2

We are given 3 equations:

$x^2+\sqrt3 xy + y^2 = 25$

$y^2 + z^2 = 9$

$x^2 +xz+ z^2 = 16$.

$x,y,z$ are positive real numbers.

Then we have to find value of $xy + 2yz + \sqrt3 xz$.

Dev
  • 35

5 Answers5

5

Based on system of equations being sides of a right angled triangle and a point P
inside the triangle such that -

$\angle BPC = 90^0, \angle APC = 150^0, \angle APB = 120^0 $
and, $AP = x, CP = y, BP = z$

We know area of a triangle is $\frac{1}{2} \times$ length of side 1 $\times$ length of side 2 $\times \sin \theta$
where $\theta$ is the angle between side 1 and side 2.

Now sum of area, $\triangle APC + \triangle BPC + \triangle APB = \triangle XYZ$

$\frac{1}{2}(xy\sin150^0 + yz\sin90^0 + xz\sin 120^0) = \frac{1}{2} \times 3 \times 4$

$xy \times \frac{1}{2} + yz + xz \times \frac{\sqrt3}{2} = 12$

$xy + 2yz + \sqrt3 xz = 24$

Math Lover
  • 51,819
  • nice solution because it gets to the value without finding x, y, z separately! – Prags Sep 06 '20 at 09:00
  • @prags yes if we try and find x, y, z first then it is similar to solving the system of equations, avoiding which was the idea in the first place. – Math Lover Sep 06 '20 at 09:12
  • Good; I will look at this some more. – Will Jagy Sep 06 '20 at 13:28
  • What would you do if it was $26$ instead of $25$ ? – Claude Leibovici Sep 07 '20 at 12:09
  • @ClaudeLeibovici of course you are right and I am aware. For $26$, at least it is still a triangle. There will be set of values when this cannot be expressed as squares of the sides of a triangle. So of course it does not apply then. But given it applies here, given the way question is asked, given the expression for which value has been asked, this is an efficient way to solve this specific problem. – Math Lover Sep 07 '20 at 12:38
  • 1
    I am just kidding. I don't know if you look at my edit. I found the work amazing (even if tedious). Cheers. – Claude Leibovici Sep 07 '20 at 12:41
  • @ClaudeLeibovici No I was not following this question anymore. I wish there was a way to notify changes (at least certain types of changes, new answers) to a post where we answered in the past, bookmarked or showed interest. I will definitely check when I get time. – Math Lover Sep 07 '20 at 12:48
3

Note that you have $$\forall x,y,z \in \mathbb{R}^{+}: \left\{\begin{aligned} x^{2}+\sqrt{3} xy + y^2 &=& 25\\ y^{2} + z^{2} &=& 9\\ x^{2} +xz+ z^{2} &=& 16 \end{aligned} \right.$$ if, and only if, $$ \forall x,y,z \in \mathbb{R}^{+}: \left\{\begin{aligned} x^{2}+\sqrt{3} xy + y^2 &=& \color{blue}{5}^{2}\\ y^{2} + z^{2} &=& \color{blue}{3}^{2}\\ x^{2} +xz+ z^{2} &=& \color{blue}{4}^{2} \end{aligned} \right. $$ Now, we can approach this problem as an algebraic geometry problem. Indeed, consider a triangle $\bigtriangleup XYZ$ with side lengths $3,4,5$ and draw a point $P$ inside the triangle such that $XP=x$, $YP=y$, and $ZP=z$. Now, you can considerer he equations in the context of the law of cosines.

Can you continue from here?

3

Just for the fun !

