1
$\begingroup$

Consider the function on $\mathbb{R}^2$: \begin{equation*} V(x,y) = (x^2+y^2)^2-a(x^2+y^2)+a^2/4 \end{equation*} with $a$ real. I want to decide weather the critical points are max, min or saddle points.

We can see that the point $(0,0)$ and the points on the circumference $x^2+y^2 = \frac{a}{2}$ are critical points. If we calculate the hessian matrix we have \begin{equation*} \mathcal{H} = \begin{pmatrix} 4(3x^2+y^2)-2a & 8xy \\ 8xy & 4(x^2+3y^2)-2a \\ \end{pmatrix} \end{equation*} and \begin{equation*} \det \mathcal{H} = 16 \cdot \left[ 3 (x^2+y^2)^2-2a(x^2+y^2)+\frac{a^2}{4} \right] \end{equation*} Then $\det \mathcal{H}_{(0,0)} = 4a^2$ so the point $(0,0)$ is max if $a>0$ and min if $a<0$. And by deduction a min for $a=0$.

How can I decide the point on the circumference $x^2+y^2 = \frac{a}{2}$?

$\endgroup$
6
  • $\begingroup$ Compute the eigenvalues of $H$ instead of just the determinant. This will give you more detailed information. $\endgroup$
    – Marcel
    Commented Nov 23, 2021 at 17:09
  • $\begingroup$ And what can I do with that? $\endgroup$
    – Spectree
    Commented Nov 23, 2021 at 17:34
  • $\begingroup$ A positive eigenvalue corresponds to a direction along which you have a minimum; a zero eigenvalue a direction along which the function is constant; a negative eigenvalue a direction along which the function has a maximum. Try it. $\endgroup$
    – Marcel
    Commented Nov 23, 2021 at 17:45
  • $\begingroup$ Did you observe that $f(x,y)=(x^2+y^2-a/2)^2$? $\endgroup$ Commented Nov 23, 2021 at 20:22
  • $\begingroup$ yes, but for the points on the circumference I get $\lambda_1=0$ and $\lambda_2=4a$, so I have one eigenvalue equal to zero. That means I get flatness on the points? @Marcel $\endgroup$
    – Spectree
    Commented Nov 24, 2021 at 10:18

0

You must log in to answer this question.

Browse other questions tagged .