Lecture 4: Linear Operators on Vector Spaces Flashcards
Linear Operator Matrix y = L(x), B = Aa
- Linear operator satisfies superposition
- Relation between input and output representations
- Used to find other relations
Properties of Representation Matrix A
ith Column of A is the representation of yi = L(ei) with respect to output basis
Progression of Linear Operator Matrix Equations
- input as matrix form
- output as matrix form
- output columns
- x = [e1 … en] * [a1;…;an]
- y = L(x) = a1*L(e1) + … + anL(en) = UAa
- yi = L(ei) = [e1 … en] * ai
Relationship between two representations and basis of x
- a’ = Pa
2. ith column of P is the representation of original basis vector ei with new basis vectors e’
Relationship between two representations and basis of y
- B’ = TB
2. ith column of T is the representation of original basis vector ui with new basis vectors u’
Relationship Between Matrix Representations
A = inverse(T) * A' * P A' = T * A * inverse(P)
Similarity Transformation
A and A’ are similar if there exists a nonsingular matrix P satisfying
A = inverse(P) * A’ * P
A’ = P * A * inverse(P)
Operators as Spaces
- Set of all linear operators is a linear space
- Follows axioms of Norms
- Norm used to find size of operator
Matrix Norms
- Supremum
- A1 Norm
- Euclidean Norm
- Infinity Norm
- Frobenius Norm
- Least upper bound
- Largest sum across columns
- sqrt(Max(x’T *A’T Ax))
- Largest sum across rows
- sqrt(tr(A’T *A)
Bounded Linear Operator
||Ax|| <= k ||x|| for all vectors x in space.
If gain ||Ax|| / ||x|| is bounded
Adjoint Operator L*
Must satisfy = for all vectors x and y (<> denotes inner product)
ie. A represents operator L where an orthonormal basis to obtain matrix representation A, then Adjoint of L is transpose(A’).
Self-Adjoint Operators
If adjoint operator is the same as the operator to begin with it is called the hermitian operator. Operator must have orthonormal basis.
If hermitian is real then A = transpose(A).