When one writes

$$f(\lambda x + (1-\lambda)y) \le \lambda f(x) + (1-\lambda)f(y)$$

for $x,y\in \mathbb{R}^n$, $\lambda\in(0,1)$ what does this mean?

1) Does it mean that the function is jointly convex in $x = (x_1,\ldots,x_n)$ or convex in each argument $x_i$, $i=1,2,\ldots,n$?

2) Furthermore, what is the difference between joint convexity and element-wise convexity?

3) Is showing the above inequality the same as showing the Hessian is positive semi-definite?

  • 2
    $\begingroup$ Joint convexity means that the restriction of the function to any segment is a convex function of a single variable. Element-wise convexity means that only vertical and horizontal segments are considered (in 2d - in higher dimensions there are more directions but the idea is the same) $\endgroup$ Apr 3 '14 at 20:15

1) Jointly convex.

2) Element-wise convexity works for one variable only, other variables are supposed to be constant; joint convexity implies element-wise convexity, but the other implication is false. Consider a function $x^2+y^2-4xy$.

3) If the function is twice differentiable. For example, $f(x)=\|x\|$ is convex, but has troubles with Hessian in zero.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.