This preview shows pages 1–5. Sign up to view the full content.
This preview has intentionally blurred sections. Sign up to view the full version.
View Full DocumentThis preview has intentionally blurred sections. Sign up to view the full version.
View Full Document
Unformatted text preview: EECS 221 A Linear Operators A Notation B Linear Operators C Change of Basis, Range Spaces, and Null Spaces D Eigenvectors and Eigenvalues E Matrix polynomials and functions F Hermitian and Definite Matrices G The SingularValue Decomposition H Operator Theory 1 A. Notation R ( A ) range space of the operator A N ( A ) null space of the operator A A the transpose of the matrix A A * the adjoint of the operator A , or the complexconjugatetranspose of the matrix A A > a positivedefinite matrix A ≥ a positivesemidefinite matrix λ i ( A ) i th eigenvalue of A Spec( A ) the set of eigenvalues of A ρ ( A ) spectral radius of A = max i  λ i ( A )  σ i ( A ) i th singular value of A (in descending order) σ largest singular value σ smallest nonzero singular value 2 B. Linear Operators 1 Let V and W be vector spaces over the same base field F . Definition A linear operator is a mapping M : V→ W such that for all v 1 ,v 2 ∈ V and all α ∈ F (a) M ( v 1 + v 2 ) = M ( v 1 ) + M ( v 2 ) (additivity) (b) M ( αv 1 ) = α M ( v 1 ) (homogeneity) 2 Examples The following operators are linear: ƒ M : C n → C m : v → Av where A ∈ C m × n ƒ M : C (∞ , ∞ ) → R : f ( t ) → f (0). Is the operator M above familiar ? ƒ Suppose h ( t ) ∈ L 2 [ a,b ] and consider M : L 2 [ a,b ] → R : f ( t ) → Z b a h ( t ) f ( t ) dt Why do we insist that f ∈ L 2 [ a,b ] above ? ƒ Let A,B,X ∈ R n × n and consider M : R n × n → R n × n : X → AX + XB 3 Linear operators on finite dimensional vector spaces are matrices with the action of matrixvector multiplication. Theorem Consider a linear operator M : V → W where dim ( V ) = n,dim ( W ) = m . Let B = { b 1 , ··· ,b n } and C = { c 1 , ··· ,c n } be basis for V and W respectively. Suppose M ( b j ) = m X i =1 α i,j c i . Let v = ∑ j γ j b j . Then, M ( v ) = m X i =1 τ i c i where τ = τ 1 . . . τ m = α 1 , 1 ··· α 1 ,n . . . ··· . . . α m. 1 ··· α m,n β 1 . . . β n = Mβ The above result says that if we make the canonical (bijective) association of v with β and of M ( v ) with τ , then the action of the operator M corresponds to matrixvector multiplication as τ = Mβ . The matrix M clearly depends on the particular choice of basis made and is called the matrix representation of M with respect to these basis. 3 Let us explain this result another way. Since V is of dimension n , it is isomorphic to C n . In other words (see Notes on Vector Spaces, B11 ) , there exists a bijection φ : V → C n such that for all v 1 ,v 2 ∈ V and α ∈ C φ ( v 1 + v 2 ) = φ ( v 1 ) + φ ( v 2 ) , and φ ( αv 1 ) = αφ ( v 1 ) This isomorphism can be exhibited explicity as follows. Fix a basis B = { b 1 , ··· ,b n } for V . Any vector v ∈ V can be expressed as a (unique) linear combination v = n X i =1 α i b i ....
View
Full
Document
This note was uploaded on 07/25/2008 for the course ME 859 taught by Professor Choi during the Spring '08 term at Michigan State University.
 Spring '08
 CHOI

Click to edit the document details