Finding Linear Maps in Linear Algebra- Complete Tutorial
What Is a Linear Map, Anyway?
A linear map (also called a linear transformation) is a function between vector spaces that preserves two operations: vector addition and scalar multiplication. That's it. No curves, no weird exceptions—just clean, predictable behavior.
In plain terms: if T is a linear map from vector space V to vector space W, then for any vectors u and v in V and any scalar c:
- T(u + v) = T(u) + T(v)
- T(cu) = cT(u)
If a transformation doesn't satisfy both conditions, it's not linear. Period.
Why Finding Linear Maps Matters
Linear maps show up everywhere in mathematics, physics, computer science, and engineering. They describe rotations, reflections, projections, and scaling. If you're working with vector spaces, you're working with linear maps whether you realize it or not.
The real question is: how do you identify one when you see it?
The Two-Part Test for Linearity
Before you can find or verify a linear map, you need to check these two properties:
1. Additivity
Take any two vectors x and y in the domain. Compute T(x + y) and compare it to T(x) + T(y). If they're equal, additivity holds. If not, you're done—it's not linear.
2. Homogeneity (Scalar Multiplication)
Take any vector x and any scalar c. Compute T(cx) and compare it to cT(x). Equal? Good. Not equal? Also a definitive answer.
Both conditions must hold for every possible input. One failure means the whole thing fails.
How to Find a Linear Map: The Matrix Approach
Here's the practical part. If you know where a basis maps to, you can construct the entire linear map.
Step 1: Identify the Domain and Codomain Dimensions
Say you have a transformation T: ℝⁿ → ℝᵐ. You need an n×m matrix to represent it.
Step 2: Apply T to Basis Vectors
Take the standard basis vectors of the domain. Apply T to each one. These outputs become your matrix columns.
For T: ℝ² → ℝ³, if:
- T(1,0) = (2, 1, 0)
- T(0,1) = (1, -1, 3)
Then your matrix is:
[2 1]
[1 -1]
[0 3]
Step 3: Write the General Formula
For any vector (x, y), the transformation is:
T(x, y) = x·T(1,0) + y·T(0,1)
Plug in the values and simplify. You'll get a formula you can use on any input.
Common Examples That Actually Work
Zero Transformation
T(v) = 0 for all v. Linear? Yes. Always. The zero map sends everything to the zero vector. It's boring but valid.
Identity Transformation
T(v) = v. Also always linear. The identity map leaves vectors unchanged.
Projection Onto the x-axis
T(x, y) = (x, 0). This projects any point onto the x-plane. Linear—satisfies both conditions.
Rotation by θ
T(x, y) = (x·cosθ - y·sinθ, x·sinθ + y·cosθ). Rotations are linear. The matrix is the standard rotation matrix.
Examples That Are NOT Linear
Students often make these mistakes. Don't be that person.
Translation
T(x, y) = (x + 1, y + 2). This is not linear. There's a constant term. Linear maps must pass through the origin—T(0) must equal 0.
Quadratic Transformations
T(x, y) = (x², y). Squares, cubes, or any nonlinear function of inputs break linearity. The squaring operation doesn't preserve the structure.
Absolute Value
T(x) = |x|. Fails homogeneity for negative scalars. T(-1·x) ≠ -1·T(x) in general.
Methods for Verifying Linearity
Here's a comparison of the main approaches:
| Method | Best For | Speed | Drawback |
|---|---|---|---|
| Direct definition check | Formal proofs | Slow | Must test all vectors |
| Matrix representation | Concrete calculations | Fast | Only works for finite dimensions |
| Kernel/Image analysis | Understanding structure | Medium | Doesn't verify linearity itself |
| Geometric intuition | Simple transformations | Fastest | Limited to ℝ² and ℝ³ |
Quick Checklist: Is It Linear?
Run through these before you commit to an answer:
- Does T(0) = 0? If not, stop here—it's not linear.
- Does T(u + v) = T(u) + T(v) for your test vectors?
- Does T(cu) = cT(u) for c = 2, -1, and 0?
- Is the transformation defined by matrix multiplication?
If all answers are yes, you have yourself a linear map.
Getting Started: Practice Problem
Determine if T: ℝ² → ℝ² defined by T(x, y) = (2x + y, x - 3y) is linear.
Solution:
Check T(0, 0) = (0, 0). ✓
Pick two vectors: u = (1, 2), v = (3, 4).
T(u) = (2(1) + 2, 1 - 3(2)) = (4, -5)
T(v) = (2(3) + 4, 3 - 3(4)) = (10, -9)
T(u) + T(v) = (14, -14)
T(u + v) = T(4, 6) = (2(4) + 6, 4 - 3(6)) = (14, -14)
Matches. ✓
Now check homogeneity with c = 3:
T(3u) = T(3, 6) = (2(3) + 6, 3 - 3(6)) = (12, -15)
3T(u) = 3(4, -5) = (12, -15)
Matches. ✓
Conclusion: T is linear. The matrix is [2 1; 1 -3].
What Linear Maps Cannot Do
You need to know the limits. Linear maps cannot:
- Translate points (that requires an affine transformation)
- Change the origin's location
- Curve straight lines into curves
- Increase dimension arbitrarily (rank ≤ min(dim domain, dim codomain))
If you need these things, you're outside linear algebra territory.
Finding Linear Maps Between Abstract Vector Spaces
When the vector spaces aren't ℝⁿ, the approach stays the same but looks more abstract.
Given T: V → W, and you know a basis for V and what T does to each basis vector, you can determine T completely. Any vector in V is a unique linear combination of basis vectors, so T acts predictably on everything.
Example: If V has basis {1, x, x²} (polynomials of degree ≤ 2) and T: V → V is the derivative operator, then:
- T(1) = 0
- T(x) = 1
- T(x²) = 2x
So T(ax² + bx + c) = 2ax + b. That's a linear map between polynomial spaces.
The Bottom Line
Finding linear maps comes down to two questions: does it preserve addition, and does it preserve scalar multiplication? Once you know what a transformation does to a basis, you know everything about it.
Stop overcomplicating this. Check the definition, verify the properties, construct the matrix if you need to compute things. That's the entire process.