Homework 11
-
(10 pts) Let \(A \in M_n(\mathbb{F})\). Prove that \(A\) and \(A^T\) have the same eigenvalues. Give an example where the eigenvectors are different.
Assume \(A, A^T \in M_n(\mathbb{F})\). Then \(\chi_{A}(x) = det(xI{d} - A)\)
Then \(\chi_{A^{T}}(x)=\det(xId-A^{T})=\det(xId^{T}-A^{T})=\det(xId-A)=\chi_{A}(x)\).
Thus \(A\) and \(A^T\) have same eigenvalues.
Let \(A = \begin{pmatrix} 1 & 0 \\ 1 & 1 \end{pmatrix}\), \(A^T = \begin{pmatrix} 1 & 1 \\ 0 & 1 \end{pmatrix}\) and \(\chi_A(x) = (x-1)^2 = \chi_{A^T}(x)\), then \(x=1\).
\(E_1\) of \(A\) is \(\text{ker}(A - Id) = \text{ker} \begin{pmatrix} 0 & 0 \\ 1 & 0 \end{pmatrix} = \langle (0,1) \rangle\)
\(E_1\) of \(A^T\) is \(\text{ker}(A^T - Id) = \text{ker} \begin{pmatrix} 0 & 1 \\ 0 & 0 \end{pmatrix} = \langle (1,0) \rangle\)
Thus they have different eigenvectors. -
(20 pts) Suppose \(S, T: V \to V\) are linear maps such that \(S\) is an isomorphism.
(a) Prove that \(T\) and \(S^{-1}TS\) have the same eigenvalues.
(b) What is the relationship between the eigenvectors of \(T\) and the eigenvectors of \(S^{-1}TS\)?(a)
Assume \(T\) has an eigenvalue \(\lambda\). Then \(T(v) = \lambda v\).
Since \(S\) is an isomorphism, then \(\ker S = \{0\}\), \(\text{Im } S = V\). Then \(\exists v' : S(v') = v\). \(S^{-1}(v) = v'\).
Then \(S^{-1}TS(v') = S^{-1}T(v) = S^{-1}\lambda v = \lambda S^{-1}(v) = \lambda v'\).
Thus they have same eigenvalues.(b)
For \(T\):
\(E_\lambda = \ker(T - \lambda Id) = \ker([T - \lambda Id]_E)\)
For \(S^{-1}TS\):
\(E_\lambda = \ker(S^{-1}TS - \lambda Id) = \ker(S^{-1}(T - \lambda Id)S)\) \(= \ker([T - \lambda Id]_{B})\) where \(S\) is change basis.
Thus \([T - \lambda Id]_E \cdot [v]_E = 0\) iff \(S^{-1}[T - \lambda Id]_E \cdot [v]_E \cdot S = 0\) iff \([T - \lambda Id]_B \cdot [v]_B = 0\).
Thus \([v]_E\) is eigenvector of \(T \iff [v]_B\) is eigenvector of \(S^{-1}TS\) where \(S\) is change of basis -
(50 pts) Decide whether or not the following linear maps \(T\) are diagonalizable and if so, find an invertible matrix \(P\) such that \(P^{-1}[T]_C P\) is a diagonal matrix. Consider them first as linear maps in \(\mathbb{R}\) and then in \(\mathbb{C}\).
(a) \(T(x, y) = (2x + 3y, -x + y)\).
In \(\R\)
First, we need \([T]_{C}=\begin{pmatrix}2&3\\-1&1\end{pmatrix}.\)
Then we calculate eigenvalues, let \(\det(T-\lambda I)=\det\begin{pmatrix}2-\lambda&3\\-1&1-\lambda\end{pmatrix}=(2-\lambda)(1-\lambda)+3=\lambda^2-3\lambda+5=0\;\Rightarrow\;\lambda=\frac{3\pm i\sqrt{11}}2.\)
Thus there is no eigenvalue in \(\R\), thus \(T\) is not diagonalizable in \(\R\)
In \(\mathbb{C}\)
First, we need \([T]_{C}=\begin{pmatrix}2&3\\-1&1\end{pmatrix}.\)
Then we calculate eigenvalues, let \(\det(T-\lambda I)=\det\begin{pmatrix}2-\lambda&3\\-1&1-\lambda\end{pmatrix}=(2-\lambda)(1-\lambda)+3=\lambda^2-3\lambda+5=0\;\Rightarrow\;\lambda=\frac{3\pm i\sqrt{11}}2.\)
Thus we find the basis of eigenvectors
For \(\displaystyle\lambda=\frac{3+i\sqrt{11}}2\): \(\text{Null}(T-\lambda I)=\text{Null}\begin{pmatrix}\tfrac{1-i\sqrt{11}}2&3\\-1&\tfrac{-1-i\sqrt{11}}2\end{pmatrix}=\bigl\langle\bigl(-\tfrac{1+i\sqrt{11}}2,\,1\bigr)\bigr\rangle.\)
For \(\displaystyle\lambda=\frac{3-i\sqrt{11}}2\): \(\text{Null}(T-\lambda I)=\text{Null}\begin{pmatrix}\tfrac{1+i\sqrt{11}}2&3\\-1&\tfrac{-1+i\sqrt{11}}2\end{pmatrix}=\bigl\langle\bigl(-\tfrac{1-i\sqrt{11}}2,\,1\bigr)\bigr\rangle.\)
Now, \(\mathcal{B}=\{(-\tfrac{1+i\sqrt{11}}2,1),\,(-\tfrac{1-i\sqrt{11}}2,1)\}\) is my basis of eigenvectors of \(T\).
