"Anatomy" of Dimensions

\( A^{m \times n} \)

\[ \begin{bmatrix} a_{11} & a_{12} & \cdots & a_{1n} \\ a_{21} & a_{22} & \cdots & a_{2n} \\ \vdots & & & \vdots \\ a_{m1} & a_{m2} & \cdots & a_{mn} \end{bmatrix} \]

\( m \) rows

\( n \) columns

"Anatomy" of Dimensions

\( A^{m \times n} \longrightarrow \dashrightarrow R \) ((R)REF of \( A \))

Staircase pattern of leading 1s with asterisks to the right and zeros below in a bracketed matrix
Figure 1
A large hand-drawn pair of square brackets encloses a schematic echelon-form matrix. Inside, a descending staircase of dashed step lines runs from the upper left to the lower right. On each step a leading entry 1 is written, followed by asterisks to its right: the first step shows 1 then one asterisk, the second step shows 1 then one asterisk, the third step shows 1 then two asterisks, the fourth step shows 1 then one asterisk, and the fifth, lowest step shows 1 then one asterisk near the right bracket. Below the entire staircase, in the open region at the bottom of the matrix, is written the label O's, indicating that all entries beneath the staircase are zeros.

"Anatomy" of Dimensions

\( A^{m \times n} \longrightarrow \dashrightarrow R \) ((R)REF of \( A \))

Echelon matrix with circled leading 1s, asterisks, zeros below; red arrows to pivot variables, magenta to free variables
Figure 2
A large bracketed matrix shown in row echelon form. Five leading entries equal to 1, each circled in red, descend in a staircase from the top-left toward the bottom-right. Immediately to the right of each circled 1 are asterisks representing arbitrary nonzero entries; the third circled 1 has two asterisks to its right. Dashed horizontal lines trace the staircase boundary beneath each leading 1. Below and to the left of the staircase the region is labeled with a handwritten 0 followed by an apostrophe and the letter s, indicating zeros. Red curved arrows run from the circled leading 1s down to the lower-left where the red label reads pivot variables. Magenta curved arrows run from the asterisk columns down to the lower-right where the magenta label reads free variables. Above the matrix is the expression \( A^{m \times n} \) with a solid arrow, then a dashed arrow, leading to \( R \), annotated as (R)REF of \( A \).

pivot variables

free variables

"Anatomy" of Dimensions

\[ \begin{gathered} \text{Rank}(A) = \#\text{ of pivot variables} \\[0.5em] \text{Nullity}(A) = \#\text{ of free variables} \end{gathered} \]

"Anatomy" of Dimensions

  1. \[ \begin{gathered} \text{Col}(A) = \text{Column space of } A \\[0.5em] = \text{Span}\{\text{columns of } A\} \subseteq \mathbb{R}^m \\[0.5em] \dim(\text{Col}(A)) = \#\text{ of pivots} = \text{Rank} \end{gathered} \]
  2. \[ \begin{gathered} \text{Null}(A) = \{X : AX = \vec{0}\} \subseteq \mathbb{R}^n \\[0.5em] \dim(\text{Null}(A)) = \#\text{ of free var.} = \text{Nullity} \end{gathered} \]

"Anatomy" of Dimensions

  1. \[ \begin{gathered} \text{Row}(A) = \text{Row Space of } A \\[0.5em] = \text{Span}\{\text{rows of } A\} \subseteq \mathbb{R}^n \end{gathered} \]

If \( A \xrightarrow{\ \text{elementary row operations}\ } \dashrightarrow B \) then \( \text{Row}(A) = \text{Row}(B) \)

\[ A \longrightarrow \longrightarrow \longrightarrow \dashrightarrow R \quad \text{((R)REF of } A\text{)} \]

We have

\[ \text{Row}(A) = \text{Row}(R) \]

"Anatomy" of Dimensions

  1. \[ \begin{gathered} \text{Row}(A) = \text{Row Space of } A \\[0.5em] = \text{Span}\{\text{rows of } A\} \\[0.5em] = \text{Span}\{\text{rows of } R\} \subseteq \mathbb{R}^n \\[0.5em] \dim(\text{Row}(A)) = \#\text{ of non-zero rows of } R \\[0.5em] = \#\text{ of pivots} \\[0.5em] = \text{Rank}(A) \quad (= \dim(\text{Col}(A))) \end{gathered} \]
\[ \dim(\text{Col}(A)) = \dim(\text{Row}(A)) \]

Relationship to Lin Ind.

Given \( A^{m \times n} \),

  1. Columns of \( A \) are lin. ind.

    \[ \begin{gathered} \Longleftrightarrow \text{No free variables} \\[0.5em] \Longleftrightarrow \text{Rank}(A) = n \end{gathered} \]
  2. Rows of \( A \) are lin. ind

    \[ \begin{gathered} \Longleftrightarrow \text{No zero rows in } R \\[0.5em] \Longleftrightarrow \text{Rank}(A) = m \end{gathered} \]

Relationship to Solving AX=B

Given \( A^{m \times n} \),

  1. \( AX = B \) is solvable for any \( B \)

    \[ \begin{gathered} \Longleftrightarrow \text{no zeros row in } R \\[0.5em] \Longleftrightarrow \text{Rank}(A) = m \end{gathered} \]
  2. \( AX = B \) has unique solution (if exists)

    \[ \begin{gathered} \Longleftrightarrow \text{no free variables, i.e. only pivots} \\[0.5em] \Longleftrightarrow \text{Rank}(A) = n \end{gathered} \]

Non-Singular Matrix

\( A^{m \times n} \) is called a non-singular matrix if for any \( B \), \( AX = B \) has a unique solution

\[ \begin{gathered} \Longrightarrow \quad \text{Rank}(A) = m \text{ and } \text{Rank}(A) = n \\[0.5em] \Longrightarrow \quad \underline{m = n}, \text{ i.e. } \boxed{A \text{ is a square matrix}} \end{gathered} \]

Non-Singular Matrix

\( A^{m \times n} \) is called a non-singular matrix if for any \( B \), \( AX = B \) has a unique solution

\[ \begin{gathered} \Longrightarrow \quad \text{Rank}(A) = m \text{ and } \text{Rank}(A) = n \\[0.5em] \Longrightarrow \quad \underline{m = n}, \text{ i.e. } \boxed{A \text{ must be a square matrix.}} \end{gathered} \]

Non-Singular Matrix (Given \( A^{n \times n} \))

The following statements are equivalent:

  1. \( A \) is non-singular;
  2. \( \text{Null}(A) = \{0\} \), ie. no free variables;
  3. \( AX = B \) has at most one solution;
  4. \( AX = B \) is solvable for any \( B \);
  5. \( A \) can be row reduced to \( I \)

Non-Singular Matrix (Given \( A^{n \times n} \))

The following statements are equivalent:

  1. \( \text{Rank}(A) = n \); \( (\dim(\text{Col}(A)) = n) \)
  2. Columns of \( A \) are lin ind;
  3. Columns of \( A \) span \( \mathbb{R}^n \)
  4. \( \text{Col}(A) = \mathbb{R}^n \)
  5. Columns of \( A \) form a basis of \( \mathbb{R}^n \)

Non-Singular Matrix (Given \( A^{n \times n} \))

The following statements are equivalent:

  1. Rows of \( A \) are lin ind;
  2. Rows of \( A \) span \( \mathbb{R}^n \)
  3. \( \text{Row}(A) = \mathbb{R}^n \)
  4. Rows of \( A \) form a basis of \( \mathbb{R}^n \)