3

Find the left and right polar decomposition of the matrix: \begin{align*} A = \left[\begin{matrix} 1 & 0 \\ 1 & 1\end{matrix}\right] \end{align*}

So a polar decomposition would be of the form $A = U\sqrt{A^\dagger A} = \sqrt{AA^\dagger}\,U$ for some unitary $U$.

I'm trying to take the square root by finding the spectral decomposition $\lbrace|\lambda\rangle\rbrace$ of $A^\dagger A$ and using: \begin{align*} \sqrt{A^\dagger A} = \sum_{\lambda}\sqrt{\lambda}|\lambda \rangle \langle \lambda | \end{align*} but since we have repeated eigenvalues I'm not sure how to proceed

Paradox
  • 659
  • Ok sorry that's embarrasing. To salvage this question i was wondering how you would find the square root of $A$ itself. – Paradox Mar 10 '14 at 04:27

1 Answers1

0

A brute force (numerical) method for obtaining the right polar decomposition has been presented in this answer at MSE. When applied to the OP's problem (Test3) we get an outcome like this: $$ \begin{bmatrix} 1.000000 & 0.000000 \\ 1.000000 & 1.000000 \end{bmatrix} = \\ \begin{bmatrix} 0.894427 & -0.447214 \\ 0.447214 & 0.894427 \end{bmatrix} \begin{bmatrix} 1.341641 & 0.447214 \\ 0.447214 & 0.894427 \end{bmatrix} $$ However, when searching for the number $0.447214$ on the internet, we find that it's approximately equal to What is 1 over square root of 5?. So let's do an educated guess: $$ \begin{bmatrix} 1 & 0 \\ 1 & 1 \end{bmatrix} = \begin{bmatrix} 2/\sqrt{5} & -1/\sqrt{5} \\ 1/\sqrt{5} & 2/\sqrt{5} \end{bmatrix} \begin{bmatrix} 3/\sqrt{5} & 1/\sqrt{5} \\ 1/\sqrt{5} & 2/\sqrt{5} \end{bmatrix} $$ Finding the left polar decomposition is left as an exercise to the reader.

EDIT.
However, the square root of a real-valued general symmetric positive definite $2 \times 2$ matrix can be calculated exactly: $$ \begin{bmatrix} a & b \\ b & c \end{bmatrix}^2 = \begin{bmatrix} A & B \\ B & C \end{bmatrix} \quad \Longrightarrow \\ \begin{bmatrix} a & b \\ b & c \end{bmatrix}\begin{bmatrix} a & b \\ b & c \end{bmatrix} = \begin{bmatrix} a^2+b^2 & b(a+c) \\ b(a+c) & b^2+c^2 \end{bmatrix} = \begin{bmatrix} A & B \\ B & C \end{bmatrix} $$ It follows that: $$ \begin{cases} a^2+b^2 = A \\ b(a+c) = B \\ b^2+c^2 = C \end{cases} \quad \Longrightarrow \quad \begin{cases} a^2 = A - b^2 \\ b^2(a^2+2ac+c^2) = B^2 \\ c^2 = C - b^2 \end{cases} $$ Using the determinants gives an additional equation: $$ (ac-b^2)^2 = AC-B^2 \quad \Longrightarrow \quad ac = \sqrt{AC-B^2}+b^2 $$ Making smart combinations already gives the end-result: $$ b^2(a^2+2ac+c^2) = B^2 \quad \Longrightarrow \\ b^2([A - b^2]+2[\sqrt{AC-B^2}+b^2]+[C - b^2]) = B^2 \quad \Longrightarrow \\ b^2(A+2\sqrt{AC-B^2}+C) = B^2 \quad \Longrightarrow \\ b = \frac{B}{\sqrt{A+2\sqrt{AC-B^2}+C}} $$ Together with $$ a = \sqrt{A-b^2} \quad ; \quad c = \sqrt{C-b^2} $$ So here comes a much simpler solution. First form the transpose times the original matrix: $$ \begin{bmatrix} 1 & 0 \\ 1 & 1 \end{bmatrix}^T\begin{bmatrix} 1 & 0 \\ 1 & 1 \end{bmatrix} = \begin{bmatrix} 1 & 1 \\ 0 & 1 \end{bmatrix} \begin{bmatrix} 1 & 0 \\ 1 & 1 \end{bmatrix} = \begin{bmatrix} 2 & 1 \\ 1 & 1 \end{bmatrix} = \begin{bmatrix} A & B \\ B & C \end{bmatrix} = \begin{bmatrix} a & b \\ b & c \end{bmatrix}^2 $$ With the above formulas: $$ b = \frac{B}{\sqrt{A+2\sqrt{AC-B^2}+C}} = \frac{1}{\sqrt{2 + 2\sqrt{1} + 1}} = 1/\sqrt{5} \\ a = \sqrt{A-b^2} = \sqrt{2 - 1/5} = 3/\sqrt{5} \\ c = \sqrt{C-b^2} = \sqrt{1 - 1/5} = 2/\sqrt{5} $$ So we have, indeed: $$ \sqrt{\begin{bmatrix} 1 & 0 \\ 1 & 1 \end{bmatrix}^T\begin{bmatrix} 1 & 0 \\ 1 & 1 \end{bmatrix}} = \begin{bmatrix} 3/\sqrt{5} & 1/\sqrt{5} \\ 1/\sqrt{5} & 2/\sqrt{5} \end{bmatrix} $$ At last, the orthogonal matrix is found with: $$ \begin{bmatrix} 1 & 0 \\ 1 & 1 \end{bmatrix} \begin{bmatrix} 3/\sqrt{5} & 1/\sqrt{5} \\ 1/\sqrt{5} & 2/\sqrt{5} \end{bmatrix}^{-1} = \begin{bmatrix} 1 & 0 \\ 1 & 1 \end{bmatrix} \begin{bmatrix} 2/\sqrt{5} & -1/\sqrt{5} \\ -1/\sqrt{5} & 3/\sqrt{5} \end{bmatrix} / 1 = \begin{bmatrix} 2/\sqrt{5} & -1/\sqrt{5} \\ 1/\sqrt{5} & 2/\sqrt{5} \end{bmatrix} $$ Which leads to the same conclusion: $$ \begin{bmatrix} 1 & 0 \\ 1 & 1 \end{bmatrix} = \begin{bmatrix} 2/\sqrt{5} & -1/\sqrt{5} \\ 1/\sqrt{5} & 2/\sqrt{5} \end{bmatrix} \begin{bmatrix} 3/\sqrt{5} & 1/\sqrt{5} \\ 1/\sqrt{5} & 2/\sqrt{5} \end{bmatrix} $$

Han de Bruijn
  • 17,070