Global extrema of $f(x,y)=3x^2-y^3$ on the circle $x^2+y^2 \leq 25$

analysiscalculuslagrange multipliermultivariable-calculusreal-analysis

I wanted to know if my method and my results are correct. There isn't a solution to that question, so I don't know if it's correct.

Given $f(x,y)=3x^2-y^3$ and I need to find its global extremas on the circle $x^2+y^2 \leq 25$.


So inside the domain, I have $\nabla f =(6x, -3y^2)= (0,0)$ which yields the trivial case $x=y=0$
Now on the border, we have $x^2+y^2=25$. I use lagrange multiplier : $\nabla f = \lambda \nabla g$ where $g(x,y) =x^2+y^2-25$

This gives $(6x, -3y^2)= \lambda (2x, 2y)$.
If $x \neq 0$ and $y \neq 0$, then $\lambda = 3$, $y=-2$ and $x= \pm \sqrt{21}$. If $x=0, y = \pm 5$. If $y=0, x = \pm 5$. So if we plug our values into our function, we get $0, 71, -125, 125, 75$ So there is a global maximum of 125 at $(0,-5)$ and a global minimum of $-125$ at $(0, 5)$.
Are my solutions correct ?

Also, I guess I could have used polar coordinates on the border given that it's a circle. But if we do that ($r=\sqrt{25}=5$) and try to take the derivative of our function in polar coordinates, we get $5^2\frac{d}{d \theta}(3cos^2(\theta)-5sin^3(\theta))$ which isn't nice to derivate. Do you agree that it's much simpler to use Lagrange multiplier in this case ?

Thanks for your help !

Best Answer

Your solutions are correct and Lagrange is simpler than polar coordinates.

Another simple method: from $g(x,y)=0$ we get $x^2=25-y^2$, hence $f(x,y)=75-3y^2-y^3$.

Now discuss the function $h(y)=75-3y^2-y^3$ on the intervall $[-5,5]$.