We can also use a linear algebra method called Gaussian elimination to solve the matrix equation
\begin{equation*}
\begin{pmatrix}
1 \amp -1 \amp -1 \\ R_1+R_4 \amp R_2 \amp 0\\ R_1+R_4 \amp 0 \amp R_3
\end{pmatrix}
\begin{pmatrix} I_1 \\ I_2 \\ I_3 \end{pmatrix} =
\begin{pmatrix} 0 \\ V_b \\ V_b \end{pmatrix}\text{.}
\end{equation*}
Gaussian elimination relies on two facts:
If we multiply one row of \(\mathbf{A}\) and the same row of \(\mathbf{v}\) by some constant \(q\text{,}\) the solutions for \(\mathbf{x}\) remain unchanged.
Any linear combination of two rows will give a new correct row.
Using the two rules above, our goal is to find an equivalent \(\mathbf{A}'\mathbf{x}=\mathbf{v}'\) such that
\begin{equation*}
\mathbf{A}' =
\begin{pmatrix}
1 \amp A_{12}' \amp A_{13}'\\
0 \amp 1 \amp A_{13}'\\
0 \amp 0 \amp 1
\end{pmatrix}\text{.}
\end{equation*}
Often, the first step in using Gaussian elimination is to divide the first row by \(A_{11}\text{.}\) In this case, both \(\mathbf{A}\) and \(\mathbf{v}\) remain unchanged since \(A_{11}=1\text{.}\) Next, we’ll subtract \(\left[A_{21} * \left(\text{first row of $\textbf{A}$ and $\bf{v}$}\right)\right]\) from row 2 to get
\begin{equation*}
\begin{pmatrix}
1 \amp -1 \amp -1 \\ 0 \amp R_2 + R_1 + R_4 \amp R_1 + R_4\\ R_1+R_4 \amp 0 \amp R_3
\end{pmatrix}
\begin{pmatrix} I_1 \\ I_2 \\ I_3 \end{pmatrix} =
\begin{pmatrix} 0 \\ V_b \\ V_b \end{pmatrix}\text{.}
\end{equation*}
Then, we’ll subtract \(\left[A_{31} * \left(\text{first row of $\textbf{A}$ and $\bf{v}$}\right)\right]\) from row 3 to get
\begin{equation*}
\begin{pmatrix}
1 \amp -1 \amp -1 \\ 0 \amp R_2 + R_1 + R_4 \amp R_1 + R_4\\ 0 \amp R_1 + R_4 \amp R_3 + R_1 + R_4
\end{pmatrix}
\begin{pmatrix} I_1 \\ I_2 \\ I_3 \end{pmatrix} =
\begin{pmatrix} 0 \\ V_b \\ V_b \end{pmatrix}\text{.}
\end{equation*}
If we now divide row 2 by \(A_{22}\text{,}\) we get
\begin{equation*}
\begin{pmatrix}
1 \amp -1 \amp -1 \\ 0 \amp 1 \amp \frac{R_1 + R_4}{R_2 + R_1 + R_4}\\ 0 \amp R_1 + R_4 \amp R_3 + R_1 + R_4
\end{pmatrix}
\begin{pmatrix} I_1 \\ I_2 \\ I_3 \end{pmatrix} =
\begin{pmatrix} 0 \\ \frac{V_b}{R_2 + R_1 + R_4} \\ V_b \end{pmatrix}
\end{equation*}
Subtracting \(\left[A_{32} * \left(\text{second row of $\textbf{A}$ and $\bf{v}$}\right)\right]\) from row 3 will eliminate \(A_{32}\) while leaving \(A_{31}=0\)
\begin{equation*}
\begin{pmatrix}
1 \amp -1 \amp -1 \\ 0 \amp 1 \amp \frac{R_1 + R_4}{R_2 + R_1 + R_4}\\ 0 \amp 0 \amp R_3 + R_1 + R_4 - \frac{\left(R_1 + R_4\right)^2}{R_2 + R_1 + R_4}
\end{pmatrix}
\begin{pmatrix} I_1 \\ I_2 \\ I_3 \end{pmatrix} =
\begin{pmatrix} 0 \\ \frac{V_b}{R_2 + R_1 + R_4} \\ V_b - \frac{V_b\left(R_1 + R_4\right)}{R_2 + R_1 + R_4} \end{pmatrix}\text{.}
\end{equation*}
Then, divide row 3 by \(A_{33}\) to get
\begin{equation*}
\begin{pmatrix}
1 \amp -1 \amp -1 \\ 0 \amp 1 \amp \frac{R_1 + R_4}{R_2 + R_1 + R_4}\\ 0 \amp 0 \amp 1
\end{pmatrix}
\begin{pmatrix} I_1 \\ I_2 \\ I_3 \end{pmatrix} =
\begin{pmatrix}
0 \\ \frac{V_b}{R_2 + R_1 + R_4} \\
\left[V_b - \frac{V_b\left(R_1 + R_4\right)}{R_2 + R_1 + R_4}\right] / \left[R_3 + R_1 + R_4
- \frac{\left(R_1 + R_4\right)^2}{R_2 + R_1 + R_4}\right] \end{pmatrix}
\end{equation*}
which simplifies to
\begin{equation*}
\begin{pmatrix}
1 \amp -1 \amp -1 \\ 0 \amp 1 \amp \frac{R_1 + R_4}{R_2 + R_1 + R_4}\\ 0 \amp 0 \amp 1
\end{pmatrix}
\begin{pmatrix} I_1 \\ I_2 \\ I_3 \end{pmatrix} =
\begin{pmatrix}
0 \\ \frac{V_b}{R_2 + R_1 + R_4} \\
\frac{V_b R_2}{R_2 + R_1 + R_4}\ \frac{R_3 + R_1 + R_4}{R_3^2+2 R_3\left(R_1 + R_4\right)} \end{pmatrix}\text{.}
\end{equation*}
Now, we can read off an expression for \(I_3\) and subsequently calculate \(I_1\) and \(I_2\)
\begin{align*}
I_3 \amp = \frac{V_b R_2}{R_2 + R_1 + R_4}\ \frac{R_3 + R_1 + R_4}{R_3^2+2 R_3\left(R_1 + R_4\right)} = 0.92\text{mA}\\
I_2 \amp = \frac{V_b}{R_2 + R_1 + R_4} - \frac{R_1 + R_4}{R_2 + R_1 + R_4}I_3 = 1.85\text{mA}\\
I_1 \amp = I_2 + I_3 = 2.77\text{mA}\text{.}
\end{align*}