroup$
  • $\begingroup$ ! I am looking for general solution which will list all possible solutions in integers. $\endgroup$ – rosy milton Apr 20 '15 at 8:14
4
$\begingroup$

Here's how I would start (don't know if I will finish):

If $m^2-n^5 = m - n$, then $m^2-m = n^5-n$ or $m(m-1) =n(n^4-1) =n(n^2-1)(n^2+1) =n(n-1)(n+1)(n^2+1) $.

The RHS needs to factor into two terms that differ by $1$. This seems important, but I don't know how to make use of this.

Also, multiplying by 4 and adding $1$ (to make the left side a square), $(2m-1)^2 =4m(m-1)+1 =4n(n-1)(n+1)(n^2+1)+1 $.

If $n$ is even, the RHS is of the form $8k+1$ where $k$ is odd.

If $n$ is odd, the RHS is of the form $64k+1$ where $k$ might be even.

I don't know where to go from here, so I'll leave it at this.

$\endgroup$
  • $\begingroup$ ! you have made very good observation. I do not know how to proceed from here. I want to know the general solution for this equation. Still I am trying... $\endgroup$ – rosy milton Apr 20 '15 at 8:12
3
$\begingroup$

This equation was solved by Bugeaud, Mignotte, Siksek, Stoll and Tengely, see e.g. http://arxiv.org/pdf/0801.4459v4.pdf

The proof uses linear forms in logarithms and Mordell-Weil sieving. I doubt that there is an easy approach to the problem.

$\endgroup$
  • $\begingroup$ ! the cited above paper is not clear to me. I think there may be a better and easy approach is there. Please let me know if you guess. $\endgroup$ – rosy milton Apr 21 '15 at 4:57
  • 1
    $\begingroup$ There is no reason to believe that there is an elementary approach to solve this problem. Sometimes Diophantine equations are just hard. $\endgroup$ – Mike Bennett Apr 21 '15 at 14:59
  • $\begingroup$ Theorem 1 of the linked paper gives all the pairs as $(X, Y ) = (−1, 0), (−1, 1), (0, 0), (0, 1), (1, 0), (1, 1), (2, −5), (2, 6), (3, −15), (3, 16), (30, −4929), (30, 4930)$ $\endgroup$ – Ross Millikan Apr 21 '15 at 15:08
  • $\begingroup$ As the authors mention, the problem was listed as an "open problem" in math.leidenuniv.nl/~evertse/07-workshop-problems.pdf . That's a pretty good indication that there is no easy approach. $\endgroup$ – Robert Israel Apr 21 '15 at 15:10
  • $\begingroup$ @Ross Millikan! As you said, according to paper link, those 12 set of pairs are solutions. But, I dint find the derivation of getting those pairs. Could you please explain how the paper or you got those pairs? $\endgroup$ – rosy milton Apr 22 '15 at 3:40
-2
$\begingroup$

Let start by eliminating the trivial solutions of the equation: $$m^2-n^5 = m - n$$ which is obviously equivalent to $$m^2-m=n^5-n$$ For instance, $m=n$ or $mn=0$ For I am sure you are looking the non-trivial ones. So we will suppose that $\gcd(m,n)=1$

First observation: According to Fermat's Little Theorem, we have: $$m^2 \equiv m \pmod 2$$ $$n^5 \equiv n \pmod 5$$ In other words, there exist 2 non-zero integers $s,t \in \mathbb{Z}$ such that: $$m^2=m+2s$$ $$n^5=n+5t$$ Therefore, $$m^2-n^5=m-n+2s-5t=m-n$$ As a result,$$2s=5t$$ From this, it is clear that $2|t$ and $5|s$ Therefore, $$2s=5t\equiv 0 \pmod {10}$$ Then, we can write $$ m^2-m=10k$$ $$n^5-n=10k$$ with $k \in\mathbb{Z} \neq 0$ The first equation can be easily solved by using the quadratic formula: $$(m-\frac{1}{2})^2=10k+\frac{1}{4}=\frac{40k-1}{4}$$ We must have $40k+1=u^2$ ($u \in\mathbb{Z} \neq 0$),in other words:$$40k=u^2-1=(u-1)(u+1)$$ Clearly, $u$ is odd. Hence, $\gcd(u-1),u+1)=2$. We deduct that one of those prime factors is a multiple of $20k$. Then we write: $$u \pm1=20k'$$ or $$u=20k'\pm 1$$ We then obtain the following values of $m$ $$m= \pm{\frac{20k'+1}{2}}+\frac{1}{2}=\pm 10k'+1$$ $$m= \pm{\frac{20k'-1}{2}}+\frac{1}{2}=\pm 10k'$$

Second Observation: From the system of equations$$ m^2-m=10k$$ $$n^5-n=10k$$ by equating $10k$ we obtain after factorization: $$m(m-1)=n(n^4-1)$$. According to Gauss lemma, since $\gcd(m,n)=1$ then, $n| m-1$ So there exists an non-zero integer $h$ such that $$m-1=nh$$ From that, we can deduct the values of $n$ knowing the above values of $m$. NOT SURE WHERE TO GO FROM HERE.

$\endgroup$

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.