QR Factorization - Linear Algebra

  1. What condition on the columns of AA guarantees a QR factorization with an invertible RR?
    The columns of AA must be linearly independent, so Gram-Schmidt never produces a zero vector.
  2. What is RR in the QR factorization of A=[[1,1],[1,−1]]A = [[1, 1], [1, -1]] (rows), with positive diagonal?
    R=[[2,0],[0,2]]R = [[\sqrt{2}, 0], [0, \sqrt{2}]]. The columns are orthogonal with norm 2\sqrt{2}, so r12=0r_{12} = 0.
  3. Why is RR upper triangular in A=QRA = QR built by Gram-Schmidt?
    Each column aj\mathbf{a}_j lies in the span of q1,…,qj\mathbf{q}_1, \dots, \mathbf{q}_j, so its components along later qi\mathbf{q}_i are zero.
  4. In A=QRA = QR, the columns of AA are a1=(2,0,0)\mathbf{a}_1 = (2, 0, 0) and a2=(3,1,0)\mathbf{a}_2 = (3, 1, 0). What is r12r_{12}?
    33. Here q1=(1,0,0)\mathbf{q}_1 = (1, 0, 0), so r12=q1⋅a2=3r_{12} = \mathbf{q}_1 \cdot \mathbf{a}_2 = 3.
  5. The first column of AA is (0,3,4)(0, 3, 4). In A=QRA = QR with positive diagonal in RR, what are q1\mathbf{q}_1 and r11r_{11}?
    q1=(0,35,45)\mathbf{q}_1 = (0, \frac{3}{5}, \frac{4}{5}) and r11=5r_{11} = 5. Here r11r_{11} is the norm of the column and q1\mathbf{q}_1 is the column divided by it.
  6. Which process applied to the columns of AA produces the columns of QQ in A=QRA = QR?
    The Gram-Schmidt process, followed by normalizing each resulting vector.
  7. If the columns of AA are already orthonormal, what are QQ and RR in A=QRA = QR?
    Q=AQ = A and R=IR = I. Gram-Schmidt leaves the columns unchanged.
  8. How does A=QRA = QR simplify finding the least-squares solution of Ax=bA\mathbf{x} = \mathbf{b}?
    The normal equations reduce to Rx^=QTbR\hat{\mathbf{x}} = Q^T\mathbf{b}, which is solved by back substitution.
  9. In a Gram-Schmidt QR factorization, what is the entry rijr_{ij} of RR for i≤ji \le j?
    rij=qi⋅ajr_{ij} = \mathbf{q}_i \cdot \mathbf{a}_j, the component of column aj\mathbf{a}_j along qi\mathbf{q}_i.
  10. In the reduced QR factorization A=QRA = QR of an m×nm \times n matrix with independent columns, what are QQ and RR?
    QQ is m×nm \times n with orthonormal columns, and RR is n×nn \times n, upper triangular and invertible.
  11. Once QQ is known in A=QRA = QR, how is RR computed from AA?
    R=QTAR = Q^TA, because QTQ=IQ^TQ = I.
  12. A square invertible matrix A=QRA = QR has positive diagonal entries in RR. What is ∣det⁡A∣|\det A| in terms of RR?
    The product r11r22⋯rnnr_{11}r_{22}\cdots r_{nn}. det⁡Q=±1\det Q = \pm 1 and the determinant of triangular RR is its diagonal product.

Study these flashcards