If $A$ be an $n\times n$ positive definite matrix, then there exists a unique positive definite matrix $B$ such that $B^2=A$.
My question is how to get this $B$. What is the name of the algorithm for finding $B$?
If $A$ be an $n\times n$ positive definite matrix, then there exists a unique positive definite matrix $B$ such that $B^2=A$.
My question is how to get this $B$. What is the name of the algorithm for finding $B$?
Copyright © 2021 JogjaFile Inc.
The positive definite matrix $A$ is diagonalizable. Thus, it can be written in the form $A = V \cdot D\cdot V^{-1}$, where $D$ is a diagonal matrix with the positive eigenvalues on the main diagonal and $V$ contains $n$ linearly independent right hand eigenvectors as its columns. Actually, we can consider the matrix $V$ to be an orthogonal one (i.e. $V^{-1} = V^T$).
So: $$B=A^{1/2} = V\cdot D^{1/2}\cdot V^{-1} = V\cdot D^{1/2}\cdot V^T$$
In that case, $B$ is a positive definite matrix. Indeed, consider any vector $x\neq \mathbf 0$. Then: $$x^TBx=x^T VD^{1/2}V^Tx = (x^T V) \cdot D^{1/2} \cdot (x^TV)^T >0, $$ due to $D^{1/2}$ is a positive definite matrix.
Yes, this method uses diagonalization of matrix $A$.