Definite matrices
Summary
- \(A\) is a square \(n×n\) matrix and \(x\) is an \(n×1\) vector. We say that \(A\) is
- Positive definite if \(x'Ax>0\) for all \(x≠0\)
- Negative definite if \(x'Ax<0\) for all \(x≠0\)
- Positive semidefinite if \(x'Ax≥0\) for all \(x≠0\)
- Negative semidefinite if \(x'Ax≤0\) for all \(x≠0\)
- Indefinite: \(x'Ax\) may take values of different sign
- Result: If \(A\) is diagonal then \(A\) is
- Positive definite if \(A_{ii}>0\) for \(i=1, \ldots ,n\)
- Negative definite if \(A_{ii}<0\) for \(i=1, \ldots ,n\)
- Positive semidefinite if \(A_{ii}≥0\) for \(i=1, \ldots ,n\)
- Negative semidefinite if \(A_{ii}≤0\) for \(i=1, \ldots ,n\)
- Result: for \(n=2\) and
\[A=\begin{bmatrix}a & c \\ c & b\end{bmatrix} x=\begin{bmatrix}x_1 \\ x_2\end{bmatrix}\]
- we have
\[x'Ax=ax_1^2+bx_2^2+2cx_1x_2\]
- and \(A\) is
- Positive definite if \(\left| A \right|>0\) and \(a>0\)
- Negative definite if \(\left| A \right|>0\) and \(a<0\)
- Positive semidefinite if \(\left| A \right|≥0\) and \(a≥0\)
- Negative semidefinite if \(\left| A \right|≥0\) and \(a≤0\)
- where \(\left| A \right|\) is the Determinant of \(A\) .