Eigenvalues and eigenvectors: Eigenvalues and eigenvectors
Solving an eigenvalue problem
When you determine eigenvalues of a matrix and corresponding eigenspaces, then you solve an eigenvalue problem for a matrix. Below is an example.
Solve the eigenvalue problem for the matrix \[
A = \matrix{4 & 0 \\ 6 & 2}
\]In other words, determine the eigenvalues and vectors.
A = \matrix{4 & 0 \\ 6 & 2}
\]In other words, determine the eigenvalues and vectors.
The characteristic equation is \[\det\matrix{4-\lambda & 0 \\ 6 & 2-\lambda }=0\] We first rewrite the characteristic polynomial of \(A\):
\[ \begin{aligned}
\det(A-\lambda I) = \left\vert \begin{array}{cc} 4-\lambda & 0 \\ 6 & 2-\lambda \end{array} \right\vert &= (4-\lambda)(2-\lambda)-0\cdot6 \\
&= (4-\lambda)(2-\lambda)
\end{aligned}\] So, the eigenvalues are \(\lambda_1 = 2\) and \(\lambda_2 = 4\).
Let \(\lambda = 2\) be an eigenvalue of the matrix \[A=\matrix{4 & 0 \\ 6 & 2}\] Then there must be a vector \(\vec{v}\) such that \(A\vec{v}=2\vec{v}\), this is, \[(A-2 I)\vec{v}=\vec{0}\text.\] In other words, we must find the kernel of the matrix \(A-2 I\). We can do this through row reduction of the matrix \[A-2 I = \matrix{4 & 0 \\ 6 & 2} - \matrix{2 & 0 \\0& 2 }=\matrix{ 2 & 0 \\ 6 & 0}\] This can be done as follows:
\[\begin{aligned}
\matrix{2&0\\6&0\\}&\sim\matrix{1&0\\6&0\\}&{\blue{\begin{array}{c}{{1}\over{2}}R_1\\\phantom{x}\end{array}}}\\\\ &\sim\matrix{1&0\\0&0\\}&{\blue{\begin{array}{c}\phantom{x}\\R_2-6R_1\end{array}}} \end{aligned}\] So the eigenspace for \(\lambda = 2\) equals \(\left\{ r \cv{0\\-1} \middle|\;r\in\mathbb R\right\}=\left\langle\cv{0\\-1}\right\rangle\).
Let \(\lambda = 4\) be an eigenvalue of the matrix \[A=\matrix{4 & 0 \\ 6 & 2}\] Then there must be a vector \(\vec{v}\) such that \(A\vec{v}=4\vec{v}\), that is, \[(A-4 I)\vec{v}=\vec{0}\text.\] In other words, we must find the kernel of the matrix \(A-4 I\). We can do this through row reduction of the matrix \[A-4 I = \matrix{4 & 0 \\ 6 & 2} - \matrix{4 & 0 \\0& 4 }=\matrix{ 0 & 0 \\ 6 & -2}\] This can be done as follows:
\[\begin{aligned}
\matrix{0&0\\6&-2\\}&\sim\matrix{6&-2\\0&0\\}&{\blue{\begin{array}{c}R_2\\R_1\end{array}}}\\\\ &\sim\matrix{1&-{{1}\over{3}}\\0&0\\}&{\blue{\begin{array}{c}{{1}\over{6}}R_1\\\phantom{x}\end{array}}} \end{aligned}\] So the eigenspace for \(\lambda = 4\) equals \(\left\{ r \cv{1\\3} \middle|\;r\in\mathbb R\right\}=\left\langle\cv{1\\3}\right\rangle\).
If possible, we avoided fractions in the solution.
\[ \begin{aligned}
\det(A-\lambda I) = \left\vert \begin{array}{cc} 4-\lambda & 0 \\ 6 & 2-\lambda \end{array} \right\vert &= (4-\lambda)(2-\lambda)-0\cdot6 \\
&= (4-\lambda)(2-\lambda)
\end{aligned}\] So, the eigenvalues are \(\lambda_1 = 2\) and \(\lambda_2 = 4\).
Let \(\lambda = 2\) be an eigenvalue of the matrix \[A=\matrix{4 & 0 \\ 6 & 2}\] Then there must be a vector \(\vec{v}\) such that \(A\vec{v}=2\vec{v}\), this is, \[(A-2 I)\vec{v}=\vec{0}\text.\] In other words, we must find the kernel of the matrix \(A-2 I\). We can do this through row reduction of the matrix \[A-2 I = \matrix{4 & 0 \\ 6 & 2} - \matrix{2 & 0 \\0& 2 }=\matrix{ 2 & 0 \\ 6 & 0}\] This can be done as follows:
\[\begin{aligned}
\matrix{2&0\\6&0\\}&\sim\matrix{1&0\\6&0\\}&{\blue{\begin{array}{c}{{1}\over{2}}R_1\\\phantom{x}\end{array}}}\\\\ &\sim\matrix{1&0\\0&0\\}&{\blue{\begin{array}{c}\phantom{x}\\R_2-6R_1\end{array}}} \end{aligned}\] So the eigenspace for \(\lambda = 2\) equals \(\left\{ r \cv{0\\-1} \middle|\;r\in\mathbb R\right\}=\left\langle\cv{0\\-1}\right\rangle\).
Let \(\lambda = 4\) be an eigenvalue of the matrix \[A=\matrix{4 & 0 \\ 6 & 2}\] Then there must be a vector \(\vec{v}\) such that \(A\vec{v}=4\vec{v}\), that is, \[(A-4 I)\vec{v}=\vec{0}\text.\] In other words, we must find the kernel of the matrix \(A-4 I\). We can do this through row reduction of the matrix \[A-4 I = \matrix{4 & 0 \\ 6 & 2} - \matrix{4 & 0 \\0& 4 }=\matrix{ 0 & 0 \\ 6 & -2}\] This can be done as follows:
\[\begin{aligned}
\matrix{0&0\\6&-2\\}&\sim\matrix{6&-2\\0&0\\}&{\blue{\begin{array}{c}R_2\\R_1\end{array}}}\\\\ &\sim\matrix{1&-{{1}\over{3}}\\0&0\\}&{\blue{\begin{array}{c}{{1}\over{6}}R_1\\\phantom{x}\end{array}}} \end{aligned}\] So the eigenspace for \(\lambda = 4\) equals \(\left\{ r \cv{1\\3} \middle|\;r\in\mathbb R\right\}=\left\langle\cv{1\\3}\right\rangle\).
If possible, we avoided fractions in the solution.
Unlock full access