Biostatistics · Intermediate

Change of Basis

Written by SignalsLab

Early edition: these lessons are being reviewed.

Content

Summary

Change of basis establishes a unique coordinate representation for vectors relative to an ordered basis, where the order of basis vectors strictly determines the order of coordinates. Transitioning between coordinate systems involves matrix multiplication, specifically using the inverse of the basis matrix to move from standard to non-standard coordinates. This framework allows for the diagonalization of linear transformations, simplifying complex operations like computing matrix powers in dynamical systems.

Intuition

Think of a vector as a physical arrow in space, which exists independently of any coordinate system. However, to communicate its location, we must assign it numbers (coordinates) based on a chosen set of reference directions (a basis). The critical insight is that these numbers are not properties of the arrow itself, but of the relationship between the arrow and the basis. When we change the basis, we are essentially changing the ruler or the grid we use to measure the arrow. Mathematically, this change is mediated by a transition matrix. If we know how the new basis vectors are constructed from the old ones, we can build a matrix that translates coordinates from one system to the other. This is particularly powerful in applications like population dynamics, where switching to an 'eigenbasis' (a special set of directions defined by the system's natural modes) turns a complex, coupled system of equations into simple, independent scaling operations.

Key equations

[v]E=(c1,…,cn)T where v=∑civi
(1)

Definition of coordinate vector relative to ordered basis E.

x=Uc
(2)

Conversion from non-standard coordinates c to standard vector x.

c=U−1x
(3)

Conversion from standard vector x to non-standard coordinates c.

S=U−1V
(4)

Transition matrix from basis V to basis U.

d=Sc
(5)

Conversion of coordinates from basis V (c) to basis U (d).

Derivation

  1. 1. Define an ordered basis E={v1,…,vn} for a vector space V.
  2. 2. Express any vector v∈V uniquely as a linear combination v=c1v1+…+cnvn.
  3. 3. Identify the scalars ci as the coordinate vector [v]E.
  4. 4. Construct matrix U with basis vectors as columns.
  5. 5. Derive x=Uc by recognizing that matrix-vector multiplication sums scaled columns.
  6. 6. Invert the relationship to find c=U−1x for standard-to-non-standard conversion.
  7. 7. For two bases V and U, equate Vc=Ud (both equal the standard vector x).
  8. 8. Solve for d to get d=U−1Vc, defining the transition matrix S=U−1V.

Worked example

Problem

Find the coordinates of vector x=(7,4)T relative to the basis u1=(3,2)T and u2=(1,1)T.

  1. Construct the basis matrix U using the basis vectors as columns: U=(3121).
  2. Calculate the inverse of U: U−1=(1−1−23).
  3. Multiply the inverse matrix by the standard vector x: c=U−1x=(1−1−23)(74).
  4. Compute the result: c=(3−2).

Answer The coordinate vector is c=(3,−2)T, meaning x=3u1−2u2.

Common mistakes

  • Neglecting the ordered nature of bases; swapping basis vectors without swapping coordinate positions leads to incorrect linear combinations.
  • Confusing the direction of the transition matrix; U converts coordinates to standard, while U−1 converts standard to coordinates.
  • Assuming the transition matrix is always the identity; it is only the identity if the bases are identical and ordered the same way.

Animation

Animation coming soon

The animated version of this lesson is being prepared for the web. In the Signals AI app, lessons are drawn step by step.

Code

This is a text lesson: it has no animation code yet.

ccBYNC© 2026 SignalsLab. Lesson text shared as is under CC BY-NC 4.0.