Linear Algebra Lecture 8

Linear Algebra Lecture 8

1. Complete solution of $Ax=b$

2. Rank r

---

Example 1

$\begin{cases} x_1 + 2x_2 + 2x_3 + 2x_4 =b_1 \\ 2x_1 + 4x_2 + 6x_3 + 8x_4 = b_2\\ 3x_1 + 6x_2 + 8x_3 + 10x_4 = b_3\end{cases}$

Changed the equations to augmented matrix(增广矩阵)
Augmented matrix = $\begin{bmatrix}A&b\end{bmatrix}$

$\begin{bmatrix}1&2&2&2&b_1\\ 2&4&6&8&b_2 \\ 3&6&8&10&b_3\end{bmatrix}$

Then do elimination to the augmented matrix.

$\begin{bmatrix}1&2&2&2&b_1\\ 2&4&6&8&b_2 \\ 3&6&8&10&b_3\end{bmatrix} \to \begin{bmatrix}1&2&2&2&b_1\\ 0&0&2&4&b_2-2b_1 \\ 0&0&2&4&b_3-3b_1\end{bmatrix} \to \begin{bmatrix}1&2&2&2&b_1\\ 0&0&2&4&b_2-2b_1 \\ 0&0&0&0&b_3-b_2-b_1\end{bmatrix}$

After the elimination, we can find 2 pivot columns, column 1 and 3.
And for the last row we can find $0=b_3-b_2-b_1$, that’s the condition for solvability.

Suppose $b=\begin{bmatrix}1\\ 5 \\ 6\end{bmatrix}$, then

$\begin{bmatrix}1&2&2&2&b_1\\ 0&0&2&4&b_2-2b_1 \\ 0&0&0&0&b_3-b_2-b_1\end{bmatrix} \to \begin{bmatrix}1&2&2&2&1\\ 0&0&2&4&3 \\ 0&0&0&0&0\end{bmatrix}$

Solvability

What are the conditions on b that make the equation solvable?
$Ax=b$ is solvable exactly when b is in the column space of A.
If a combination of the rows of A gives the zero row, then the same combination of the components of b has to give zero.

Find complete solutions

Step 1, find a particular solution(特解), X particular.
Set all free variables to zero, then solve $Ax=b$ for the pivot variables.

$\begin{cases} x_1 + 2x_2 + 2x_3 + 2x_4 =1 \\ 2x_3 + 4x_4 = 3\end{cases} \to \begin{cases} x_1 + 2x_3 =1 \\ 2x_3 = 3\end{cases} \to x_p = \begin{bmatrix}-2\\0\\\frac{3}{2}\\0\end{bmatrix}$

The particular solution comes first check to have zero equals zero, and then set all the free variables to zero, solve for the pivot variables, then get a particular solution.

Step 2, add on null space. The complete solution is one particular solution plus all different vectors out of the null space.
$x = x_{particular} + x_{nullspace}$

$\begin{cases}Ax_p = b\\ Ax_n=0\end{cases} \to A(x_p+x_n) = b+0 =b \to x = x_p+x_n$

Special solutions in the null space take 1 and 0 in the free variables, then solve the pivot variables. The null space consists of all combinations of the special solutions.

$x_{complete} =\begin{bmatrix}-2\\0\\\frac{3}{2}\\0\end{bmatrix} + c_1\begin{bmatrix}-2\\1\\0\\0\end{bmatrix} + c_2\begin{bmatrix}2\\0\\-2\\0\end{bmatrix}$

Algorithm
The algorithm to solve $Ax=b$ is just go through elimination and find the particular solution, and then find those special solutions.

---

Rank r

$m$ by $n$ matrix $A$ of rank $r$ (know $r\le m$ , $r \le n$)

Full column rank means $r = n$, a pivot in every column, means there are $n$ pivot columns and no free variable.

If there are no free variables to give value, then the null space of $A$ is only the zero vector. $N(A)$={zero vector}

Solution x to $Ax=b$, if there is a solution, $x = x_{particular}$, it’s unique solution if it exists , 0 or 1 solutions.

Full row rank means $r = m$, every row has a pivot, get m pivots.
After do elimination, and don’t get any zero rows, so there aren’t any requirements on $b$, so can solve $Ax=b$ for every $b$.Left with $n-r=n-m$ free variables.

Square matrix with full rank $r=m=n$ , means the matrix is invertible.
The reduced row echelon form for an invertible matrix is the identity. $R=I$.
The null space is zero vector only, and can solve $Ax=b$ for every $b$.

---

$r=m=n$ , $R=I$, 1 solution to $Ax=b$

$r=n<m$, $R = \begin{bmatrix}I\\0\end{bmatrix}$, 0 or 1 solution to $Ax=b$

$r=m<n$, $R=\begin{bmatrix}I&F\end{bmatrix}$, $\infty$ solution to $Ax=b$

$r<m, r<n$, $R=\begin{bmatrix}I&F\\ 0 & 0\end{bmatrix}$,0 or $\infty$ solution to $Ax=b$

The rank r tells you everything about the number of solutions, except the exact entries in the solutions.