Show that the inverse of a triangular matrix is triangular

184 Views Asked by At

Consider the system of $d$ equations contained in the rows of $R\bar{x} = \bar{e_k}$ for the $d × d$ upper-triangular matrix $R$, where $\bar{e_k}$ is a $d$-dimensional column vector with a single value of $1$ in the $kth$ entry and $0$ in all other entries. Discuss why solving for $\bar{x} = [x_1 \cdots x_d]^T$ is simple in this case by solving for the variables in the order $x_d, x_{d−1}, \dots, x_1$. Furthermore, discuss why the solution for $R\bar{x} = \bar{e_k}$ must satisfy $x_i = 0$ for $i > k$. Why is the solution $x$ equal to the $kth$ column of the inverse of $R$? Discuss why the inverse of $R$ is also upper-triangular.

I am confused by the first part of the question and don't know where to start. For the last two questions, my answer for Why is the solution $x$ equal to the $kth$ column of the inverse of $R$? is that solving the equation $\bar{x}=R^{-1}\bar{e_k}$, all the columns of $R^{-1}$ would be multiplied by the scalars $ith$ entries $0$ of $\bar{e_k}$ except for the $kth$ entry of $\bar{e_k}$ where it is equal to $1$. For the last part Discuss why the inverse of R is also upper-triangular. The polynomials and inverses of triangular matrices are also triangular matrices of the same type.

1

There are 1 best solutions below

0
On

It is all about definitions. Any matrix is equivalent to its reduced row echelon form. In order to guarantee that $R$ is invertible, none of its diagonal entries can be zero (See Exercise 1.6.9 in Hoffman and Kunze). Now performing elementary row operations (Only operation types 1 and 2. In other words, only scaling and shearing, no switching is needed) on $R$ will convert $R$ to $I$. Since each elementary row operation works on the upper part of $R$, it is equivalent to multiply an upper elementary matrix on the left side of $R$. Hence, $E_s \cdots E_1 R=I$, which implies $R^{-1}=E_s\cdots E_1$. The product of upper-triangular matrices is still upper-triangular.