Skip to content

6.4-exam3-2007-johnson-b

Quiz 3, 2007 (johnson) · Section 6.4: Complex Numbers and Vectors and Matrices Source: exam3_1.pdf, pp. 122-123

Problem

3 (30 pts.) (a) Suppose \(A = e^{iB}\) where \(B\) is Hermitian; what is \(A^H A\)? Hence \(A\) is a __ matrix.

(b) For the recurrence relation \(\vec{f}_{k+1} = e^{iB} \vec{f}_k\), what is \(\|\vec{f}_k\|^2/\|\vec{f}_0\|^2\)? [Hint: part (a) is useful.]

(c) Compute \(\vec{f}_k\) explicitly [i.e. no matrix exponentials or powers of matrices] for \(B = \begin{bmatrix} -3 & 4 \\ 4 & 3 \end{bmatrix}\) and \(\vec{f}_0 = \begin{bmatrix} 1 \\ 0 \end{bmatrix}\). The eigenvectors of this \(B\) are \(\vec{x}_1 = \begin{bmatrix} 1 \\ 2 \end{bmatrix}\) and \(\vec{x}_2 = \begin{bmatrix} 2 \\ -1 \end{bmatrix}\) with eigenvalues \(\lambda_1 = 5\) and \(\lambda_2 = -5\), respectively.

(d) Check that your answer from (b) is true for your answer from (c).

Solution (a) \(A^H = e^{(iB)^H} = e^{-iB^H} = e^{-iB}\). Hence \(A^H A = e^{-iB} e^{iB} = e^{-iB + iB} = e^0 = I\). (Note that \(iB\) and \(-iB\) obviously commute, which is why we can combine the exponentials like this.) Hence \(A\) is unitary. (b) As in class, \(\vec{f}_k = A^k \vec{f}_0\). Hence \[ \|\vec{f}_k\|^2 = \vec{f}_k^H \vec{f}_k = \vec{f}_0^H (A^k)^H A^k \vec{f}_0 = \vec{f}_0^H \vec{f}_0 = \|\vec{f}_0\|^2 \] [using the result from part (a) to cancel the \(A^H A\) factors in the middle], and hence \(\|\vec{f}_k\|^2/\|\vec{f}_0\|^2 = 1\). Equivalently, the product of unitary matrices is unitary, so \(A^k\) is unitary, so it preserves lengths. (c) We first have to expand the initial condition in terms of the eigenvectors: \[ \vec{f}_0 = \begin{bmatrix} 1 \\ 0 \end{bmatrix} = \frac{\begin{bmatrix} 1 \\ 2 \end{bmatrix} + 2 \begin{bmatrix} 2 \\ -1 \end{bmatrix}}{5}. \] Then \(\vec{f}_k = A^k \vec{f}_0 = e^{iBk} \vec{f}_0\), multiplying each eigenvector by \(e^{i\lambda k}\): \[ \vec{f}_k = \frac{\begin{bmatrix} 1 \\ 2 \end{bmatrix} e^{i5k} + 2 \begin{bmatrix} 2 \\ -1 \end{bmatrix} e^{-i5k}}{5} = \frac{\begin{bmatrix} e^{i5k} + 4e^{-i5k} \\ 2e^{i5k} - 2e^{-i5k} \end{bmatrix}}{5}. \] (d) Using orthogonality to eliminate the cross terms, \[ \|\vec{f}_k\|^2 = \vec{f}_k^H \vec{f}_k = \frac{\left\| \begin{bmatrix} 1 \\ 2 \end{bmatrix} \right\|^2 |e^{i5k}|^2 + 2^2 \left\| \begin{bmatrix} 2 \\ -1 \end{bmatrix} \right\|^2 |e^{-i5k}|^2}{5^2} = \frac{5 + 4 \cdot 5}{25} = 1 = \|\vec{f}_0\|^2. \]

Keywords: matrix exponential, unitary matrix, Hermitian matrix, orthogonality, complex eigenvalues