Using algebra, you can entirely solve the problem. Let $$X=x^2 \qquad Y=y^2 \qquad Z=z^2 \qquad a= xy \qquad b=xz$$So, the equations are now $$X+\sqrt{3} a+Y=25 \tag 1$$ $$Y+Z=9\tag 2$$ $$ X +b+ Z=16\tag 3$$ $$a^2=XY\tag 4$$ $$b^2=XZ\tag 5$$

Using $(1)$,$(2)$ and $(3)$ which are linear equations, we have $$X=-\frac{\sqrt{3} a}{2}-\frac{b}{2}+16\qquad Y=-\frac{\sqrt{3} a}{2}+\frac{b}{2}+9\qquad Z=\frac{\sqrt{3} a}{2}-\frac{b}{2}$$ Replacing, equations $(4)$ and $(5)$ become $$-a^2-50 \sqrt{3} a-b^2+14 b+576=0 \tag 6$$ $$-3 a^2+32 \sqrt{3} a-3 b^2-32 b=0\tag 7$$ Using $(7)$ $$b=\frac{1}{3} \left(\sqrt{-9 a^2+96 \sqrt{3} a+256}-16\right)$$ Plug in $(6)$ to get $$74 \sqrt{-9 a^2+96 \sqrt{3} a+256}-546 \sqrt{3} a+4000=0$$

One squaring step to get

$$a=\frac{96 \left(148+177 \sqrt{3}\right)}{6553}\qquad \implies \qquad b=\frac{384 \left(91 \sqrt{3}-72\right)}{6553}$$ $$x=32 \sqrt{\frac{91-24 \sqrt{3}}{6553}}\qquad y=3 \sqrt{\frac{2185+1152 \sqrt{3}}{6553}} \qquad z=12 \sqrt{\frac{3 \left(91-24 \sqrt{3}\right)}{6553}}$$

Now, you can compute the value of any function $f(x,y,z)$.

Edit

In comments, @WillJagy reported difficulties for the more general case where the equations would be $$X+\alpha\, a+Y=\beta \tag 1$$ $$Y+Z=9\tag 2$$ $$ X +b+ Z=16\tag 3$$ $$a^2=XY\tag 4$$ $$b^2=XZ\tag 5$$

Repeating the same steps, I have $$X=x^2=-\frac{1}{6} \sqrt{-3 a^2 \alpha ^2+6 a \alpha \beta -54 a \alpha -3 \beta ^2+54 \beta +781}-\frac{a \alpha }{2}+\frac{\beta }{2}+\frac{37}{6}$$ $$Y=y^2=+ \frac{1}{6} \sqrt{-3 a^2 \alpha ^2+6 a \alpha \beta -54 a \alpha -3 \beta ^2+54 \beta +781}-\frac{a \alpha }{2}+\frac{\beta }{2}-\frac{37}{6}$$ $$Z=z^2=-\frac{1}{6} \sqrt{-3 a^2 \alpha ^2+6 a \alpha \beta -54 a \alpha -3 \beta ^2+54 \beta +781}+\frac{a \alpha }{2}-\frac{\beta }{2}+\frac{91}{6}$$ The problem now is that the last equation in $a$ is a quartic and no more a quadratic.

To reduce it to a cubic would require $$16 \left(\alpha ^2-3\right)^2 \implies \alpha=\pm \sqrt 3$$ which in fact reduces the last equation to a quadratic the solution of which being $$a=\frac{(\beta +7) \left(\sqrt{3} \left(4 \beta ^2-55 \beta -63\right)+\sqrt{(49-\beta ) (\beta -1)}\right)}{2 \left(12 \beta ^2-54 \beta +403\right)}$$ So, we have the analytical solution for any $1\leq \beta \leq 49$, the variables $(x,y,z)$ being non-negative in this range.

