Row Operation is Equivalent to Pre-Multiplication by Product of Elementary Matrices
Theorem
Let $\map \MM {m, n}$ be a metric space of order $m \times n$ over a field $K$.
Let $\mathbf A \in \map \MM {m, n}$ be a matrix.
Let $\Gamma$ be a row operation which transforms $\mathbf A$ to a new matrix $\mathbf B \in \map \MM {m, n}$.
Then there exists a unique invertible square matrix $\mathbf R$ of order $m$ such that:
- $\mathbf R \mathbf A = \mathbf B$
where $\mathbf R$ is the product of a finite sequence of elementary row matrices.
Proof
The proof proceeds by induction.
By definition, $\Gamma$ is a finite sequence of elementary row operations on $\mathbf A$.
Let $\sequence e_k$ denote a finite sequence of elementary row operations $\tuple {e_1, e_2, \ldots, e_k}$ applied on $\mathbf A$ in order: first $e_1$, then $e_2$, then $\ldots$, then $e_k$.
Let $\Gamma_k$ be the row operation which consists of $\sequence e_k$.
Let $\mathbf E_k$ denote the elementary row matrix of order $m$ formed by applying $e_k$ to the unit matrix $I_m$.
For all $r \in \Z_{>0}$, let $\map P r$ be the proposition:
- For all $\Gamma_r$, there exists a unique invertible square matrix $\mathbf R_r$ of order $m$ such that:
- $\mathbf R_r \mathbf A = \mathbf B_r$
- where:
- $\Gamma_r$ is a row operation which transforms $\mathbf A$ to a new matrix $\mathbf B_r \in \map \MM {m, n}$.
- $\mathbf R_r$ is the product of the finite sequence of elementary row matrices:
- $\mathbf R_r = \mathbf E_r \mathbf E_{r - 1} \dotsb \mathbf E_2 \mathbf E_1$
Basis for the Induction
$\map P 1$ is the case where $\Gamma_1$ is a single-term sequence consisting of one elementary row operation $e_1$.
Let $e_1$ be an elementary row operation operating on $\mathbf A$, which transforms $\mathbf A$ into $\mathbf B_1$.
By definition, there exists exactly one elementary row matrix $\mathbf E_1$ of order $m$ such that $\mathbf E_1$ is the result of applying $e_1$ to the unit matrix $\mathbf I$ of order $m$.
From the corollary to Elementary Row Operations as Matrix Multiplications:
- $\mathbf E_1 \mathbf A = \mathbf B_1$
By Elementary Row Matrix is Invertible, $E_1$ is invertible.
Thus $\map P 1$ is seen to hold.
This is the basis for the induction.
Induction Hypothesis
Now it needs to be shown that if $\map P k$ is true, where $k \ge 1$, then it logically follows that $\map P {k + 1}$ is true.
So this is the induction hypothesis:
- For all $\Gamma_k$, there exists a unique invertible square matrix $\mathbf R_k$ of order $m$ such that:
- $\mathbf R_k \mathbf A = \mathbf B_k$
from which it is to be shown that:
- For all $\Gamma_{k + 1}$, there exists a unique invertible square matrix $\mathbf R_{k + 1}$ of order $m$ such that:
- $\mathbf R_{k + 1} \mathbf A = \mathbf B_{k + 1}$
Induction Step
This is the induction step:
By definition, $\Gamma_{k + 1}$ is a row operation consisting of a finite sequence of elementary row operations $\tuple {e_1, e_2, \ldots, e_k, e_{k + 1} }$ applied on $\mathbf A$ in order.
Thus $\Gamma_{k + 1}$ consists of the finite sequence of elementary row operations $\tuple {e_1, e_2, \ldots, e_k}$ applied on $\mathbf A$ in order, followed by a further elementary row operation $e_{k + 1}$.
By the induction hypothesis, there exists a unique invertible square matrix $\mathbf R_k$ of order $m$ such that:
- $\mathbf R_k \mathbf A = \mathbf B_k$
where $\mathbf B_k \in \map \MM {m, n}$ is the result of applying $\sequence e_k$ to $\mathbf A$ in order.
Let $e_{k + 1}$ be applied to $\mathbf B_k$.
By definition, there exists exactly one elementary row matrix $\mathbf E_{k + 1}$ of order $m$ such that $\mathbf E_{k + 1}$ is the result of applying $e_{k + 1}$ to the unit matrix $\mathbf I$ of order $m$.
Then:
\(\ds \mathbf B_{k + 1}\) | \(=\) | \(\ds \mathbf E_{k + 1} \mathbf B_k\) | Corollary to Elementary Row Operations as Matrix Multiplications | |||||||||||
\(\ds \) | \(=\) | \(\ds \mathbf E_{k + 1} \paren {\mathbf R_k \mathbf A}\) | ||||||||||||
\(\ds \) | \(=\) | \(\ds \paren {\mathbf E_{k + 1} \mathbf R_k} \mathbf A\) | Matrix Multiplication is Associative |
By Product of Matrices is Invertible iff Matrices are Invertible, $\mathbf E_{k + 1} \mathbf R_k$ is invertible.
We have that $\mathbf R_k$ is the unique invertible square matrix resulting from the application of $\sequence e_k$ on $\mathbf I_m$.
Thus $\mathbf E_{k + 1} \mathbf R_k$ is the unique invertible square matrix resulting from the application of $\sequence e_{k + 1}$ on $\mathbf I_m$.
So $\map P k \implies \map P {k + 1}$ and the result follows by the Principle of Mathematical Induction.
Therefore, for every row operation $\Gamma$ which transforms $\mathbf A$ to a new matrix $\mathbf B \in \map \MM {m, n}$, there exists a unique invertible square matrix $\mathbf R$ of order $m$ such that:
- $\mathbf R \mathbf A = \mathbf B$
where:
- $\mathbf R$ is the product of a finite sequence of elementary row matrices.
$\blacksquare$
Also see
Sources
- 1998: Richard Kaye and Robert Wilson: Linear Algebra ... (previous) ... (next): Part $\text I$: Matrices and vector spaces: $1$ Matrices: $1.5$ Row and column operations