Assignment 5 - Nonlinear Optimization - Fall 2003 | ISYE 6663, Assignments of Linear Algebra

Material Type: Assignment; Class: Nonlinear Optimization; Subject: Industrial & Systems Engr; University: Georgia Institute of Technology-Main Campus; Term: Spring 2003;

Typology: Assignments

Pre 2010

Uploaded on 08/05/2009

koofers-user-hn7-1
koofers-user-hn7-1 🇺🇸

10 documents

1 / 1

Toggle sidebar

This page cannot be seen from the preview

Don't miss anything!

bg1
ISyE 6663 Optimization III
Spring 2003
Assignment 5
Issued: March 13, 2003
Due: March 20, 2003
Problem 1
Nocedal and Wright, Problem 5.1
Problem 2
Nocedal and Wright, Problem 5.4
Problem 3
Nocedal and Wright, Problem 5.11
Problem 4
Consider a symmetric positive definite matrix QRn×n, and the associated norm xQ:= xTQx.
Consider Q-conjugate directions d0,d
1,...,d
n1Rngenerated from linearly independent vectors
p0,p
1,...,p
n1Rn. Show that, for each k=1,...,n1, dk=pkˆpk, where ˆpkis the projection
of pkonto the subspace spanned by p0,...,p
k1(or the subspace spanned by d0,...,d
k1)with
respect to the ·
Q-norm, that is,
ˆpk=argmin{pkpQ:p[p0,...,p
k1]}
That is, dkis the part of pkthat remains after we subtract the projection of pkonto the subspace
spanned by p0,...,p
k1.

Partial preview of the text

Download Assignment 5 - Nonlinear Optimization - Fall 2003 | ISYE 6663 and more Assignments Linear Algebra in PDF only on Docsity!

ISyE 6663 Optimization III

Spring 2003

Assignment 5

Issued: March 13, 2003

Due: March 20, 2003

Problem 1 Nocedal and Wright, Problem 5.

Problem 2 Nocedal and Wright, Problem 5.

Problem 3 Nocedal and Wright, Problem 5.

Problem 4 Consider a symmetric positive definite matrix Q ∈ Rn×n, and the associated norm ‖x‖Q :=

xT^ Qx. Consider Q-conjugate directions d 0 , d 1 ,... , dn− 1 ∈ Rn^ generated from linearly independent vectors p 0 , p 1 ,... , pn− 1 ∈ Rn. Show that, for each k = 1,... , n − 1, dk = pk − pˆk, where ˆpk is the projection of pk onto the subspace spanned by p 0 ,... , pk− 1 (or the subspace spanned by d 0 ,... , dk− 1 ) with respect to the ‖ · ‖Q-norm, that is,

pˆk = arg min {‖pk − p‖Q : p ∈ [p 0 ,... , pk− 1 ]}

That is, dk is the part of pk that remains after we subtract the projection of pk onto the subspace spanned by p 0 ,... , pk− 1.