1

What is the non-zero integer general solution to $x^3+y^2=z^2$ ? I guess it is already solved in some book or paper, in that case plz help me to find that.

Edit:

  1. One solution is - $n^3 = [(n)(n+1)/2]^2 - [(n)(n-1)/2]^2$ Found in here.

  2. $(x,y,z)=\left(abuv,\frac{ab(bu^3-av^3)}{2},\frac{ab(bu^3+av^3)}{2}\right)$ where $bu^3\equiv av^3\pmod{2}$

  • Could they be zero? – J. W. Tanner Mar 12 '20 at 19:01
  • https://www.wolframalpha.com/input/?i=integer+solution+to+x%5E3%2By%5E2%3Dz%5E2 – manooooh Mar 12 '20 at 19:02
  • 3
    https://www.nctm.org/tmf/library/drmath/view/60941.html – JMoravitz Mar 12 '20 at 19:03
  • It is not hard to find infinite many non-trivial solutions to this equation (hint : consider pythagorean triples) , but to find the complete solution probably is hard. – Peter Mar 12 '20 at 19:03
  • @JMoravitz how could you find a post on the subject so quickly? did u know this before? or there is a technique to search such equations? – Consider Non-Trivial Cases Mar 12 '20 at 19:05
  • 2
    I googled "difference of squares is a cube" (as it made more sense to me to move the squares to one side) and this was the second result. The first was irrelevant, talking about the identities $a^3-b^3=(a-b)(a^2+ab+b^2)$ and similar – JMoravitz Mar 12 '20 at 19:07
  • As for the question of if there exist multiple nontrivial distinct choices for $y,z$ for some $x$, nontrivial in the sense of zero not allowed, distinct in the sense that changing the signs on $y,z$ doesn't count as different either, I don't yet know. Probably. – JMoravitz Mar 12 '20 at 19:10
  • $$15^3=60^2-15^2=120^2-105^2$$ so solutions are not necessarily distinct. – SlipEternal Mar 12 '20 at 19:19
  • The most commonly appreciated example might be $2^3+1^2=3^2$ – Keith Backman Mar 12 '20 at 19:35
  • Find the solution in integers of $,n=x^3,$ and $,n = z^2-y^2,$ (see OEIS sequence A042965). – Somos Mar 12 '20 at 20:07
  • @InterstellarProbe I got ur first equation, after that it is not clear, if possible please use simpler notation. – Consider Non-Trivial Cases Mar 12 '20 at 20:56
  • 1
    I will post it as an answer so I have more room. – SlipEternal Mar 12 '20 at 21:01
  • 1
    All integer solutions $(x,y,z)$ to the Diophantine equation $x^3+y^2=z^2$ are given by $$(x,y,z)=\left(abuv,\frac{ab(bu^3-av^3)}{2},\frac{ab(bu^3+av^3)}{2}\right)$$ for some $a,b,u,v\in\mathbb{Z}$ such that either $ab$ is even or $bu^3\equiv av^3\pmod{2}$. – Batominovski Mar 12 '20 at 21:44
  • @Batominovski Do you have a reference? Also, just a minor simplification; if $ab$ is not even then $a\equiv b\equiv1\pmod{2}$ and so the requirement that $bu^3\equiv av^3\pmod{2}$ is equivalent to $u\equiv v\pmod{2}$. – Servaes Mar 26 '20 at 11:58
  • @Servaes I already wrote a proof for this. Please take a look below. – Batominovski Mar 26 '20 at 18:36
  • @Batominovski I feel silly now; I think I got confused by the fact that the same user has asked practically the same question twice, and linked them. Anyway +1 to you, nice solution. – Servaes Mar 26 '20 at 19:19

4 Answers4

2

We have: $x^3+y^2 = z^2 \Longrightarrow x^3 = z^2-y^2$.

We are looking for non-trivial integer solutions. So, let's assume $x>0,y>0,z>0$. Thus, $z>y>0$. Let's write $z = y+n$.

