Automatic Control (AC)
Question 3: Optimization
August 2016
1.(20 pts) In some of the optimization methods, when minimizing a given function f(x), we select an intial guess $ x^{(0)} $ and a real symmetric positive definite matrix $ H_{0} $. Then we computed $ d^{(k)} = -H_{k}g^{(k)} $, where $ g^{(k)} = \nabla f( x^{(k)} ) $, and $ x^{(k+1)} = x^{(k)} + \alpha_{k}d^{(k)} $, where
$ \alpha_{k} = arg\min_{\alpha \ge 0}f(x^{(k)} + \alpha d^{(k)}) . $
Suppose that the function we wish to minimize is a standard quadratic of the form,
$ f(x) = \frac{1}{2} x^{T} Qx - x^{T}b+c, Q = Q^{T} > 0. $
(i)(10 pts) Find a closed form expression for $ \alpha_k $ in terms of $ Q, H_k, g^{(k)} $, and $ d^{(k)}; $
(ii)(10 pts) Give a sufficient condition on $ H_k $ for $ \alpha_k $ to be positive.
- Click here to view student answers and discussions
