Lagrangian Optimization: equality constraint vs inequality constraint?

109 Views Asked by At

For minimization problem

$minimize$ $x_1+x_2$ with constraint $x_1^2+x_2^2-2=0$

How does it differ from

$minimize$ $x_1+x_2$ with constraint $-x_1^2-x_2^2+2 \geq 0$ ?

For the equality constraint, I got the Lagrangian

$\mathcal{L}(x_1,x_2,\lambda)=x_1+x_2+\lambda(x_1^2+x_2-2)$

and the KKT conditions

$1+2\lambda x_1=0$

$1+2\lambda x_2=0$

$x_1^2+x_2^2-2=0$

So I got possible points are $(1,1),(-1,-1)$, and indeed $(-1,-1)$ is the minimizer.

Is there any difference with $\geq$ inequality?