Part 2 · Tensors
2.8 Change of basis
Source: beyond the notes (on the syllabus; standard, as in Gurtin, in the notes’ notation)
Objectives
- Understand that changing basis changes components, never the vector or the tensor.
- Recipe: and .
- Picture: turning the basis until the matrix of a symmetric tensor is diagonal.
- Vocabulary: change of basis, invariant.
Everything so far was defined without a basis, then written in . Nothing forces that choice. Take a second orthonormal basis : how do the components change?
The tensor that turns one basis into the other
The change of basis from to is the tensor Column of is written in the old basis.
It is orthogonal: (Proposition 2.2.7). Its columns are orthonormal, as Recipe 2.6.3 requires.
Proofbeyond the notes
. And (Theorem 2.1.6).
Loading the demo…
The vector and the tensor do not change; only their components do. Anything defined without a basis gives the same number from either matrix: , , , the eigenvalues, . For instance .
Choosing as the eigenbasis of a symmetric makes diagonal: the spectral theorem in matrix language.
Do not confuse in components with the tensor . The first is the same tensor, described from a turned basis. The second is a different tensor: physically rotated, as when a stretched body is turned. That one appears with frame indifference.
In the demo, turning by leaves the ellipse where it is. Rotating the body by would turn the ellipse, , and leave alone.
Exercises
In the plane, take , . Find the components of in the new basis.
Solution
. , and times that is . Diagonal: the new basis is the eigenbasis of Exercise 2.7.1. Trace and determinant in both.
Show is the same in every orthonormal basis, using the change-of-basis formula.
Solution
. The same as Exercise 2.6.1.