what's the condition of the complex convex problem?

80 Views Asked by At

In most books or papers, the convex problems are usually defined on the Real field. However, there are also a lot of works focus on the problem defined on the Complex field but also use the KKT conditions to solve the close-form solution.

My problem is how to judge that if a function with complex variable is convex function. As is known to all, for real variables, the first order condition is given by $f(y) \geqslant f(x)+\nabla f(x)^{T}(y-x)$

The second order condition is given by $\nabla^{2} f(x) \succeq 0$.

So, what's the conditions of the convex problem when vairable is complex? Thanks a lot in advance.

1

There are 1 best solutions below

1
On BEST ANSWER

Typically, most of the functions map the complex field to real field, i.e., $f: \mathbb{C}^n \mapsto \mathbb{R}$.

The first order condition for the complex field reads $f(y) \geq f(x) + 2 \Re\left\{ \nabla f(x)^H \left( y-x \right) \right\}$, where $(\cdot)^H$ is Hermitian transpose.

The second order condition for the complex field (and real field) reads same, i.e., $\nabla^2 f(x) \succeq 0$ for all $x$.


Reference:

Ali H. Sayed (2014), "Adaptation, Learning, and Optimization over Networks", Foundations and Trends® in Machine Learning: Vol. 7: No. 4-5, pp 311-801. http://dx.doi.org/10.1561/2200000051.