Created: 2017-02-28 Tue 10:45
A vector is an ordered set of numbers arranged in a column. An n-dimensional column vector \(\mathbf{a}\) is
A matrix is a set of column vectors. An \(n \times k\) matrix \(\mathbf{A}\) is
A diagonal matrix: a square matrix in which all off-diagonal elements equal zero, that is,
An identity matrix: a diagonal matrix in which all diagonal elements are 1. A subscript is sometimes included to indicate its size, e.g. \(\mathbf{I}_4\) indicate a \(4 \times 4\) identity matrix.
A triangular matrix: have only zeros either above or below the main diagonal. A lower triangular matrix looks like
The transpose of a matrix \(\mathbf{A}\), denoted \(\mathbf{A}^{\prime}\), is obtained by creating the matrix whose kth row is the kth column of the original matrix. That is,
\[ \mathbf{B} = \mathbf{A}^{\prime} \Leftrightarrow b_{ik} = a_{ki} \text{ for all } i \text{ and } k \]
For two matrices \(\mathbf{A}\) and \(\mathbf{B}\) with the same dimensions, that is both are \(n \times k\).
\[\mathbf{A} + \mathbf{B} = [a_{ij} + b_{ij}] \text{ for all } i \text{ and } j\]
The inner product of two \(n \times 1\) column vector \(\mathbf{a}\) and \(\mathbf{b}\) is \[ \mathbf{a}^{\prime} \mathbf{b} = \sum^n_{i=1} a_i b_i \]
Since both \(\mathbf{a}\) and \(\mathbf{b}\) are \(n \times 1\) vectors, it must hold that \(\mathbf{a}^{\prime} \mathbf{b} = \mathbf{b}^{\prime} \mathbf{a}\).
Suppose that \(\mathbf{A}\) is an \(n \times m\) matrix and \(\mathbf{B}\) is an \(m \times k\) matrix, then the product \(\mathbf{C} = \mathbf{AB}\) is an \(n \times k\) matrix, where the \((i,j)\) element of \(\mathbf{C}\) is \(c_{ij} = \sum_{l=1}^m a_{il} b_{lj}\).
In other words, if we write \(\mathbf{A}\) and \(\mathbf{B}\) with vectors, that is,
where \(\mathbf{a}_i = [a_{i1}, a_{i2}, \cdots, a_{im}]^{\prime}\) is the ith row of \(\mathbf{A}\) for \(i = 1, 2, \ldots, n\), and \(\mathbf{b}_j = [b_{1j}, b_{2j}, \ldots, b_{mj}]^{\prime}\) is the jth column of \(\mathbf{B}\) for \(j = 1, 2, \ldots, k\).
Let \(a^{ik}\) be the ikth element of \(\mathbf{A}^{-1}\). The general formula for computing an inverse matrix is
\[ a^{ik} = \frac{|\mathbf{C}_{ki}|}{|\mathbf{A}|} \]
where \(| \mathbf{A} |\) is the determinant of \(\mathbf{A}\), \(| \mathbf{C}_{ki} |\) is the kith cofactor of \(\mathbf{A}\), that is, the determinant of the matrix \(\mathbf{A}_{ki}\) obtained from \(\mathbf{A}\) by deleting row \(k\) and column \(i\), pre-multiplied by \((-1)^{(k + i)}\).
We need to use the following results of matrix calculus in the future lectures.
When \(\mathbf{A}\) is symmetric, then \((\partial \mathbf{x}^{\prime} \mathbf{A} \mathbf{x}) / (\partial \mathbf{x}) = 2\mathbf{A} \mathbf{x}\)