Find the global maxima and minima of $f(x,y) = x^2 + 2y^2 - 4y$ subject to the constraint $x^2 + y^2 = 9$, using Lagrange multipliers.
Kush Motwani, AIR 11, 2023
Lagrange function $L(x,y,\lambda) = f(x,y) + \lambda\, g(x,y)$ (function + λ constraint)
Wkt, at extreme values, $L_x = 0,\ L_y = 0,\ L_\lambda = 0$. → ②
$L_x = 0 \equiv 2x + 2\lambda x = 0 \Rightarrow 2x(1+\lambda) = 0 \Rightarrow x = 0$ or $\lambda = -1$ → ③
$L_y = 0 \equiv 4y - 4 + 2\lambda y = 0 \Rightarrow y(2+\lambda) = 2 \Rightarrow y = \frac{2}{(2+\lambda)}$ → ④
$L_\lambda = 0 \Rightarrow x^2 + y^2 = 9$ → ⑤ (ie, the constraint).
Case 1: (from ③) $x = 0$
$\Rightarrow$ from ⑤: $y = \pm 3$
$\Rightarrow$ extreme points will be $(0,3)$ (Ⓐ) and $(0,-3)$ (Ⓑ)
$\therefore \pm 3 = \frac{2}{2+\lambda} \Rightarrow 2+\lambda = \pm\frac{2}{3} \Rightarrow \lambda = \frac{-8}{3}$ or $\frac{-4}{3}$
Case II: (from ③) $\lambda = -1$:
Using ④: $y = \frac{2}{(2+\lambda)} = 2$.
$\Rightarrow$ from ⑤: $x = \pm\sqrt{5}$
$\Rightarrow$ extreme points: $(\sqrt{5}, 2)$ (Ⓒ) and $(-\sqrt{5}, 2)$ (Ⓓ)
Now:
For Ⓐ: $(0, 3)$.
negative /0
$\Downarrow$
[neither maxima, nor minima]
For Ⓑ: $(0, -3)$.
$\Downarrow$
$L_{xx} < 0$, $\begin{vmatrix} L_{xx} & L_{xy} \\ L_{yx} & L_{yy} \end{vmatrix} > 0$ and $-M < 0$.
$\Downarrow$
[Maxima] → ⑥
For (C): $(\sqrt{5}, 2)$.
$\Downarrow$
Minima
For (D): $(-\sqrt{5}, 2)$
$\Downarrow$
Minima → (7)
For maxima, consider:
$f(x,y) = (x^2+y^2) + (y^2-4y) = 9 + (y^2 - 4y)$
subject to $x^2+y^2 = 9 \rightarrow y \in [-3,3]$.
$\Downarrow$
$\max (y^2-4y) = 21$ (at $y = -3$)
Thus: $f_{max} = 30$ at $(0,-3)$ (as already shown in (6))
Thus: global minima: $(\pm\sqrt{5}, 2) \Rightarrow f_{min} = 5$
global maxima: $(0,-3) \Rightarrow f_{max} = 30$
Hence solved.