Abstract¶ The definition of the adjoint operator.
Keywords: Vectors Vector Spaces Inner Products Adjoint Transpose Complex Conjugate. ¶ Adjoint Definition ¶ Let V V V be an inner product space defined over C \mathbb{C} C , with inner product ( ⋅ , ⋅ ) : C n ⊗ C n → C 1 (\cdot, \cdot): \mathbb{C}^{n} \otimes \mathbb{C}^{n} \rightarrow \mathbb{C}^{1} ( ⋅ , ⋅ ) : C n ⊗ C n → C 1 , and L : V → V L: V \rightarrow V L : V → V a linear operator.
The adjoint operator of the linear operator L L L , denoted by L † L^{\dagger} L † , is the operator that satisfies the following relation
( L † ( x ) , y ) = ( x , L ( y ) ) , (L^{\dagger}(\bm{x}), \bm{y}) = (\bm{x}, L(\bm{y})), ( L † ( x ) , y ) = ( x , L ( y )) , where x , y ∈ V \bm{x}, \bm{y}~\in~V x , y ∈ V .
Associated Matrix of the Adjoint ¶ Let M L B M^{\mathfrak{B}}_{L} M L B be the matrix associated to the linear map L L L with respect to the orthonormal basis B \mathfrak{B} B and the element in the i i i th row and j j j th colunm be a i j a_{ij} a ij .
The matrix associated to the linear map L † L^{\dagger} L † with respect to the orthonormal basis B \mathfrak{B} B , M L † B M_{L^{\dagger}}^{\mathfrak{B}} M L † B , is given by taking the complex conjugate of all elements of M L B M_{L}^{\mathfrak{B}} M L B and then transposing. Hence, the element in the i i i th row and j j j th column of M L † B M_{L^{\dagger}}^{\mathfrak{B}} M L † B will be a j i ∗ a^{*}_{ji} a ji ∗ .
For this reason, the adjoint is also sometimes referred to as the conjugate transpose operations.
Properties of the Adjoint ¶ Let V , ( ⋅ , ⋅ ) V, (\cdot, \cdot) V , ( ⋅ , ⋅ ) be an inner product space defined over C \mathbb{C} C , let L , T : V → V L, T: V \rightarrow V L , T : V → V be linear operators and λ ∈ C 1 \lambda~\in~\mathbb{C}^{1} λ ∈ C 1 .
( L + T ) † = L † + T † (L+T)^{\dagger} = L^{\dagger} + T^{\dagger} ( L + T ) † = L † + T † .( λ L ) † = λ ∗ L † (\lambda L)^{\dagger} = \lambda^{*} L^{\dagger} ( λ L ) † = λ ∗ L † , where ( ⋅ ) ∗ (\cdot)^{*} ( ⋅ ) ∗ is the complex conjugate.( L T ) † = T † L † . (LT)^{\dagger} = T^{\dagger}L^{\dagger}. ( L T ) † = T † L † . ( L † ) † = L (L^{\dagger})^{\dagger} = L ( L † ) † = L .If L − 1 L^{-1} L − 1 exists, then ( L † ) − 1 = ( L − 1 ) † (L^{\dagger})^{-1} = (L^{-1})^{\dagger} ( L † ) − 1 = ( L − 1 ) † . Matrix Properties From the Adjoint ¶
Let A ∈ M n m A~\in~\mathbb{M}_{nm} A ∈ M nm .
A A A is unitary if A † A = A A † = I A^{\dagger}A=AA^{\dagger}=\mathbb{I} A † A = A A † = I ,
where I \mathbb{I} I is the identity.
Properties
Let A A A be unitary, then:
A − 1 A^{-1} A − 1 is unitary.∣ ∣ U x ∣ ∣ = ∣ ∣ x ∣ ∣ \vert \vert U \bm{x} \vert \vert = \vert \vert \bm{x} \vert \vert ∣∣ U x ∣∣ = ∣∣ x ∣∣ , where x ∈ V \bm{x}~\in~V x ∈ V and ∣ ∣ ⋅ ∣ ∣ \vert \vert \cdot \vert \vert ∣∣ ⋅ ∣∣ is a norm .If λ \lambda λ is an eigenvalue of A A A , then ∣ λ ∣ = 1 \vert \lambda \vert = 1 ∣ λ ∣ = 1 , meaning A A A has complex eigenvalues of magnitude one. The columns of A A A form an orthonormal basis . Let A ∈ M n m A~\in~\mathbb{M}_{nm} A ∈ M nm .
A A A is hermitian, or self-adjoint, if A = A † A=A^{\dagger} A = A † .
Properties
Let A A A be hermitian, then:
A A A has only real eigenvalues .A A A ’s eigenvectors with different eigenvalues are orthogonal.A A A ’s eigenvectors form a complete orthonormal basis.Let A ∈ M n m A~\in~\mathbb{M}_{nm} A ∈ M nm .
A A A is normal if A † A = A A † A^{\dagger}A = AA^{\dagger} A † A = A A †
Properties
Let A A A be normal, then:
If A x = λ x A\bm{x} = \lambda \bm{x} A x = λ x such that λ \lambda λ is an eigenvalue of A A A , then A † x = λ ∗ x , A^{\dagger} \bm{x} = \lambda^{*} \bm{x}, A † x = λ ∗ x , where λ ∗ \lambda^{*} λ ∗ is the complex conjugate of λ \lambda λ and x ∈ V \bm{x}~\in~V x ∈ V .
A A A ’s eigenvectors with different eigenvalues are orthogonal.The space in which A A A is defined over will have an orthonormal basis of eigenvectors of A A A . There exists a unitary matrix, U U U , that can diagonalise A A A , U † A U = ( λ 1 0 … 0 0 λ 2 … 0 ⋮ ⋮ ⋮ ⋮ 0 0 … λ n ) , U^{\dagger}AU = \begin{pmatrix}
\lambda_{1} & 0 & \ldots & 0 \\
0 & \lambda_{2} & \ldots & 0 \\
\vdots & \vdots & \vdots & \vdots \\
0 & 0 & \ldots & \lambda_n \\
\end{pmatrix}, U † A U = ⎝ ⎛ λ 1 0 ⋮ 0 0 λ 2 ⋮ 0 … … ⋮ … 0 0 ⋮ λ n ⎠ ⎞ , where Spec A = ( λ 1 , … , λ n ) \textrm{Spec}A = (\lambda_1, \ldots, \lambda_n) Spec A = ( λ 1 , … , λ n ) . The matrix U U U will have the eigenvectors of A A A as its columns.