Thus \([T]_{\mathcal{B}}=\begin{pmatrix}\tfrac{3+i\sqrt{11}}2&0\\0&\tfrac{3-i\sqrt{11}}2\end{pmatrix}.\)
Then \(P= \begin{pmatrix} -\tfrac{1+i\sqrt{11}}{2} & -\tfrac{1-i\sqrt{11}}{2} \\ 1 & 1 \end{pmatrix},\quad P^{-1}[T]_{C}P=[T]_{\mathcal{B}}.\)
(b) \(T(x, y, z) = (-x - 2y + 2z, 4x + 3y - 4z, -2y + z)\).
In \(\R\)
First, we need \([T]_{C}=\begin{pmatrix}-1 & -2 & 2\\ 4 & 3 & -4\\ 0 & -2 & 1\end{pmatrix}\)
Then we calculate eigenvalues, let \(\det(T-\lambda I)=\det\begin{pmatrix}-1-\lambda & -2 & 2\\ 4 & 3-\lambda & -4\\ 0 & -2 & 1-\lambda\end{pmatrix}=0\)
\(=(-1-\lambda)(3-\lambda)(1-\lambda)+4\cdot (-2)\cdot 2-(-4)\cdot(-2)(-1-\lambda)-(1-\lambda)\cdot(-2)\cdot4\)
\(=-\lambda^3+3\lambda^2+\lambda-3=0\Rightarrow\lambda=1,-1,3\)
Thus we find the basis of eigenvectors
For \(\lambda = -1\): \(\text{Null}\left(T+I\right)=\text{Null}\begin{pmatrix}0 & -2 & 2\\ 4 & 4 & -4\\ 0 & -2 & 2\end{pmatrix}=\text{Null}\begin{pmatrix}0 & 1 & -1\\ 1 & 0 & 0\\ 0 & 0 & 0\end{pmatrix}=\langle(0,1,1)\rangle\).
For \(\lambda = 1\): \(\text{Null}(T-I)=\text{Null}\begin{pmatrix}-2 & -2 & 2\\ 4 & 2 & -4\\ 0 & -2 & 0\end{pmatrix}=\text{Null}\begin{pmatrix}1 & 0 & -1\\ 0 & 1 & 0\\ 0 & 0 & 0\end{pmatrix}=\langle(1,0,1)\rangle\).
For \(\lambda = 3\): \(\text{Null}(T-3I)=\text{Null}\begin{pmatrix}-4 & -2 & 2\\ 4 & 0 & -4\\ 0 & -2 & -2\end{pmatrix}=\text{Null}\begin{pmatrix}0 & 1 & 1\\ 1 & 0 & -1\\ 0 & 0 & 0\end{pmatrix}=\langle(1,-1,1)\rangle\).
Now, \(\mathcal{B}=\{(0,1,1),\left(1,0,1),(1,-1,1\right)\}\) is my basis of eigenvectors of \(T\).
Thus \([T]_{\mathcal{B}}=\begin{pmatrix}-1 & 0 & 0\\ 0 & 1 & 0\\ 0 & 0 & 3\end{pmatrix}\)
Then \(P= \begin{pmatrix} 0 & 1 & 1 \\ 1 & 0 & -1 \\ 1 & 1 & 1 \end{pmatrix}\), then \(P^{-1}[T]_CP=[T]_B\)
Thus \(T\) is diagonalizable in \(\R\), then \(T\) is also diagonalizable in \(\mathbb{C}\)
(c) \(T(x, y, z) = (6x - 3y - 2z, 4x - y - 2z, 10x - 5y - 3z)\).
