0

In each of the following three equations I need help in finding all solutions in positive integers :

i) $\dfrac 1x+\dfrac 2y-\dfrac3z=1 $

ii) $\dfrac 1{x^2}+\dfrac 2{y^2}+\dfrac 3{z^2}=\dfrac 23$

iii) $p^x-y^3=1$ , where $p$ is a given prime

Please help

Souvik Dey
  • 8,535
  • Please, post only one question in one post. Posting several questions in the same post is discouraged and such questions may be put on hold, see meta. – Martin Sleziak Jun 19 '14 at 11:43
  • 1
    @martin many interesting questions are unfortunately put on hold or even closed. – Peter Jun 19 '14 at 11:54
  • 1
    Posting several questions as one: http://meta.math.stackexchange.com/questions/6464/posting-multiple-questions-as-one? – Martin Sleziak Jun 19 '14 at 12:37
  • @Peter What I mean is that according to recommendation from the meta thread I've linked, this question should be closed even if the three parts of it are fine when posted as separate questions. – Martin Sleziak Jun 19 '14 at 12:38
  • 1
    A hint for the last equation. $$p^x=y^3+1=(y+1)(y^2-y+1)$$ What can you say about the greatest common divisor of $y+1$ and $y^2-y+1$? That will tell you how you must select $p$. The rest is easy. – Jyrki Lahtonen Jun 19 '14 at 12:48
  • A similar equation was decided there. http://math.stackexchange.com/questions/450280/erdös-straus-conjecture/831830#831830 If it is not satisfied with tell. – individ Jun 19 '14 at 13:32

3 Answers3

2

For the 1st equation, we have $${1\over x}+{2\over y}>1$$ so we can't have both $x\ge3$ and $y\ge3$. If $x=2$ then $y$ must be 1, 2, or 3 for the inequality to hold; check these in turn, and find $(x,y,z)=(2,1,2),(2,2,6),(2,3,18)$. If $x=1$, you get $${2\over y}={3\over z}$$ which gives an infinity of solutions $(1,2k,3k)$. If $y=2$ then $${1\over x}={3\over z}$$ and you get $(k,2,3k)$. If $y=1$, then $3/z>1$ and all you get is $(2,1,2)$ again. So that should be it.

Similar methods apply to the second equation. We need $z\ge3$, $y\ge2$, $x\ge2$, but, aside from $(3,3,3)$, at least one of $x,y$ must be less than 3. Try all the cases --- there aren't that many.

For the third one, $$p^x=y^3+1=(y+1)(y^2-y+1)$$ so we must have $y+1=p^r$ and $y^2-y+1=p^s$ for some $r,s$. Then $$p^{2r}-3p^r=y^2-y-2=p^s-3$$ so $r=0$ or $s=0$ or $p=3$. It should be easy from there.

Gerry Myerson
  • 185,413
1

i) Multiplying with xyz gives $$yz + 2xz -3xy = xyz$$

Beside (2,1,2) and (2,3,18) I only found solutions with x = 1 or y = 2, but I have no proof that there are no more. It is probably difficult to find all solutions here.

ii) Multiplying with $3x^2y^2z^2$ gives $$3y^2z^2 + 6x^2z^2 + 9x^2y^2 = 2x^2y^2z^2$$

I checked all triples with x,y,z$ \le 1000$ and (3,3,3) is the only solution. But I have no proof that there are no more solutions.

iii) Catalan's conjecture (now proven) states that the only nontrivial solution of $a^b-c^d=1$ is $3^2-2^3=1$ (nontrivial means a,b,c,d > 1)

Peter
  • 86,576
1

i) We'll check two cases:

Case 1: $x\ge y$

Then we have:

$$1 = \frac 1x + \frac 2y - \frac 3z \le \frac 3y - \frac 3z$$ $$\frac 13 \le \frac{z-y}{yz}$$ $$3z - 3y \ge yz$$

Obviously $y\le2$, otherwise the inequality wouldn't hold.

If $\fbox{$y=1$}$ we get:

$$1 = \frac 1x + 2 - \frac 3z$$ $$\frac 3z - \frac 1x = 1$$ $$3x - z = zx$$

Obviosuly $z\le2$. plugging all possible values we get $\fbox{$x=z=2$}$.

If $\fbox{$y=2$}$ we get:

$$1 = \frac 1x + 1 - \frac 3z$$ $$\frac 3z = \frac 1x$$

Hence all solutions are: $\fbox{$(k,2,3k); k \in \mathbb{N}$}$

Simular method for $$y\le x$$

ii) Use Garry's hint and note that at at least one of the variables have to be divisible by 3.

iii) The equation is equivalent to:

$$p^x = (y+1)(y^2 - y + 1)$$

Now we have two cases:

**Case 1:**$(y+1, y^2 - y + 1) = 1$

This means that this numbers are coprime and that means that one of them has to be equal to one. Checking both cases yields: $$(p,x,y) = (2,1,1)$$

Case 2:$p$ divides both numbers

Then we have:

$$p\mid y^2 - y + 1 -(y+1) = y(y-2)$$

Since $p\mid y+1$ we get $p\not\mid y$. Hence $p\mid y-2$. Then we have: $p \mid y+1 - y + 2=3 \implies p = 3$

$y=1$ doesn't yield solution and $y=2$ gives us $p=3, x=2$. Now let $y\ge 3$. Obviously $y^2 - y + 1 > y + 1$ for $y>2$ and hence we have:

$$y+1 = 3^n \quad y^2 - y + 1 = 3^m, \quad m>n$$

Now check all possible combinations for y modulo 3 and you'll get a contradiction fairly easy.

Stefan4024
  • 36,357