substitution of minimization problem restrictions gives wrong answer

36 Views Asked by At

Just got out of classes and the tutor couldn't find out what was wrong (though he solved it his way):

Find the points of the curve that is the intersection of equations

$$g^1: x^2 + y^2 -xy - z^2 = 1$$

$$g^2: x^2 + y^2 = 1$$

closest to the origin.

Well, that's $$\min_{\left(x,y,z\right)\in\mathbb R^3} (x^2+y^2+z^2)\quad\text{st } g^1, g^2$$

Both restrictions have to be satisfied, than the restrictions reduces to $z^2 = -xy$. Well, if the restrictions are true, than the problem reduces to

$$\min_{\left(x, y, z\right)\in\mathbb R^3}( 1 - xy)$$

If $x$ is minimum, then it has to follow that $1-y =0$ and $1-x=0$, so I get $x=1$, $y=1$ and $z^2 = -1$, and that's a problem, cause z has to be a real value (and dont satisfy restrictions).

I understood Lagrange will give me the right results, but it's cumbersome to do with two restrictions. Any clues what I got wrong?

1

There are 1 best solutions below

3
On

Here's an elementary (no differential calculus) solution: name $(P_\lambda)$ the horizontal plane having $z=\lambda$ as cartesian equation ($\lambda\in\Bbb{R}$). Now consider any $M\in (g_1\cap g_2)\cap(P_\lambda)$. By Pythagoras' theorem the distance from $M$ to the origin is $OM=\sqrt{1+\lambda^2}\geqslant1$.

If you take $\lambda=0$, any $M\in g_1\cap g_2$ with $z=0$ is such that $\mathbf{OM=1}$, which thus achieves a minimum. Now if you report $z=0$ in the equation for $g_1$ and $g_2$, you easily find out that the solutions are given by $xy=0$ and $x^2+y^2=1$, thus $(x,y)=(\pm1,0)$ or $(0,\pm1)$.

Conversely it is immediately checked that the four points $(\pm1,0,0)$ and $(0,\pm1,0)$ belong to $g_1\cap g_2$.enter image description here