Andrew Thangaraj
Aug-Nov 2020
A linear map \(T:V\to W\) is said to be invertible if there exists a linear map \(S:W\to V\) (called inverse of \(T\)) such that \(ST:V\to V\) is the identity map on \(V\) and \(TS:W\to W\) is the identity map on \(W\).
Properties
An invertible linear map has a unique inverse.
Suppose \(S_1,S_2\) are two inverses. \[S_1=S_1I=S_1(TS_2)=(S_1T)S_2=IS_2=S_2\]
\(T^{-1}\): denotes unique inverse when it exists
\(T\) is invertible iff it is injective and surjective.
This is a classic general result about maps.
See book for proof.
Maps that are not injective are non-invertible
Maps that are not surjective are non-invertible
Polynomials: multiplication by \(x^2\)
Polynomials: left shift
An invertible linear map is called an isomorphism. Two vector spaces are called isomorphic if there is an isomorphism between them.
Two finite-dimensional vector spaces over the same field \(F\) are isomorphic iff they have the same dimension.
A finite-dimensional vector space \(V\) is isomorphic to \(F^n\), where \(n=\) dim \(V\).
dim \(V=n\), dim \(W=m\), \(\mathcal{L}(V,W)\): vector space of linear maps, \(F^{m,n}\): vector space of \(m\times n\) matrices
\(\mathcal{L}(V,W)\) and \(F^{m,n}\) are isomorphic.
Fix bases for \(V\) and \(W\)
\(\mathcal{M}\): linear, injective, surjective
dim \(\mathcal{L}(V,W)=\text{dim}(V)\text{dim}(W)\)
A linear map from a vector space to itself is called an operator.
\(\mathcal{L}(V)\): set of all operators on \(V\).
Operators are the most important linear maps.
Invertible operators or invertible square matrices are an important class.
When are operators injective, surjective and invertible?
Let \(T:V\to V\) be an operator and let \(V\) be finite-dimensional. Then the following are equivalent: (a) \(T\) is invertible; (b) \(T\) is injective; (c) \(T\) is surjective.
Proof
a implies b: by definition
b implies c: Fundamental theorem
c implies a: Fundamental theorem
How to find if a matrix is invertible?
Matrix has to be square
\(m>n\): not surjective
\(m<n\): not injective
If square, the column rank has to be full
Find dimension of column space
Gaussian elimination