Introduction

Matrices are organized numbers.

A vector stores one ordered list of numbers. A matrix stores numbers in rows and columns. That simple change lets us organize many vector operations at once.

This chapter keeps the purpose practical:

a matrix is:
a shape-aware table of numbers
that can transform vectors

The phrase shape-aware matters. Matrix arithmetic is not only about values. It is also about whether the rows, columns, and vector lengths fit together.

Why Matrices Come After Vectors

Vectors taught us how to read ordered coordinates. Matrices add a second direction of organization.

A row can interact with a vector through a dot product. Many rows can produce many outputs. Later, many matrices can stack into layers.

That is the bridge:

vector operation
-> many organized vector operations

This is why matrices become central in neural networks. A single dot product can score one pattern in an input vector. A matrix can hold many such patterns as rows, so one matrix-vector product can compute many scores at once.

What This Chapter Teaches

You will learn to:

• read matrices as tables
• name rows, columns, and entries
• reason about shape
• compute matrix-vector products
• compute small matrix products
• recognize identity, diagonal, and transpose
• read batch shapes
• see matrices as linear maps
• connect matrices to neural network layers

We will avoid rank, eigenvalues, SVD, PCA, and deeper decomposition ideas for now. Those belong later, after basic matrix computation is fluent.

Next, we start with the simplest reading: a matrix as a table.