This gives:

$$x^3 = (y+n)^2 - y^2 = 2ny + n^2$$

So, $x^3 \equiv n^2 \pmod{2n}$ implies there exists a solution in this form. Let's try it out:

$$n=1: x^3 = (y+1)^2-y^2 = 2y+1 \Longrightarrow x^3\equiv 1 \pmod{2}\text{ and }y = \dfrac{x^3-1}{2}$$

Examples:

$$3^3 = \left(\dfrac{3^3-1}{2}+1\right)^2 - \left(\dfrac{3^3-1}{2}\right)^2 = 14^2-13^2 \\ 5^3 = \left(\dfrac{5^3-1}{2}+1\right)^2 - \left(\dfrac{5^3-1}{2}\right)^2 = 63^2-62^2 \\ \vdots$$


$$n=2: x^3 = (y+2)^2-y^2 = 4y+4 \Longrightarrow x^3\equiv 0 \pmod{4} \text{ and }y = \dfrac{x^3-4}{4}$$

Examples:

$$4^3 = \left(\dfrac{4^3-4}{4}+2\right)^2 - \left(\dfrac{4^3-4}{4}\right)^2 = 17^2-15^2 \\ 8^3 = \left(\dfrac{8^3-4}{4}+2\right)^2 - \left(\dfrac{8^3-4}{4}\right)^2 = 129^2-127^2 \\ \vdots$$


$$n=3: x^3 = (y+3)^2-y^2 = 6y+9 \Longrightarrow x^3 \equiv 3 \pmod{6} \text{ and }y = \dfrac{x^3-9}{6}$$

Examples:

$$3^3 = \left(\dfrac{3^3-9}{6}+3\right)^2-\left(\dfrac{3^3-9}{6}\right)^2 = 6^2-3^2 \\ 9^3 = \left(\dfrac{9^3-9}{6}+3\right)^2-\left(\dfrac{9^3-9}{6}\right)^2 = 123^2-120^2 \\ \vdots$$

Note: $(2n)^2 \equiv 0 \pmod{4n}$ and $(2n-1)^2 \equiv n \pmod{2(2n-1)}$ for all positive integers $n$. This yields:

$$\begin{array}{c|c}n & \text{equivalence} \\ \hline 1 & x^3 \equiv 1\pmod{2} \\ 2 & x^3 \equiv 0\pmod{4} \\ 3 & x^3\equiv 3 \pmod{6} \\ 4 & x^3\equiv 0\pmod{8} \\ 5 & x^3\equiv 5 \pmod{10} \\ 6 & x^3\equiv 0 \pmod{12} \\ 7 & x^3\equiv 7 \pmod{14} \\ 8 & x^3\equiv 0 \pmod{16} \\ 9 & x^3\equiv 9 \pmod{18} \\ \vdots & \vdots\end{array}$$

SlipEternal
  • 10,560
2

Let $(x,y,z)\in\mathbb{Z}^3$ be such that $x^3+y^2=z^2$. Then, $$x^3=z^2-y^2=(z-y)(z+y)\,.$$ Write $$z-y=s\,\prod_{i=1}^l\,p_i^{3t_i}\,\prod_{j=1}^m\,q_j^{3u_j+1}\,\prod_{k=1}^n\,r_k^{3v_k+2}\,,$$ where

  • $s\in\{-1,+1\}$,
  • $p_1,p_2,\ldots,p_l$, $q_1,q_2,\ldots,q_m$, and $r_1,r_2,\ldots,r_n$ are pairwise distinct prime natural numbers, and
  • $t_1,t_2,\ldots,t_l$, $u_1,u_2,\ldots,u_m$, and $v_1,v_2,\ldots,v_k$ are nonnegative integers.

