0

Given $a+ab+b=2+3\sqrt2$ and $a^2+b^2=6$, $a,b\in\mathbb{R}$. Does that conditions are sufficient for finding $a$ and $b$? I try as follows.

\begin{align} &(a+ab+b)^2=(2+3\sqrt2)^2\\ \iff & a^2+b^2+a^2b^2+2a^2b+2ab+2ab^2 = 4+18+12\sqrt2\\ \iff & (a^2+b^2)+a^2b^2+2a^2b+2ab+2ab^2 = 22+12\sqrt2\\ \iff & 6+a^2b^2+2ab(a+b+1) = 22+12\sqrt2\\ \iff & a^2b^2+2ab(a+b+1) = 16+12\sqrt2\\ \end{align}

It's difficult and spend many time to find $a$ and $b$, so my question is: Does two condition above are sufficient to find $a$ and $b$?

user
  • 27,958
  • You get $a=2$, or a cubic equation in $a$. – Dietrich Burde Feb 19 '25 at 18:04
  • https://math.stackexchange.com/questions/1767225/algorithm-intersection-of-two-conics – Travis Willse Feb 19 '25 at 18:15
  • If the two given equations are independant then there are always solution (possibly non-real). With your system if the radii of the circle is greater that the distance between the origine and the hyperbola,(you find it making $x=y$) then you will have two solutions or one double solution. If you reduce enough the radii of the circle your solutions will be non-real. – Ataulfo Feb 20 '25 at 00:59
  • Noting $m=2+3\sqrt2$ you have as resultant $a^4+2a^3-4a^2-(12+2m)a+m^2=0$ and it is not so easy to solve this equation. – Ataulfo Feb 20 '25 at 01:19

4 Answers4

2

Hint: $$(a+b)^2=6+2ab\ \ \ \ (1)$$ $$ab=2+3\sqrt2-(a+b)\ \ \ \ (2)$$

So, $a+b$ is a root of $$t^2=6+2(2+3\sqrt2-t)\iff t^2+2t-(10+6\sqrt2)=0$$

$t=\dfrac{-2\pm\sqrt{4+4(10+6\sqrt2)}}2=-1\pm\sqrt{11+6\sqrt2}$

Now $11+6\sqrt2=2\cdot3\sqrt2+3^2+2=(3+\sqrt2)^2$

$\implies t=-1\pm(3+\sqrt2)=2+\sqrt2$ or $-4-\sqrt2$

From $(2), $ you can get the value of $ab$ corresponding to each of the two values of $a+b$

Now if we know the values of $a+b,ab$ can we find $a,b$ separately ?

1

Yes it is possible.

From the second equation you have: $(a+b)^2=6+2ab$, which combining with the first equation gives the square equation: $$ 6+2ab=(2+3\sqrt2-ab)^2 $$ with solutions $ab=2\sqrt2$ and $ab=6+4\sqrt2$. Using this and $a^2+b^2=6$ you should be able to solve the problem.

user
  • 27,958
1

Let $\,a=m+n,\,b=m-n\,$ . Then by mental calculation, we have :

$$\begin{align}\begin{cases}2m+m^2-\color{#c00}{n^2}=2+3\sqrt 2\\ m^2+n^2=3\Leftrightarrow\color{#c00}{n^2}=3-m^2\end{cases}\end{align}$$

This means that the above system of equations leads to the quadratic equation with respect to $m.$ Thus, the answer becomes to your exact question is $\rm{“Yes”}$ .

1

The given system is an algebraic symmetric system in the unknowns $a,b$. (I.e. the polynomials involved in the two equations are symmetric in the unknowns $a,b$.) In such cases, the standard method to solve is to use the substitution $s=a+b$ (for the sum), and $p=ab$ (for the product). The given system becomes then an equivalent algebraic system in $s,p$, and the degree drops by chance: $$ \left\{ \begin{aligned} s + p &= 2+3\sqrt 2\ ,\\ s^2 - 2p &= 6\ . \end{aligned} \right. $$ We eliminate $p$ among the two equations (add two times the first to the second to eliminate $p$), and obtain an equation in $s$: $$ s^2 + 2s - 10-6\sqrt 2=0\ . $$ Its half-discriminant is $1^2+(10+6\sqrt 2)=11+6\sqrt2$. Is this a "nice square"? For instance in the ring $\Bbb Z[\sqrt 2]$.

