www-ai.cs.tu-dortmund.de/LEHRE/VORLESUNGEN/NOPT/WS16/lectures/Ch14_Duality.pdf
Untitled
i = 1, . . . ,m
!j(x) = 0, j = 1, . . . , r
Constrained Optimization
TU Dortmund, Dr. Sangkyun Lee 3
All functions are smooth, and could be non-convex
The Lagrangian function:
è Constraint set C
L(x ;↵ [...] 15
Considerations:
1. When do optimal Lagrange multipliers exist ?
2. What is the relation between
3. When can we solve a dual instead of its primal, and obtain primal
solutions from the dual solutions …