The Essence of Linear Transformations
A transformation is a rule that “turns each input vector into some output vector” — in essence it is moving every point in space.
input vector v ──transformation T──► output vector T(v)
Functions versus transformations: mathematically it is just a function, but the word “transformation” is used so that you can picture the whole space moving, rather than handling one isolated vector at a time.
Why look at the whole space
Knowing where a single vector goes is of limited use; but if you watch how the entire grid deforms, you can “see” the complete behaviour of the transformation. This is the essence of the visualisation method used in this course.
A transformation is a linear transformation if and only if it satisfies both of the following properties:
Property 1: It Maps Lines to Lines
A straight line is still a straight line after the transformation (it is never bent).
T(a straight line) = a straight line
Property 2: The Origin Stays Fixed, and the Grid Stays Parallel and Equally Spaced
The origin stays put, and the grid lines stay parallel and equally spaced.
Stated Algebraically
Equivalently, a linear transformation satisfies additivity and homogeneity:
Intuition:
- adding first and then transforming = transforming first and then adding
- scaling first and then transforming = transforming first and then scaling
The Core Intuition
A linear transformation is one that “keeps the grid lines parallel and equally spaced”. It may be a rotation, a scaling or a shear, but it cannot be a bending, a non-uniform stretching, or anything that moves the origin.
| Transformation | Linear? | Reason |
|---|---|---|
| A 90° rotation | ✅ | the grid rotates as a whole, the lines stay straight, parallel and equally spaced |
| A uniform scaling by 2 | ✅ | the grid is scaled uniformly |
| A horizontal shear | ✅ | the grid tilts, but the lines stay straight and the spacing is unchanged |
| A reflection in the diagonal | ✅ | mirror symmetry, the structure is preserved |
| A translation (moving the origin) | ❌ | the origin does not stay fixed (violating “the origin stays put”) |
| A bending (curving the grid into arcs) | ❌ | straight lines become curved |
| A non-uniform stretch (denser on one side) | ❌ | the grid spacing is not preserved |
Note: a translation is not a linear transformation! Because it moves the origin away. This is the point beginners most often confuse.
Suggested steps:
- Click “Rotate 90°” — watch the grid rotate as a whole, with the lines staying straight and the spacing unchanged
- Click “Scale (2, 0.5)” — watch the x direction stretch and the y direction compress
- Click “Horizontal shear” — watch the vertical lines tilt into a parallelogram grid
- Change the matrix entries by hand — freely explore any linear transformation
What to observe:
- the grid lines always stay straight, parallel and equally spaced — this is the sign of linearity
- wherever the basis vectors and end up, the whole space ends up accordingly
Because these two properties hold, we do not need to know the image of every vector — knowing where the basis vectors go is enough to deduce the image of every vector (covered in the next section).
any vector v = (x, y) = x·î + y·ĵ
linearity ⟹ T(v) = x·T(î) + y·T(ĵ)
Recording just the images of the two basis vectors completely describes a linear transformation.
Symbols Used in This Article
| Symbol | Type | Pronunciation/Explanation | Meaning in This Article |
|---|---|---|---|
| Mathematical symbol | transformation T | The transformation (function) under consideration, mapping each input vector to an output vector | |
| Mathematical symbol | vectors v and w | The input vectors taking part in the transformation | |
| Mathematical symbol | T of v plus w | Additivity: | |
| Mathematical symbol | T of c v | Homogeneity: | |
| Mathematical symbol | scalar c | The scalar (number) in scalar multiplication | |
| Mathematical symbol | i hat, j hat | The unit basis vectors of the two-dimensional plane |
