Extreme point of convex function

69 Views Asked by At

For twice differentiable convex function $f(x)$ on $\mathbb{R}^n$, with $x=\langle x_1,x_2,...,x_n \rangle$,

prove that, $\forall x \in \mathbb{R}^n$ and $\forall i \in \{1,2,...,n\}$, if $\frac{\partial f(x) }{\partial{} ~x_i} < 0$, then

$$x^*_i \ge x_i$$

where $x^*=\langle x^*_1,x^*_2,...,x^*_n \rangle$ is the extreme point (global minimum) of convex function $f$.

1

There are 1 best solutions below

0
On

enter image description here

A counterexample for two dimension convex function, picture shows the contour lines, where all rings are perfect circle!

A concrete example would be $f(x,y)=x^{2}+\left(y+x\right)^{2}$