Change of Basis

I. One Vector, Different Coordinates

In the standard basis {ı^,ȷ^}\{\hat{\imath}, \hat{\jmath}\}, the vector (3,2)(3, 2) means "3ı^+2ȷ^3\hat{\imath} + 2\hat{\jmath}".

But if you change to another basis {b1,b2}\{\mathbf{b}_1, \mathbf{b}_2\} (say b1=(1,1)\mathbf{b}_1 = (1, 1) and b2=(1,−1)\mathbf{b}_2 = (1, -1)), the same vector has different coordinates — expressed as a combination of b1\mathbf{b}_1 and b2\mathbf{b}_2.

The same point:
in the standard basis: v = (3, 2)
in the new basis:    v = 2.5·b₁ + 0.5·b₂   (since 2.5(1,1) + 0.5(1,-1) = (3, 2))

Coordinates are not an intrinsic property of a vector, but a statement “relative to some basis”.

II. The Change-of-Basis Matrix

From coordinates in the new basis to coordinates in the standard basis

Let the new basis be {b1,b2}\{\mathbf{b}_1, \mathbf{b}_2\} (written in coordinates relative to the standard basis). If the coordinates of a vector in the new basis are [v]B=(x′,y′)T[\mathbf{v}]_\mathcal{B} = (x', y')^T, then its coordinates in the standard basis are:

v=x′b1+y′b2=(∣∣b1b2∣∣)⏟B(x′y′)\mathbf{v} = x'\mathbf{b}_1 + y'\mathbf{b}_2 = \underbrace{\begin{pmatrix} | & | \\ \mathbf{b}_1 & \mathbf{b}_2 \\ | & | \end{pmatrix}}_{B} \begin{pmatrix} x' \\ y' \end{pmatrix}

The columns of the change-of-basis matrix BB are the new basis vectors (in coordinates relative to the standard basis).

From coordinates in the standard basis to coordinates in the new basis

Conversely, use B−1B^{-1}:

[v]B=B−1v[\mathbf{v}]_\mathcal{B} = B^{-1}\mathbf{v}
Example: B = [[1, 1], [1, -1]] (b₁ = (1,1), b₂ = (1,-1))
B⁻¹ = (1/-2)[[-1, -1], [-1, 1]] = [[0.5, 0.5], [0.5, -0.5]]
Check: B⁻¹·(3, 2) = (0.5×3+0.5×2, 0.5×3−0.5×2) = (2.5, 0.5) ✓
(consistent with 2.5b₁ + 0.5b₂ above)

III. A Linear Transformation under a Change of Coordinates: Similar Matrices

The problem

The same linear transformation TT is represented by the matrix AA in the standard basis and by the matrix A′A' in the new basis B\mathcal{B}. What is the relation between A′A' and AA?

The derivation

  1. First convert the vector from the new-basis coordinates back to the standard basis: BB
  2. Apply the transformation: AA
  3. Convert back to the new-basis coordinates: B−1B^{-1}
A′=B−1ABA' = B^{-1} A B

A′A' and AA are said to be similar.

IV. Why Change of Basis Matters

ScenarioSignificance
Choosing a “good” basisunder an eigenbasis some transformations become the simplest possible diagonal matrix
Diagonalisationif you find a set of eigenvectors to serve as a basis, the matrix becomes diagonal (the next section)
Signal processingthe Fourier transform = changing to the frequency basis
Data sciencePCA (principal component analysis) = changing to the directions of greatest variance (the eigenvectors)

The benefit of diagonalisation: computation with a diagonal matrix is extremely simple (element by element), and it also makes taking powers easy:

Ak=B(λ1k00λ2k)B−1A^k = B \begin{pmatrix} \lambda_1^k & 0 \\ 0 & \lambda_2^k \end{pmatrix} B^{-1}

V. An Intuitive Summary

ConceptExplanation
Coordinatesa statement relative to some basis, not an intrinsic property
The change-of-basis matrix BBits columns are the new basis vectors in standard-basis coordinates
Converting back to the standard basisv=B[v]B\mathbf{v} = B[\mathbf{v}]_\mathcal{B}
Converting to the new basis[v]B=B−1v[\mathbf{v}]_\mathcal{B} = B^{-1}\mathbf{v}
Similar matricesA′=B−1ABA' = B^{-1}AB (the same transformation, different coordinate systems)
The purposeto choose a “good basis” that makes the matrix simpler (e.g. diagonalisation)

Summary

Symbols Used in This Article

SymbolTypePronunciation/ExplanationMeaning in This Article
B\mathcal{B}Mathematical symbolcalligraphic BThe chosen basis (e.g. the new basis {b1,b2}\{\mathbf{b}_1,\mathbf{b}_2\})
b1,b2\mathbf{b}_1, \mathbf{b}_2Mathematical symbolb one, b twoThe new basis vectors, written in standard-basis coordinates
[v]B[\mathbf{v}]_\mathcal{B}Mathematical symbolv in basis BThe coordinates of the vector v\mathbf{v} in the basis B\mathcal{B}, such as (x′,y′)T(x',y')^T
BBMathematical symbolmatrix BThe change-of-basis matrix, whose columns are the new basis vectors
B−1B^{-1}Mathematical symbolB inverseThe inverse of the change-of-basis matrix, converting standard-basis coordinates back to the new basis
A′=B−1ABA' = B^{-1}ABMathematical symbolA prime equals B inverse A BThe matrix of the same linear transformation in the new basis; A′A' and AA are similar
λ1,λ2\lambda_1, \lambda_2Greek letterLambdaThe eigenvalues appearing in the diagonal standard form after diagonalisation

Found an issue with this content?

Tell us what could be improved.