In \(\R\)
First, we need \([T]_{C}=\begin{pmatrix}6&-3&-2\\4&-1&-2\\10&-5&-3\end{pmatrix}.\)
Then we calculate eigenvalues, let \(\det(T-\lambda I)=\det\begin{pmatrix}6-\lambda&-3&-2\\4&-1-\lambda&-2\\10&-5&-3-\lambda\end{pmatrix}=0\;\Rightarrow\;- \lambda^3+2\lambda^2-\lambda+2=0\;\Rightarrow\;\lambda^3-2\lambda^2+\lambda-2=(\lambda-2)(\lambda^2+1)=0\) \(\Rightarrow\;\lambda=2,\;i,\;-i.\)
Thus we find the basis of eigenvectors
For \(\lambda=2\): \(\text{Null}(T-2I)=\text{Null}\begin{pmatrix}4&-3&-2\\4&-3&-2\\10&-5&-5\end{pmatrix}=\bigl\langle(1,0,2)\bigr\rangle.\)
Then eigenspace for \(\lambda =2\) is less than \(3\), thus \(T\) is not diagonalizable in \(\R\)
But if in \(\mathbb{C}\)
For \(\lambda=i\): \(\text{Null}(T-iI)=\text{Null} \begin{pmatrix} 6-i & -3 & -2 \\ 4 & -1-i & -2 \\ 10 & -5 & -3-i \end{pmatrix}=\lang(3+i,3+i,5){\rangle}.\)
For \(\lambda=-i\): \(\text{Null}(T+iI)=\text{Null} \begin{pmatrix} 6+i & -3 & -2 \\ 4 & -1+i & -2 \\ 10 & -5 & -3+i \end{pmatrix}=\lang(3-i,3-i,5){\rangle}.\)
Now, \(\mathcal{B}=\{(1,0,2),\;(3+i,3+i,5),\;(3-i,3-i,5)\}\) is my basis of eigenvectors of \(T\).
Thus \([T]_{\mathcal{B}}= \begin{pmatrix} 2 & 0 & 0 \\ 0 & i & 0 \\ 0 & 0 & -i \end{pmatrix}.\)
Then \(P_{}= \begin{pmatrix} 1 & 3+i & 3-i \\ 0 & 3+i & 3-i \\ 2 & 5 & 5 \end{pmatrix},\quad P^{-1}[T]_{C}P=[T]_{\mathcal{B}}.\)
Thus \(T\) is diagonalizable in \(\mathbb{C}\) but not in \(\R\)
(d) \(T(x, y, z, w) = (-5x - 5y - 9z + 7w, 8x + 9y + 18z - 9w, -2x - 3y - 7z + 4w, 2w)\).
In \(\R\)
First, we need \([T]_{C}=\begin{pmatrix}-5&-5&-9&7\\8&9&18&-9\\-2&-3&-7&4\\0&0&0&2\end{pmatrix}.\)
Then we calculate eigenvalues, let \(\det(T-\lambda I)=(2-\lambda)\det\begin{pmatrix}-5-\lambda&-5&-9\\8&9-\lambda&18\\-2&-3&-7-\lambda\end{pmatrix}=-(\lambda-2)(\lambda+1)^3=0\;\Rightarrow\;\lambda=2,\,-1,-1,-1.\)
For \(\lambda=2\): \(\text{Null}(T-2I)=\text{Null} \begin{pmatrix} -7 & -5 & -9 & 7 \\ 8 & 7 & 18 & -9 \\ -2 & -3 & -9 & 4 \\ 0 & 0 & 0 & 0 \end{pmatrix}=\bigl\langle(-9,63,-7,27)\bigr\rangle.\)
For \(\lambda=-1\): \(\text{Null}(T+I)=\text{Null}\begin{pmatrix}-4&-5&-9&7\\8&10&18&-9\\-2&-3&-6&4\\0&0&0&3\end{pmatrix}=\bigl\langle(\tfrac32,-3,1,0)\bigr\rangle.\)
Since the eigenspace for \(\lambda=-1\) has dimension \(1 < 3\), \(T\) is not diagonalizable.
Thus \(T\) is not diagonalizable in \(\R\) and in \(\mathbb{C}\)
-
(20pts) Suppose \(T \in \mathcal{L}(\mathbb{C}^3, \mathbb{C}^3)\) is such that \(6\) and \(7\) are eigenvalues of \(T\). Furthermore, suppose \([T]_{\mathcal{B}}\) is not a diagonal matrix with respect to any basis \(\mathcal{B}\) of \(\mathbb{C}^3\). Prove that there exists \((x, y, z) \in \mathbb{C}^3\) such that \(T(x, y, z) = (17 + 8x, \sqrt{5} + 8y, 2\pi + 8z)\).
Proof
We need to prove the existence of \((x, y, z) \in \mathbb{C}^3\) such that \(T(x, y, z) = (17 + 8x, \sqrt{5} + 8y, 2\pi + 8z)\).
Then we need to prove the existence of \((x, y, z) \in \mathbb{C}^3\) such that \(T(x,y,z)-8\left(x,y,z)=(17,\sqrt5,2\pi\right)\).
Then we need to prove the existence of \(v=(x, y, z) \in \mathbb{C}^3\) such that \(\left(T-8I\right)\left(v\right)\left.=(17,\sqrt5,2\pi\right)\).
Since \(6,7\) are eigenvalues and we know \(T\) is not diagonalizable, then \(8\) is not eigenvalue.
Then \(\dim \text{Null}(T-8I)=0\), then \(T-8I\) is injective.
Since \(T-8I\) is an endomorphism, then \(T-8I\) is surjective.
Then there exists such linear map.