More Matrices

Eigenvalue/ Eigenvector

  • A
    : an
    n×n
    matrix,
    v
    : a vector
    Rn
    ,
    λ:
    a scalar
  • If there exist a nonzero vector
    v
    such that
    Av=λv
    , we say
    v
    is an eigenvector and
    λ
    is the eigenvalue for
    v
  • Fact: eigenvectors corresponding to distinct eigenvalues are linearly independent

Matrix Diagonalization

  • Fact: not all matrices are diagonalizable
  • If an
    n×n
    matrix
    A
    is diagonalizable, it can be decomposed as
    • A=PDP1
      • D
        is a diagonal matrix with
        λ1λn
        diagonal entries
      • P=[p1pn]
        is an invertible matrix
    • AP=PD
      • AP=[Ap1Apn]
      • PD=[λ1p1λnpn]
    • Api=λipi
      . Thus,
      pi
      is an eigenvector of
      A
      with eigenvalue
      λi
    • Because
      P
      is invertible, the column vectors of
      P
      (the eigenvectors of
      A
      ) are independent
    • An=PDnP1
      (computationally efficient!)

Symmetric Matrix

  • definition:
    A=A
  • Fact: symetric matrix is diagonalizable, and its eigenvectors can be chosen to form an orthonormal basis of
    Rn
  • A=UDU1
    (diagonalization)
    • U
      is an orthogonal matrix
      • its column vectors are orthonormal
      • UU=I=UU
      • U1=U
    • A=UDU
  • Quadratic form:
    xAx
    • A
      is a symmetric matrix,
      x
      is a vector
    • xAx=xUDUx
    • Let
      y=Ux
      ,
      xAx=(xU)D(Ux)=yDy=Σiλiyi2λmaxΣiyi2=λmaxyy=λmaxxx
      • proof.
        yy=(Ux)(Ux)=xUUx=xUU1x=xx
        (orthogonal maxtrix preserves norm)
      • λmax
        (resp.
        λmin
        ): the largest (resp. smallest) eigenvalues among all eigenvalues of
        A
    • λminxxxAxλmaxxx