The Inverse Matrix and Systems of Linear Equations
If the transformation stretches and rotates the plane, then the inverse transformation turns it back to what it was:
Geometric intuition: is “deforming space”, and is “restoring the deformation”. Deforming and then restoring = no change at all.
v ──M──► Mv ──M⁻¹──► v
The inverse matrix exists ⟺ det ≠ 0
- det ≠ 0: the transformation squashes nothing, and everything can be perfectly restored → invertible
- det = 0: space has been squashed flat, information has been lost, and it cannot be restored → not invertible (singular)
How to remember it:
- Swap the main diagonal:
- Negate the off-diagonal: change sign
- Divide the whole thing by the determinant
Example: M = [[2, 0], [0, 4]] (x stretched by 2, y stretched by 4)
det = 8
M⁻¹ = (1/8)[[4, 0], [0, 2]] = [[0.5, 0], [0, 0.25]]
Check: M·M⁻¹ = [[2×0.5, 0], [0, 4×0.25]] = [[1, 0], [0, 1]] = I ✓
Geometrically: the x direction is stretched by 2, so the inverse transformation compresses it back by half (×0.5); the same holds for the y direction.
When det = 0
The denominator is zero and the formula fails. This is exactly the algebraic expression of “not invertible” — as we said earlier, space has been squashed flat at this point and there is no way to restore it.
A system of linear equations can be written in matrix form:
that is, .
Geometric view: the problem becomes — “which vector lands on after the transformation ?”
Solving: multiply both sides on the left by :
Example: 2x + 0y = 2 → [[2,0],[0,4]]·(x,y) = (2, 8)
0x + 4y = 8
M⁻¹ = [[0.5,0],[0,0.25]]
(x, y) = M⁻¹·(2, 8) = (0.5×2, 0.25×8) = (1, 2) ✓
Geometric meaning: solving an equation = “running the transformation backwards”. This is exactly the view that 3Blue1Brown emphasises: a system of linear equations is not a game of symbols, but “finding the vector that reaches the target after the transformation”.
| Case | Geometry | The solutions |
|---|---|---|
| det ≠ 0 | The transformation is full-dimensional, so always has a unique preimage | a unique solution |
| det = 0 and is in the column space | lands on the squashed line, and several vectors map onto it | infinitely many solutions |
| det = 0 and is not in the column space | is off the squashed line | no solution |
” is in the column space” — this concept is exactly the subject of the next section.
| Property | Formula |
|---|---|
| The inverse of a product | (note the reversal of the order!) |
| The inverse of a transpose | |
| The inverse of an inverse | |
| The determinant |
Geometric understanding of : to undo “first and then ”, you have to undo first (with ) and then undo (with ) — so the order of the inverses is reversed.
Core Conclusions
- The inverse matrix = the reverse transformation, satisfying
- The inverse exists ⟺ det ≠ 0 ⟺ the transformation squashes nothing
- The 2×2 formula: swap the main diagonal, negate the off-diagonal, divide by the determinant
- Solving ⟺ running the transformation backwards to find the preimage:
Symbols Used in This Article
| Symbol | Type | Pronunciation/Explanation | Meaning in This Article |
|---|---|---|---|
| Mathematical symbol | M inverse | The inverse matrix: the reverse transformation that turns back to the original | |
| Mathematical symbol | identity matrix | The identity matrix (the identity transformation), satisfying | |
| Mathematical symbol | 2 by 2 matrix | The general form of a 2×2 matrix | |
| Mathematical symbol | a d minus b c | The 2×2 determinant, the denominator in the inverse matrix formula | |
| Mathematical symbol | M x equals b | The matrix form of a system of linear equations | |
| Mathematical symbol | x equals M inverse b | The solution of the system obtained with the inverse matrix | |
| Mathematical symbol | inverse of M1 M2 | The inverse of a product, equal to (the order is reversed) |
