Scribd Logo
Upload

Welcome to Scribd!

  • Matrix Norm

    Uploaded by

    Anna Francesca Marcelo
    10 views3 pages
    The document defines a matrix norm ||A|| for a k x m matrix A as ||A|| = trace(AA^T) where A^T is the transpose of A. It then shows that this matrix norm satisfies the three properties needed to be a valid norm: 1) ||A|| ≥ 0 and ||A|| = 0 if and only if A is the zero matrix, 2) ||cA|| = |c|||A|| for any scalar c, and 3) the triangular inequality ||A + B|| ≤ ||A|| + ||B||. It also shows that ||A|| = ||A^T|| and that for conformable matrices A and B, ||AB|| ≤ ||A||

    Original Description:

    Copyright

    © Attribution Non-Commercial (BY-NC)

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    Report this Document
    The document defines a matrix norm ||A|| for a k x m matrix A as ||A|| = trace(AA^T) where A^T is the transpose of A. It then shows that this matrix norm satisfies the three properties needed to be a valid norm: 1) ||A|| ≥ 0 and ||A|| = 0 if and only if A is the zero matrix, 2) ||cA|| = |c|||A|| for any scalar c, and 3) the triangular inequality ||A + B|| ≤ ||A|| + ||B||. It also shows that ||A|| = ||A^T|| and that for conformable matrices A and B, ||AB|| ≤ ||A||

    Copyright:

    Attribution Non-Commercial (BY-NC)
    Download as PDF, TXT or read online from Scribd
    Flag for inappropriate content
    Download as pdf or txt
    10 views3 pages

    Matrix Norm

    Uploaded by

    Anna Francesca Marcelo
    The document defines a matrix norm ||A|| for a k x m matrix A as ||A|| = trace(AA^T) where A^T is the transpose of A. It then shows that this matrix norm satisfies the three properties needed to be a valid norm: 1) ||A|| ≥ 0 and ||A|| = 0 if and only if A is the zero matrix, 2) ||cA|| = |c|||A|| for any scalar c, and 3) the triangular inequality ||A + B|| ≤ ||A|| + ||B||. It also shows that ||A|| = ||A^T|| and that for conformable matrices A and B, ||AB|| ≤ ||A||

    Copyright:

    Attribution Non-Commercial (BY-NC)
    Download as PDF, TXT or read online from Scribd
    Flag for inappropriate content
    Download as pdf or txt
    of 3

    p The matrix norm ||A|| = trace (AA0)

    Herman Bierens November 16, 2009


    Let A be an k m matrix. Dene the matrix norm ||.|| by p ||A|| = trace (AA0 ) ||A|| = 0 if and only if A = O, ||c.A|| = |c|.||A|| for any scalar c, ||A + B|| ||A|| + ||B|| (triangular inequality),

    (1)

    Recall that a norm needs to satisfy three conditions

    (2) (3) (4)

    where of course in the latter case the matrix B has the same size as A. Conditions (2) and (3) follow trivially from (1), but condition (4) is not obvious and will be shown below. We also have ||A|| = ||A0 || (5)

    because trace(AA0 ) = trace(A0 A) . Moreover, it will be shown that for conformable matrices A and B, ||AB|| ||A||.||B|| and thus for b Rm , ||Ab|| ||A||.||b||. (6)

    Proof of (4): To check the triangular inequality, let B be a k m matrix. Then ||A + B||2 = trace ((A + B)(A0 + B 0 )) 1

    = = =

    trace(AA0 ) + trace(AB 0 ) + trace(BA0 ) + trace(BB 0 ) ||A||2 + 2.trace(B 0 A) + ||B||2 ||A||2 + 2.||A||.||B|| + ||B||2 (||A|| + ||B||)2

    which implies (4). The inequality follows from Schwarz inequality, applied twice, as follows. Let ai.j and bi,j be the typical elements of A and B, respectively. Then m k m k X X X 1 X bj,i ai,j k bj,i ai,j |trace(B 0 A)| = k j=1 i=1 j=1 i=1 v v v v u k u k u k m m u k X u1 X u1 X X uX uX t kt b2 t a2 = b2 t a2 k i=1 j,i k i=1 i,j j=1 i=1 j,i i=1 i,j j=1 v v u m k u m k uX X uX X p p t b2 t a2 = trace(BB 0 ) trace(AA0 ) j,i i,j
    j=1 i=1 j=1 i=1

    = ||B||.||A||

    Proof of (6): Let B be an m ` matrix. Then ||AB||2 = trace (ABB 0 A0 ) = trace ((BB 0 )(A0 A)) Pm 0 We can write BB 0 = i=1 i qi qi , where the i s are the (nonnegative) and orthonormal eigeneigenvalues of BB 0 P the qi s are the correspondingP m 0 0 0 vectors. Note that i=1 qi qi = Im and trace(BB ) = m i trace(qi qi ) = i=1 Pm Pm 0 i=1 i (qi qi ) = i=1 i . Then ||AB||2 = trace ((BB 0 )(A0 A)) m m X X 0 0 0 = i trace(qi qi A A) = i qi A0 Aqi
    i=1 m X i=1 i=1

    m X i=1

    0 qi A0 Aqi m X i=1

    = trace(BB 0 )

    0 trace(A0 Aqi qi )

    = trace(BB 0 ).trace A0 A

    = trace(BB 0 ).trace (A0 A) = trace(BB 0 ).trace (AA0 ) = ||B||2 .||A||2

    m X i=1

    0 qi qi

    You might also like