Computing Transition Matrices
You'll learn a systematic technique to find the matrix that converts coordinates from one basis to another.
The Core Idea
You already know what a transition matrix does—it transforms coordinate vectors from one basis to another. But how do you actually compute it? The most reliable method uses an augmented matrix approach that leverages row reduction.
Here's the key insight: If you want the transition matrix from basis B to basis C (call it P), you need to express each vector in B using C-coordinates. Instead of solving separate systems for each basis vector, you solve them all at once using an elegant trick.
The Augmented Matrix Method
To find the transition matrix P from basis B = {b₁, b₂, ..., bₙ} to basis C = {c₁, c₂, ..., cₙ}:
-
Form an augmented matrix: Place the C-basis vectors as columns on the left, and the B-basis vectors as columns on the right:
[C | B] -
Row reduce to reduced row echelon form:
[I | P] -
The result: The right side becomes your transition matrix P!
Why This Works
When you row reduce [C | B] to [I | P], you're essentially solving C · P = B. Since C is a basis (invertible), this gives P = C⁻¹B. The matrix P contains the C-coordinates of each B-basis vector as its columns—exactly what a transition matrix needs.
Reversing Direction
Need the transition matrix from C to B instead? Simply compute the inverse: P⁻¹. Or, start with [B | C] and row reduce.
Key Takeaway: Compute transition matrices by forming the augmented matrix [new_basis | old_basis] and row reducing to [I | transition_matrix]—a single elegant operation that solves multiple coordinate conversions simultaneously.