Vol. 19, No. 2, 1966

Download this article
Download this article. For screen
For printing
Recent Issues
Vol. 344: 1  2
Vol. 343: 1  2
Vol. 342: 1  2
Vol. 341: 1  2
Vol. 340: 1  2
Vol. 339: 1  2
Vol. 338: 1  2
Vol. 337: 1  2
Online Archive
Volume:
Issue:
     
The Journal
About the journal
Ethics and policies
Peer-review process
 
Submission guidelines
Submission form
Editorial board
Officers
 
Subscriptions
 
ISSN 1945-5844 (electronic)
ISSN 0030-8730 (print)
 
Special Issues
Author index
To appear
 
Other MSP journals
Extreme copositive quadratic forms

Leonard Daniel Baumert

Vol. 19 (1966), No. 2, 197–204
Abstract

A real quadratic form Q = Q(x1,⋯,xn) is called copositive if Q(x1,⋯,xn) ≧ 0 whenever x1,⋯,xn ≧ 0. If we associate each quadratic form Q = ∑ qijxixj qij = qji (i,j = 1,⋯,n) with a point (q11,⋯,qnn,√2-q12,⋯,√2qn−1,n) of Euclidean n(n + 1)∕2 space then the copositive forms constitute a closed covex cone in this space. We are concerned with the extreme points of this cone. That is, with those copositive quadratic forms Q for which Q = Q1 + Q2 (with Q1, Q2 copositive) implies Q1 = aQ, Q2 = (1 − a)Q, 0 ≦ a ≦ 1. We show that if Q(x1,⋯,xn), n ≧ 2, is an extreme copositive quadratic form then for any index i(1 ≦ i ≦ n), Q has a zero u with u1⋯,un ≧ 0 where ui > 0. Further, using this fact, we establish an extension process whereby extreme forms in n variables can be used to construct extreme form in n′ variables, for all n′ > n.

Mathematical Subject Classification
Primary: 15.70
Secondary: 05.00
Milestones
Received: 5 April 1965
Published: 1 November 1966
Authors
Leonard Daniel Baumert