Convex Programming with Linear Equality Constraints : Existence of Dual Optimal Solution

69 Views Asked by At

Let's consider following problem:

$$\text{minimize}\ \ \ \ f(x)\\\text{subject to}\ \ x\in X , \ Ax = b$$

Then I'd like to prove following claims:

1) If $f$ is finite and that there exists $\bar x \in ri(X)$ such that $A\bar x = b$. Then $f^* = q^*$ and there exist at least one dual optimal solution.

2) There holds $f^* = q^*$ and $(x^*, \lambda^*)$ are a primal and dual optimal solution pair iff if $x^*$ is feasible and $x^* \in argmin_{x\in X}L(x,\lambda^*)$


Any advice hint to start the proof ?