6. Matrix Multiplication, Identity Matrices, and Vector Applications

Dimension Compatibility and Non-Commutativity in Matrix Multiplication

• Matrix Multiplication Compatibility:

  • Matrix multiplication $A \times B$ is defined only when the inner dimensions match, meaning the number of columns in matrix $A$ equals the number of rows in matrix $B$.

  • When calculating $A \times B$, row vectors from the left matrix $A$ are paired via dot products with column vectors from the right matrix B$.\n - **Example of compatible vs. incompatible dimension pairing:**\n - Multiplying row vectors of dimension 2$such as$(1, 2)$with two-element column vectors like$\begin{pmatrix} 7 \ 9 \end{pmatrix} is well-defined.\n - Conversely, attempting B \times A$with row vectors of length$2$like$(7, 8)$and column vectors of length$3$like$\begin{pmatrix} 1 \ 3 \ 5 \end{pmatrix}$fails because the dimensions do not match. Thus,$B \times A is undefined.\n\n- **Contextual Interpretation Shift:**\n - Swapping the order of matrix multiplication fundamentally alters the context in which the matrices are interpreted.\n - In A \times B$, matrix$A$is interpreted as a set of row vectors, while matrix$B is interpreted as a set of column vectors.\n - In B \times A$, matrix$B$is reinterpreted as row vectors and matrix$A as column vectors.\n - This contextual shift distinguishes matrix multiplication from scalar arithmetic (where switching number positions does not change their identity or meaning) and standard vector dot products (where \mathbf{u} \cdot \mathbf{v} = \mathbf{v} \cdot \mathbf{u} because vectors retain identical interpretations regardless of order).\n\n- **Non-Commutativity in Square Matrices:**\n - For square matrices, both A \times B$and$B \times A$are defined because all dimensions match, but the resulting products yield different values ($A \times B \neq B \times A).\n - **First Order Numerical Example:**\n - Taking the row vector (5, 1, 2)$dotted with the column vector$\begin{pmatrix} 5 \ 7 \end{pmatrix}$produces an output value of$19.\n - **Reversed Order Numerical Example:**\n - Taking the first row vector (5, 6)$dotted with the first column vector$\begin{pmatrix} 1 \ 3 \end{pmatrix}$produces$5 \times 1 + 6 \times 3 = 5 + 18 = 23.\n - **Conclusion:** Because 19 \neq 23, matrix multiplication is non-commutative.\n\n### Algebraic Properties and Identity Matrix Definition\n\n- **Property Analysis of Mathematical Operators:**\n - When constructing or evaluating a new mathematical operator, fundamental algebraic properties must be examined:\n - **Commutativity:** Evaluating whether A \times B = B \times A.\n - **Identity Existence:** Determining whether a universal identity element exists.\n - **Transitivity and Reflexivity:** Analyzing relational behavior where applicable.\n\n- **Formal Definition of the Identity Matrix (I):**\n - For a matrix A$, an identity matrix$I$satisfies the property$A \times I = A.\n - Identity matrices are defined strictly for square matrices (n \times n).\n - **Dimension Specificity:**\n - Each dimensional space requires a distinct identity matrix (denoted I_n$for dimension$n).\n - A single identity matrix cannot operate universally across different matrix dimensions (such as multiplying a 3 \times 2$matrix to return the exact same$3 \times 2 matrix) without matching the specific structural dimension.\n\n### Operational Mechanics: The Copying and Masking Framework\n\n- **Structure of the Two-Dimensional Identity Matrix (I_2):**\n - I_2 = \begin{pmatrix} 1 & 0 \ 0 & 1 \end{pmatrix}\n\n- **Dual Instructions of Identity Operations:**\n - The entry 1 functions as a Copy instruction (retaining the exact coordinate value).\n - The entry 0 functions as a Mask instruction (filtering out contribution so nothing is added to the dot product sum).\n\n- **Step-by-Step Multiplication Mechanics with Matrix \begin{pmatrix} 5 & 6 \ 7 & 8 \end{pmatrix}:**\n - Multiplying \begin{pmatrix} 1 & 0 \ 0 & 1 \end{pmatrix} \times \begin{pmatrix} 5 & 6 \ 7 & 8 \end{pmatrix}:\n - **Row 1 Evaluation ((1, 0)):**\n - Dot product with first column \begin{pmatrix} 5 \ 7 \end{pmatrix}$:$1 \times 5 + 0 \times 7 = 5 + 0 = 5$(The$1$copies the top row contribution$5$; the$0$masks the bottom row contribution$7).\n - Dot product with second column \begin{pmatrix} 6 \ 8 \end{pmatrix}$:$1 \times 6 + 0 \times 8 = 6 + 0 = 6$(The$1$copies top entry$6$; the$0$masks bottom entry$8).\n - **Row 2 Evaluation ((0, 1)):**\n - Dot product with first column \begin{pmatrix} 5 \ 7 \end{pmatrix}$:$0 \times 5 + 1 \times 7 = 0 + 7 = 7$(The$0$masks top entry$5$; the$1$copies bottom entry$7).\n - Dot product with second column \begin{pmatrix} 6 \ 8 \end{pmatrix}$:$0 \times 6 + 1 \times 8 = 0 + 8 = 8$(The$0$masks top entry$6$; the$1$copies bottom entry$8).\n - **Output Result:** Reconstructs the exact original matrix \begin{pmatrix} 5 & 6 \ 7 & 8 \end{pmatrix}.\n\n- **Data Processing Implications:**\n - Masking contributions from specific rows or columns while copying desired coordinates is a fundamental technique for data filtering, subsetting, and feature isolation.\n\n### Coordinate Selection and Trigonometric Connections\n\n- **Selection Conceptualization:**\n - Beyond copying and masking, the entry 1 acts as a coordinate selector.\n\n- **Connection to Trigonometric Coordinate Selection:**\n - Consider a vector (x, y)$situated at an angle$\theta measured from the x-axis:\n - \cos(\theta)$selects the$x-coordinate component from the vector.\n - \sin(\theta)$selects the$y-coordinate component from the vector.\n\n- **Column and Row Selection Demonstrations:**\n - Multiplying identity components extracts specific structural positions:\n - 1 \times 5$copies or selects$5$from the first column, while$0 \times 6 = 0$masks$6.\n - 1 \times 7$copies or selects$7$, while$0 \times 8 = 0$masks$8.\n - 0 \times 5 = 0$masks$5$,$1 \times 6 = 6$copies$6$,$0 \times 7 = 0$masks$7$, and$1 \times 8 = 8$copies$8$.

