Scribd
Upload
188 views

Principal Sub Matrix

The document discusses conditions for determining if a function has a local maximum or minimum at a point. It states that for an interior point: 1) If the derivative is 0 and the Hessian is negative/positive semidefinite, then the function has a local maximum/minimum. 2) If the derivative is 0 and the Hessian is negative/positive definite, then the function has a strict local maximum/minimum. It then discusses using the signs of principal minors of the Hessian matrix to determine if the Hessian is definite, semidefinite, or indefinite.

Uploaded by

kaliman2010
Copyright
© Attribution Non-Commercial (BY-NC)
Available Formats
Download as PDF, TXT or read online on Scribd
188 views

Principal Sub Matrix

The document discusses conditions for determining if a function has a local maximum or minimum at a point. It states that for an interior point: 1) If the derivative is 0 and the Hessian is negative/positive semidefinite, then the function has a local maximum/minimum. 2) If the derivative is 0 and the Hessian is negative/positive definite, then the function has a strict local maximum/minimum. It then discusses using the signs of principal minors of the Hessian matrix to determine if the Hessian is definite, semidefinite, or indefinite.

Uploaded by

kaliman2010
Copyright
© Attribution Non-Commercial (BY-NC)
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

Handout on Second Order Conditions 10/12/04

Theorem 1 (Second Order Conditions) Suppose f is a C 2 function on Z Rn , and x is an interior point of Z. If f has a local maximum (respectively, minimum) at x , then Dx (f ) is zero and Hf (x ) is negative (respectively, positive) semidenite. Conversely, if Dx (f ) is zero and Hf (x ) is negative (respectively, positive) denite, then f has a strict local maximum (respectively, minimum) at x . The above theorem gives almost necessary and sucient conditions for an interior optimum. Almost because no conclusions can be drawn if the Hessian is semidenite but not denite. Besides working out quadratic forms, there is another simple algorithm for testing the deniteness of a symmetric matrix like the Hessian. First, we need some denitions: Denition 2 Let A be an n n matrix. A k k submatrix of A formed by deleting n k rows of A, and the same n k columns of A, is called principal submatrix of A. The determinant of a principal submatrix of A is called a principal minor of A. Note that the denition does not specify which n k rows and columns to delete, only that their indices must be the same. Example 3 For a general 3 3 matrix, a11 a12 a13 A = a21 a22 a23 a31 a32 a33

there is one third order principal minor, namely |A|. There are three second order principal minors: a11 a12 a21 a22 , formed by deleting column 3 and row 3; a11 a13 a31 a33 , formed by deleting column 2 and row 2; a22 a23 a32 a33 , formed by deleting column 1 and row 1 1

And there are three rst order principal minors: a11 , formed by deleting the last two rows and columns a , formed by deleting the rst and third rows and columns 22 a33 , formed by deleting the rst two rows and columns

Denition 4 Let A by an n n matrix. The k th order principal submatrix of A obtained by deleting the last n k rows and columns of A is called the kth order leading principal submatrix of A, and its determinant is called the k th order leading principal minor of A. We will denoted the kth order leading principal submatrix of A by Ak , and its k th order leading principal minor by |Ak | . Now, the algorithm for testing the deniteness of a symmetric matrix: Theorem 5 Let A be an n n symmetric matrix. Then, 1. (a) A is positive denite if and only if all its n leading principal minors are (strictly) positive. (b) A is negative denite if and only if its n leading princiapl minors alternate in sign as follows: |A1 | < 0, |A2 | > 0, |A3 | < 0, etc. (c) If some kth order leading principal minor of A is nonzero but does not t either of the above sign patterns, then A is indenite. One particular failure of this algorithm occurs when some leading principal minor is zero, but the others t one of the patterns above. In this case, the matrix is not denite, but may or may not be semidenite. In this case, we must unfortunately check not only the principal leading minors, but every principal minor. Theorem 6 Let A be an n n symmetric matrix. Then, A is positive semidenite if and only if every principal minor of A is 0. A is negative semidenite if and only if every principal minor of odd order is 0 and every principal minor of even order is 0.

You might also like