KKT Conditions and Optimization Techniques
Slide: This lecture begins with an introduction to the Karush-Kuhn-Tucker (KKT) conditions, which are fundamental in constrained optimization. We will explore their geometric interpretation, derive the necessary conditions for optimality, and discuss their applications in solving optimization problems.
Lagrangian Multiplier
Optimization with Equality Constraints
Consider an optimization problem with equality constraints:
how to characterize the necessary conditions for the optimal solution \(x^*\) ?
Geometric Insight
Consider a quadratic objective function with one linear equality constraint:
\(\nabla f( x^*)\) must lie within the linear subspace spanned by \(\{\nabla h_j({x}^*)\}\); otherwise, the function value could be further decreased by moving along a feasible direction. This means there exist multipliers \(\{\nu_j^*\}\) such that:
Lagrangian Function and Optimality Conditions
We introduce the Lagrangian function as a tool to characterize optimality:
The necessary conditions for optimality can be expressed as stationarity of the Lagrangian:
Lagrangian Relaxation
A Min-Max Interpretation
The Lagrangian function also leads to a powerful dual interpretation:
The original constrained problem is equivalent to the following min-max problem:
The Dual Problem and Weak Duality
Solution for \(\min_x\max_\nu\mathcal{L}(x,\nu)\) may be non-continuous, but solution for \(\max_\nu\min_x\mathcal{L}(x,\nu)\) is easy if \(\mathcal{L}(x,\nu)\) is tractable. We can form the dual problem by swapping the order of the min and the max:
Under some conditions, equality holds, which means strong duality holds.
Uzawa's Method (Dual Ascent)
This leads to Uzawa's Method:
- Minimization (\(x\)-step): $$ x{k+1}=\arg\min_x\mathcal{L}(x,\nuk) $$
- Ascent (\(\nu\)-step): $$ \nu^{k+1} = \nuk+\alphakh(x^{k+1}) $$ where \(\alpha^k>0\) is the step size.
KKT Conditions
General Constrained Optimization
Consider a general optimization problem with equality and inequality constraints:
the question does not change: how to characterize the necessary conditions for the optimal solution \(x^*\) ?
Geometric Insight
Challenge:
- Directionality: On the boundary, \(\nabla g_i\) points towards the exterior of the feasible region. To prevent \(f\) from pushing the point into an infeasible area, \(\nabla f\) must have a component opposite to \(\nabla g_i\) \(\Longrightarrow\) \(\mu_i\ge0\).
- Activity Identification: The optimum may lie in the interior of the region with \(g_i < 0\) or on the boundary with \(g_i = 0\) \(\Longrightarrow\) \(\mu_ig_i=0\).
Summary
- Stationarity: \(0\in\partial_{ x}[f( x)+\sum_i\mu_ig_i(x)+\sum_j\nu_jh_j(x)]\)
- Complementary Slackness: \(\mu_ig_i( x)=0\)
- Primal Feasibility: \(g_i(x)\le0,h_j(x)=0\)
- Dual Feasibility: \(\mu_i\ge0\)