Change of Basis and Abstraction
Until now a “vector” has always been a finite-dimensional array of coordinates. But the truly powerful thing about linear algebra is that it generalises: any set satisfying the rules of addition and scalar multiplication is a vector space — the set of functions, the set of matrices, the set of polynomials; all of them can have bases, dimensions and coordinates.
This chapter proceeds in two steps. First, change of basis: within the same vector space, changing to a different basis gives different coordinates, but the vector itself has not changed; from this comes the notion of similar matrices — two matrices are similar when they describe the same linear transformation in different bases. This also explains why the eigenvalues are invariants of similarity: changing the basis does not change which directions the transformation “stretches” or by how much it “scales”.
Then it turns to abstract vector spaces: dropping the concrete requirement of “components” and keeping only the axioms of addition and scalar multiplication. Once you do that, functions, polynomials and matrices all become “vectors”, and their respective “bases” are seen from a unified perspective. Abstraction is not affectation — it is the reason why the same theory can handle every one of the concrete cases above.
