1. Matrices#
\(\mathbf{A} \in \mathbb{K}^{m,n} \) with usually \(\mathbb{K}^{m,n} = \mathbb{R}^{m,n}\) or \(\mathbb{C}^{m,n}\)
Hermitian matrix. The Hermitian matrix \(\mathbf{A}^*\) of a matrix \(\mathbf{A}\) is the transpose and complex conjugate matrix (if \(\mathbb{K} = \mathbb{C}\)),
with the notation of \(a^*\) for the complex conjugate of a numerical quantity.
1.1. Subspaces#
1.1.1. Range, Image#
The range of a matrix \(\mathbf{A}\) is the linear space built on the columns of \(\mathbf{A}\), since the operation \(\mathbf{A} \mathbf{x}\) represents nothing but a linear combination of the columns of matrix.
1.1.2. Null, Kernel#
1.2. Theorem#
1.2.1. Orthogonality of \(R(\mathbf{A})\) and \(K(\mathbf{A^*})\)#
The following holds,
meaning that \(\forall \mathbf{u} \in R(\mathbf{A})\) and \(\forall \mathbf{v} \in K(\mathbf{A}^*)\), \(\mathbf{u}^* \mathbf{v} = 0\).
Proof.
and thus, premultiplication by \(\mathbf{x}^*\) of the second relation gives
This theorem becomes quite useful, e.g. for constrained linear systems and projections… (e.g. N-S, or other constrained linear systems…)
todo add links