MATLAB Optimization and Integration 优化与整合教程.pdf

Matlab – Optimization and Integration Paul Schrimpf January 14, 2009 Paul Schrimpf () Matlab – Optimization and Integration January 14, 2009 1 / 43 This lecture focuses on two ubiquitous numerical techiniques: 1 Optimization and equation solving Agents maximize utility / profits Estimation 2 Integration Expectations of the future or over unobserved variables Paul Schrimpf () Matlab – Optimization and Integration January 14, 2009 2 / 43 Optimization Want to solve a minimization problem: min f (x ) x Two basic approaches: 1 Heuristic methods search over x in some systematic way 2 Model based approaches use an easily minimized approximation to f () to guide their search First, x ∈ for intuition Then, x ∈ n Paul Schrimpf () Matlab – Optimization and Integration January 14, 2009 3 / 43 Section Search Form bracket x < x < x such that f (x ) < f (x ) and f (x ) 1 2 3 2 1 3 Try new x ∈ (x ,x ), update bracket 1 3 Paul Schrimpf () Matlab – Optimization and Integration January 14, 2009 4 / 43 Quadratic Interpolation More general interpolation methods possible e.g. Matlab uses both quadratic and cubic interpolation for line search Paul Schrimpf () Matlab – Optimization and Integration January 14, 2009 5 / 43 Newton-Rhapson and Quasi-Newton Newton: use f (x ) and f (x ) to construct parabola