This preview shows pages 1–7. Sign up to view the full content.
This preview has intentionally blurred sections. Sign up to view the full version.
View Full DocumentThis preview has intentionally blurred sections. Sign up to view the full version.
View Full DocumentThis preview has intentionally blurred sections. Sign up to view the full version.
View Full Document
Unformatted text preview: Lecture 6: Closed Functions February 5, 2007 Lecture 6 Outline I LogTransformation I Jensen’s Inequality I Level Sets I Closed Functions I Convexity and Continuity I Recession Cone and Constancy Space I Optimization Terminology Convex Optimization 1 Lecture 6 LogTransformation of Variables Useful for transforming a nonconvex function to a convex one A posynomial is a function of y 1 > ,...,y n > of the form g ( y 1 ,...,y n ) = by a 1 1 ··· y a n n with scalars b > and a i > for all i . I A posynomial need not be convex I Logtransformation of variables x i = ln y i for all i I We have convex function f ( x ) = be a 1 x 1 ··· e a n x n = be a x Convex Optimization 2 Lecture 6 LogTransformation of Functions Replacing f with ln f [when f ( x ) > over dom f ] Useful for: I Transforming nonseparable functions to separable ones Example : (Geometric Mean) f ( x ) = (Π n i =1 x i ) 1 /n for x with x i > for all i is nonseparable. Using F ( x ) = ln f ( x ) , we obtain a separable F , F ( x ) = 1 n n X i =1 ln x i I Separable structure of objective function is advantageous in distributed optimization Convex Optimization 3 Lecture 6 General Convex Inequality Basic convex inequality : For a convex f , we have for x,y ∈ dom f and α ∈ (0 , 1) , f ( αx + (1 α ) y ) ≤ αf ( x ) + (1 α ) f ( y ) General convex inequality : For a convex f and any convex combination of the points in dom f , we have f m X i =1 α i x i ! ≤ m X i =1 α i f ( x i ) ( x i ∈ dom f and α i > for all i , ∑ m i =1 α i = 1 , m > integer) A convex combination ∑ m i =1 α i x i can be viewed as the expectation of a random vector z having outcomes z = x i with probability α i Convex Optimization 4 Lecture 6 Jensen’s Inequality General convex inequality can be interpreted as: I For a convex f and a (finite) discrete random variable z with outcomes z i ∈ dom f , we have f ( E z ) ≤ E [ f ( z )] General Jensen’s inequality : The above relation holds for a convex f and,...
View
Full
Document
This note was uploaded on 08/22/2008 for the course GE 498 AN taught by Professor Angelianedich during the Spring '07 term at University of Illinois at Urbana–Champaign.
 Spring '07
 AngeliaNedich

Click to edit the document details