Take the 2-minute tour ×
Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

I have reduced a certain equation (in positive integers) to the equation

$$A^2 + B^2 + 4 = C^2 + D^2. \quad(\star)$$

Assume the positive integers $(a,b,c,d)$ are any solution to $(\star)$. Are there any algebraic restrictions on [i.e., relationships between] the elements which are well-known or easy to prove? I'm thinking of things like Pythagorean means, or other relative size or congruence restrictions.

Thank you,

share|improve this question
One special case $D=2\implies A,B,C $ are Pythagorean triple (en.wikipedia.org/wiki/Pythagorean_triple) –  lab bhattacharjee Sep 15 '13 at 17:03

5 Answers 5

One can pick up some things from congruence considerations. Here is a sample. Let us work modulo $8$.

If $a$ and $b$ are odd, the left side is congruent to $6$ modulo $8$. But the right side cannot be congruent to $6$ modulo $8$. (We used the fact that if $x$ is odd, then $x^2\equiv 1\pmod{8}$.)

Suppose now that $a$ is even and $b$ is odd. If $a$ is divisible by $4$, then $a^2+b^2+4$ is congruent to $5$ modulo $8$, which means that one of $c$ or $d$ is odd, and the other is even but not divisibl by $4$.

Let $a$ and $b$ be both even. Then $c$ and $d$ also must be. We end up with an equation of shape $s^4+t^4 +1=u^2+v^2$.

We could also pick up information by working modulo $3$, and other moduli.

share|improve this answer
Thanks! One point: the right-hand side can be congruent to $5$ modulo $8$, if e.g. $2 \parallel c$ and $d$ is odd. –  Kieren MacMillan Sep 15 '13 at 18:29
Thanks, that was careless. –  André Nicolas Sep 15 '13 at 18:37
By the way, this was already quite helpful: using my reformulation and your congruence suggestion, I've already been able to eliminate one of the two possible cases I need to prove! Hopefully your "working modulo 3 and other moduli" will prove equally helpful. –  Kieren MacMillan Sep 15 '13 at 23:52
I am not optimistic about general solution. However, one should be able to get a parametric family of solutions. –  André Nicolas Sep 16 '13 at 0:01

A subset of integer solutions to,

$$A^2+B^2+4 = C^2+D^2\tag{1}$$

can be reduced to solving the Pell equation $x^2-ny^2 = 1$ for any non-square $n$ (where $n=5$ will involve the Fibonacci numbers). Consider the more general,

$$x_1^2+x_2^2+x_3^2 = y_1^2+y_2^2+y_3^2\tag{2}$$

The complete solution is,

$$(a+b)^2+(c+d)^2+(e+f)^2 = (a-b)^2+(c-d)^2+(e-f)^2\tag{3}$$


$$ab+cd+ef = 0\tag{4}$$

It is easy to make the last term of $(3)$ vanish. Let $e=f=t$. Thus,

$$(a+b)^2+(c+d)^2+(2t)^2 = (a-b)^2+(c-d)^2\tag{5}$$

However, if you set $a,b,c = x+y,\, -x+y,\, y^2$, then condition $(4)$ becomes,

$$x^2-(d+1)y^2 = t^2\tag{6}$$

If you set $t = 1$, then any Pell equation will give integer solutions to $(1)$. For a nice example, let $d = 4$ so one has to solve,

$$x^2-5y^2 = 1$$

The solution is $x_n = \tfrac{1}{2}F_{6n}+F_{6n-1}$ (A023039) and $y_n = \tfrac{1}{2}F_{6n}$ (A060645), where $F_n$ are the Fibonacci numbers. After some tweaking so that all terms won't be even, a Fibonacci identity which satisfies $(1)$ is then,

$$(F_{6n+3})^2+( \tfrac{1}{4}F_{6n+3}^2+4)^2+4 = (F_{6n+2}+F_{6n+4})^2+(\tfrac{1}{4}F_{6n+3}^2-4)^2\tag{7}$$

where $n=1$ is,

$$34^2+293^2+4 = 76^2 +285^2$$

and so on.

share|improve this answer

I'm not sure you can say a great deal. If $A^2+B^2=C^2$ (pythagoren triple), then we can have $D=2$.

We have $4=(C+A)(C-A)+(D+B)(D-B)$ so that $C+A$ and $D+B$ are both odd or both even.

In the even case, both products are divisible by $4$ and we have some expression like $4=56-52=14\times 4 - 26\times 2$ which gives $C=9, A=5, B=14, D=12$

In the odd case we get something like $4=55-51=5\times 11-3\times 17$ which gives $C=8, A=3, D=7, B=10$

share|improve this answer

The conditions for a number $n$ to be a sum of two squares are well-understood: $n$ must have no factors congruent to $3\bmod 4$ which occur to odd powers; in other words, if $p\equiv 3\pmod 4, p^{2k+1}|n$, then $p^{2k+2}|n$ (and in particular, if $p|n$ then $p^2|n$). This is a direct consequence of Fermat's Two-Square Theorem and the rule $(a^2+b^2)(c^2+d^2) = (ac-bd)^2+(ad+bc)^2$ for expressing the product of two numbers, each a sum of two squares, as a sum of two squares.

Applied to the given problem, this says that both $n=A^2+B^2$ and $n+4$ have no factors of their squarefree parts congruent to $3\bmod 4$; unfortunately, since addition plays notoriously poorly with factorization (for instance, beyond the trivial lack of shared factors, there's essentially nothing known about any correlation between the factorizations of $n$ and $n+1$) then it's unlikely that you'll be able to say anything about $A,B,C,D$ aside from the simplest congruence implications.

share|improve this answer

For the equation: $X^2+Y^2+4=Z^2+D^2$

You can draw a simple formula.















$c,b,q,a$ - what some integers.

share|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.