Take $$a:=\prod_{j=1}^m\,q_j\,,$$ $$b:=\prod_{k=1}^n\,r_k\,,$$ and $$c:=s\,\prod_{i=1}^l\,p_i^{t_i}\,\prod_{j=1}^m\,q_j^{u_j}\,\prod_{k=1}^n\,r_k^{v_k}\,.$$ Then, $z-y=ab^2c^3$ with $a$ and $b$ being squarefree positive integers, and $c$ being an integer. Because $x^3=(z-y)(z+y)$, it follows immediately that $$z+y=a^2bd^3\,,$$ for some integer $d$. Consequently, $$x=\sqrt[3]{(z-y)(z+y)}=abcd\,,\tag{*}$$ $$y=\frac{z+y}{2}-\frac{z-y}{2}=\frac{ab(ad^3-bc^3)}{2}\,,\tag{#}$$ and $$z=\frac{z+y}{2}+\frac{z-y}{2}=\frac{ab(ad^3+bc^3)}{2}\,.\tag{$\star$}$$ For $y$ and $z$ to be integers, we need that $$ab(c-d)\equiv ab(ad^3\pm bc^3)\equiv 0\pmod{2}\,.$$

Thus, a solution $(x,y,z)$ is uniquely determined by

  • two squarefree positive integers $a$ and $b$ which are relatively prime, and
  • two integers $c$ and $d$ such that $ab(c-d)$ is divisible by $2$

via the formulae (*), (#), and ($\star$) above. If we relax the first condition by allowing $a$ and $b$ to be arbitrary integers, then more than one quadruples $(a,b,c,d)$ may produce the same triple $(x,y,z)$.

Here are some infinite families presented by other users. The infinite family $$(x,y,z)=\left(N,\dfrac{N(N-1)}{2},\dfrac{N(N+1)}{2}\right)$$ for $N\in \mathbb{Z}$ corresponds to $(a,b,c,d)=(N,1,1,1)$. The infinite family $$(x,y,z)=\left(N^2,\dfrac{N(N^4-1)}{2},\dfrac{N(N^4+1)}{2}\right)$$ for $N\in \mathbb{Z}$ corresponds to $(a,b,c,d)=(N,1,1,N)$.

Batominovski
  • 50,341
  • Note that there are no conditions on $c$ and $d$ other than having $ab(ad^3\pm bc^3)$ be even. So in fact you can take any four integers $e$, $f$, $g$ and $h$ and set $$(a,b,c,d):=\left( \operatorname{rad}\Big(\tfrac{e}{\gcd(e,f)}\Big), \operatorname{rad}\Big(\tfrac{f}{\gcd(e,f)}\Big), \operatorname{rad}\Big(\tfrac{e}{\gcd(e,f)}\Big)^{-1}eg, \operatorname{rad}\Big(\tfrac{f}{\gcd(e,f)}\Big)^{-1}fh \right),$$ to get a quadruple of the form you describe. Now you only need the requirement that $ef(g-h)$ is even. You could circumvent this by multiplying by $$t:=\gcd(2,e+1,f+1,g+h+1).$$ – Servaes Mar 26 '20 at 19:45
  • This shows that the integral solutions are parametrized by $$(x,y,z)=\left(abcd,\tfrac t2ab(ad^3-bc^3),\tfrac t2ab(ad^2-bc^3)\right),$$ where $e,f,g,h\in\Bbb{Z}$ and $t=\gcd(2,e+1,f+1,g+h+1)$. – Servaes Mar 26 '20 at 19:47
1

Equation $(x^3+y^2=z^2)$ has parametric solution:

$x=(2k-1)$

$y=(k-1)(2k-1)$

$z=k(2k-1)$

For, $k=7$ we get:

$(x,y,z)=(13,78,91)$

Sam
  • 11
0

There are a lot of solutions. For example, parameterize a set of such solutions.

We have modulo $7$ the only cubes are $0$ and $\pm1$ and besides $t^6=1$ for all $t$ non divisible by $7$. Take for example the cube $1$ so we have $$1\equiv(z-y)(z+y)\pmod7$$ We can choose for example $z+y=t^5$ and $z-y=t$ so we get $$z=\frac{t^5+t}{2}\hspace{15mm}y=\frac{t^5-t}{2}$$ This allow us to get the parameterization in integers $$\begin{cases}x=t^2\\y=\dfrac{t^5-t}{2}\\z=\dfrac{t^5+t}{2}\end{cases}$$in relation with the identity $$(t^2)^3+\left(\frac{t^5-t}{2}\right)^2=\left(\frac{t^5+t}{2}\right)^2$$

Ataulfo
  • 32,657