If yes, $11+6\sqrt 2=(m+n\sqrt 2)^2$.

We obtain passing to the norm on both sides (where the norm of $m+n\sqrt2$ is $(m+n\sqrt2)(m-n\sqrt 2)=m^2-n^2\in\Bbb Z$) the relation:

$49=11^2-2\cdot 6^2=(m^2-2n^2)^2$. So we search (up to a sign) for $m,n\in\Bbb Z$ with $m^2-2n^2=7$ and $11+ 6\sqrt 2 =(m+n\sqrt2)^2=(m^2+2n^2)+(2mn)\sqrt 2$. We have luck, the equations $m^2+2n^2=11$, $m^2-2n^2=7$ give us $m^2=9$, $n^2=1$, and we need also $2mn=6$. Yes, the luck shows we have a discriminant equal to $11+6\sqrt 2=(3+\sqrt 2)^2$. So $s = -1\pm (3+\sqrt 2)$.

The two possible values for the sum $s$ are: $2+\sqrt 2$ and $-4-\sqrt 2$.

The corresponding values for $p = 2+3\sqrt 2-s$ are respectively: $2\sqrt 2$ and $6+4\sqrt 2$. The last value has to be rejected because of $a,b\in\Bbb R$, which leads to $0\le (a- b)^2=a^2+b^2- 2ab=6-2ab=6-2p$, which is invalid for p=6+4\sqrt 2$.

So the only chance is: $a+b=s=2+\sqrt 2$ and $ab=p=2\sqrt 2$. Then (by Vieta) $a,b$ are the roots of the equation in a (new) variable $T$: $$T^2 - sT+p=0\ ,\qquad\text{ i.e. }\qquad T^2-(2+\sqrt 2)T+2\sqrt 2=0\ . $$ We either "see the roots", they are $2$ and $\sqrt 2$, or we compute them.

Conclusion: The given symmetric system has two solutions: $(a,b)=(2,\sqrt 2)$, and $(a,b)=(\sqrt2, 2)$. We cannot "determine" $a,b$, but the set $\{a,b\}$ is determined.



Second solution: (Can be given if we know the solution $a=2$. If not, this is not really a good path.)

The first equation, $a+ab+b=2+3\sqrt 2$ is linear in $b$, so we can extract $b$ linearly from it and plug in into the second equation. This implies successively: $$ \begin{aligned} b &=\frac{2+3\sqrt 2-a}{a+1}\ ,\\ 0 &= a^2 -6 + \frac{(2+3\sqrt 2-a)^2}{(a+1)^2}\ ,\\ 0 &= a^2(a+1)^2 - 6(a+1)^2 + (a-2-3\sqrt 2)^2\ ,\\ 0 &= \underbrace{a^4 + 2a^3 - 4a^2 -(16+6\sqrt 2)a + (16+12\sqrt 2)}_{P(a)}\ . \end{aligned} $$ We may plot the polynomial (function) $P$ or compute its values in some points trying to identify some intervals with roots. When computing its values in $0,1,2,3,\dots$ we have the chance to see that $a=2$ is a root, $P(2)=0$. This leads to $b=\sqrt 2$.

But the given system is symmetric, so together with $(a,b)$ then $(b,a)$ is also a solution. So $a=\sqrt 2$ must be a root, and we easily check this. Then $b=2$ is the corresponding value.

We have two $a$-roots. The other roots are obtained after a division of $P$ by $(a-2)(a-\sqrt 2)$. Let us denote by $u$ the value $u=\sqrt 2$ for easy typing. Then the Horner scheme for this division is: $$ \begin{array}{r|ccccc} & 1 & 2 & -4 & -16-6u & 16 + 12u\\\hline 2 & 1 & 4 & 4 & -8-6u & \boxed 0\\\hline u & 1 & 4+u & 6+4u & \boxed 0\\\hline \end{array} $$ so the remained roots are roots of the polynomial with the coefficients from the last line: $a^2+(4+u)a+(6+4u)$. But its discriminant is negative, $(4+u)^2-4(6+4u)=4u-6<0$, so there are no further real solutions.

J. W. Tanner
  • 63,683
  • 4
  • 43
  • 88
dan_fulea
  • 37,952