Higher-Dimensional Generalization and Computer Graphics Applications

• Structural Rule for General $n \times n$ Identity Matrices:

  • An identity matrix of any dimension has ones ($1$s) placed along the main diagonal and zeros ($0$s) filling all off-diagonal entries.

• Three-Dimensional Identity Matrix ($I_3$):

  • $I_3 = \begin{pmatrix} 1 & 0 & 0 \\ 0 & 1 & 0 \\ 0 & 0 & 1 \end{pmatrix}$

• Operational Filtering in Three Dimensions:

  • First row $(1, 0, 0)$: Copies the 1st column while masking the 2nd and 3rd columns.

  • Second row $(0, 1, 0)$: Masks the 1st column, copies the 2nd column, and masks the 3rd column.

  • Third row $(0, 0, 1)$: Masks the 1st and 2nd columns while copying the 3rd column.

• Applications in Virtual Reality and 3D Computer Graphics:

  • Internal Spatial Representation:

  • Three-dimensional models, visual scenes, and virtual reality renderings represent objects and spatial points internally using $3 \times 3$ (or larger) matrices and vectors.

  • Vector and Pixel Manipulation:

  • Graphical pixels and point positions are represented as vectors.

  • Matrix multiplication operations are performed directly on these pixel vectors to execute geometric transformations, positional adjustments, and dynamic manipulations in 3D space.