1. Transformations Between Dimensions
So far: 2D → 2D (2×2 matrices) and 3D → 3D (3×3). It’s equally sensible to map between dimensions, e.g. 2D → 3D.
Definition 1.1 – Linear (between dimensions)
Same rule as before: grid lines stay parallel and evenly spaced, and the origin maps to the origin.
⚠️ Pitfall: Inputs and outputs now live in completely separate spaces — a 2D input isn’t “moving” within 3D; it’s a different kind of object.
2. 3×2 Matrices: 2D → 3D
Encoding
As always: write where each basis vector lands as a column.
Example. , :
- Column space: a 2D plane through the origin of 3D space.
- Still full rank: the column space has 2 dimensions = the input dimension.
Worked example.
3. Reading Shape → Dimensions
Rule 3.1
For an matrix:
- columns → basis vectors in the input → input is -dimensional.
- rows → each landing spot needs coordinates → output is -dimensional.
| Shape | Map | Picture |
|---|---|---|
| 2D → 3D | Plane tilted inside 3D space | |
| 3D → 2D | All of space flattened onto a plane | |
| 2D → 1D | Plane collapsed onto the number line | |
| 2D → 2D | Usual case |
⚠️ Pitfall: “Rows × columns” — the columns give the input dimension, the rows give the output dimension. It’s easy to swap them.
4. 2×3 Matrices: 3D → 2D
Three columns (three basis vectors in 3D), each landing on a point with only two coordinates → a squish of 3D space onto the plane.
Example.
5. 1×2 Matrices: 2D → the Number Line
1D space is just the number line, so these take a 2D vector and return a number.
Visual meaning of linearity here
Grid lines are hard to track with so much squishing, so use: a line of evenly spaced dots stays evenly spaced after landing on the number line.
Example. sends , :
These turn out to be closely tied to the dot product (Chapter 9).
6. Products of Nonsquare Matrices
still means “apply , then ”. It only makes sense when ‘s output dimension equals ‘s input dimension:
⚠️ Pitfall: A times a is undefined — the first output (2D) can’t feed an input expecting 3D.
Exercises: Think about matrix multiplication and linear systems for transformations between different dimensions — e.g. compute for the (2×3) and (3×2) above and interpret it as a 2D → 2D map (lesson page).