It's not hard. There's not much theory behind reducing a 3x3, 4x4 5x5...9x9... All you have to keep in mind is that you want to eliminate as many leading terms as possible. I'll show you an example:
Let's start with:
[tex]
\left[ <br />
\begin{array}{ccc|c} <br />
0 & 2 & 3 & 8\\ <br />
2 & 3 & 1 & 5\\ <br />
1 & -1 & -2 & -5\\ <br />
\end{array} <br />
\right] [/tex]
What not to do: It's a bad idea here to add R2 or R3 to R1 since that will change the leading zero (in a1,1) to a 2 or a 1. So we do not change R1 yet. We can begin by switching R1 for R3, we want all leading terms to be in the first few rows, and the leading zeros to be at the bottom:
[tex]
\left[ <br />
\begin{array}{ccc|c} <br />
0 & 2 & 3 & 8\\ <br />
2 & 3 & 1 & 5\\ <br />
1 & -1 & -2 & -5\\ <br />
\end{array} <br />
\right] <br />
→ <br />
\left[ <br />
\begin{array}{ccc|c} <br />
1 & -1 & -2 & -5\\ <br />
2 & 3 & 1 & 5\\ <br />
0 & 2 & 3 & 8\\ <br />
\end{array} <br />
\right] <br />
→ <br />
\left[ <br />
\begin{array}{ccc|c} <br />
1 & -1 & -2 & -5\\ <br />
0 & 5 & 5 & 15\\ <br />
0 & 2 & 3 & 8\\ <br />
\end{array} <br />
\right] <br />
→ <br />
\left[ <br />
\begin{array}{ccc|c} <br />
1 & -1 & -2 & 5\\ <br />
0 & 1 & 1 & 3\\ <br />
0 & 2 & 3 & 8\\ <br />
\end{array} <br />
\right]<br />
→ <br />
\left[ <br />
\begin{array}{ccc|c} <br />
1 & -1 & -2 & 5\\ <br />
0 & 1 & 1 & 3\\ <br />
0 & 0 & 1 & 2\\ <br />
\end{array} <br />
\right]<br />
[/tex]
What I did after changing R1 and R3 was I subtracted 2*R1 from R2, this eliminated the leading 2 in a2,1.
I then multiplied R2 by [tex]\frac {1}{5}[/tex] After that I subtracted 2*R2 from R3, which eliminated the leading 2 in a3,2.
Do you see the gist of what is going on here? On most steps, I try to eliminate one of the entries on the bottom left-half of the matrix. I want to create a "triangular" shape so that I can have a rref for my matrix. Basically, I want to be able to just read off the answer.
Can you try to finish the Gauss-Jordan reduction on the matrix that I started?