  • Nice. Can we also use complex numbers to solve these eqn.z=rcosx+irsinx. Equating real and img part. – Dev Sep 06 '20 at 07:27
  • 1
    @Dev. There are many things we could do. Why don't you post an answer ? It would be interesting. Cheers :-) – Claude Leibovici Sep 06 '20 at 07:29
  • Claude, I have been trying to solve the general system with symbols, say $x^2 - a_{12}xy + y^2 = d_{12}^2$ for $x,y$ then the other two pairs. No luck yet. – Will Jagy Sep 06 '20 at 23:50
  • @WillJagy. If we just change the first equation as $x^2+\alpha xy+y^2=\beta$ and keep the other two as they are, I did not find any problem before the last step (computing $a$). This effective leads to a monstrous quadratic (I have the explicit solutions). But the problem is that we need the positive root. I shall make an edit. Cheers. – Claude Leibovici Sep 07 '20 at 10:41
  • Claude, thank you so much. – Will Jagy Sep 07 '20 at 13:29
  • @ClaudeLeibovici this looks great. Btw the triangle method that I used will also work for $\beta \lt 49$ (basically as long as it is a triangle, we can apply cosine law and add up individual areas and equate the same way as we did for $25$. It will of course be more tedious. Only if it goes to $49$ or above, the triangle method would not work. – Math Lover Sep 07 '20 at 18:37
  • @WillJagy. Good to know ! May I confess that I am very, very bad with triangles. What is interesting (just try it) is to plot the goal function with respect to $\beta$. Cheers and have a nice day (I am going to bed now). Thanks to you, I had fun. – Claude Leibovici Sep 07 '20 at 18:51
  • Sweet Dreams... – Will Jagy Sep 07 '20 at 18:57
2

Here is an accurate picture. Draw in some extra lines......

enter image description here

Will Jagy
  • 139,541
  • Yes that what I got but how we will proceed now and find xy+2yz+√3xz from this ? – Dev Sep 06 '20 at 02:51
  • +1 from me as it was good enough hint for the OP to continue from and solve. – Math Lover Sep 06 '20 at 06:30
  • @MathLover thanks. I wondered, last night, if an argument with Law of Sines and Law of Cosines might work, introducing one or two of the remaining angles as new vertices. I still do not know. It is very likely, perhaps I will examine the other answers in detail, that using the coordinate axes, the length 3 in the traditional $x$ direction and the 4 in the traditional $y$ direction. – Will Jagy Sep 06 '20 at 13:20
  • @WillJagy please see my solution which is based on it. – Math Lover Sep 06 '20 at 13:25
  • You are right ! The last equation is a quartic in $a$. – Claude Leibovici Sep 07 '20 at 11:03
  • Have a look at my edit (if you have time to waste). We meed $\alpha=\pm\sqrt 3$ to reduce from quartic to quadratic. Now, your turn : solve the problem changing $25$ to wathever you want using your method. Cheers :-) – Claude Leibovici Sep 07 '20 at 12:08
  • @ClaudeLeibovici Thank you for looking into the matter. In my first try, I put some things into gp-pari, then tried to make sense of that. Later, it beame far too hot in my apartment to think at all, 92 Fahrenheit is about 33 Celsius ( I am on the upper floor, the (flat) roof remains hot after dark). Now, early morning, it is 81 Fahrenheit, about 27 Celsius, I may be able to think straight for a while. – Will Jagy Sep 07 '20 at 13:20
  • 1
    @WillJagy. What is amaring is that it is only for $\beta=25$ that the expression is an integer. Morover, $24$ is the maximum value of the function. Knowing it, it becomes a piece of cake. Cheers :-) – Claude Leibovici Sep 07 '20 at 13:24
1

Given a triangle with sides $l_1,l_2,l_3$ we have

$$ \cases{ l_1^2=l_2^2+l_3^2-2l_2l_3\cos\theta_1\\ l_2^2=l_1^2+l_3^2-2l_1l_3\cos\theta_2\\ l_2^2=l_1^2+l_2^2-2l_1 l_2\cos\theta_3 } $$

then making $l_1=x,l_2=y,l_3=z$

$$ \cases{ 2\cos\theta_1=0\\ 2\cos\theta_2=-1\\ 2\cos\theta_3=-\sqrt{3} } $$

It is a rectangle. Etc.

Cesareo
  • 33,252