Notes - Linear Algebra II HT23, Eigenvectors and eigenvalues


Flashcards

Let $V$ be a vector space over $\mathbb R$ and let $T : V \to V$ be a linear map. What is the definition of an eigenvector of $T$?


A vector $v$ is an eignevector of $T$ if

\[Tv = \lambda v\]

for some $\lambda \in \mathbb R$.

Let $V$ be a vector space over $\mathbb R$ and let $T : V \to V$ be a linear map. What is the definition of an eigenvalue of $T$?


$\lambda \in \mathbb R$ is an eigenvalue of $T$ if there is a $v \in V$ such that $v \ne 0$ and $Tv = \lambda v$.

What’s a quick proof that $\lambda$ is an eigenvalue of $T$ if and only if $\ker (T - \lambda I)\ne \{ 0 \}$?


\[\begin{aligned} \lambda \text{ eigenvalue} &\iff Tv = \lambda v \text{ where } v \ne 0 \\\\ &\iff (T - \lambda I) v = 0 \text{ where } v \ne 0 \\\\ &\iff \ker(T - \lambda I) \ne \\{0\\}. \end{aligned}\]

What is the characteristic polynomial of a matrix $A \in \mathbb R^{n \times n}$?


$\chi _ A (t) = \det (A - t I)$.

What’s the standard way of finding the eigenvalues of a linear transformation/matrix $T$?


Finding the roots of the characteristic polynomial $\chi _ T(t)$

What are the roots of the characteristic polynomial $\chi _ A(t)$?


The eigenvalues of $A$.

If $\lambda _ 1, \ldots, \lambda _ n$ are the eigenvalues of $A$, then what is $\det A$?


\[\lambda_1 \ldots \lambda_n\]

If $\lambda _ 1, \ldots, \lambda _ n$ are the eigenvalues of $A$, then what is $\text{Tr}(A)$?


\[\lambda_1 + \ldots + \lambda_n\]

What do the first two and last terms of the characteristic polynomial $\chi _ A(t)$ look like for some matrix $A \in \mathbb R^{n \times n}$?


\[\begin{aligned} \chi_A(t) = (-1)^n t^n &+ (-1)^{n-1}\text{Tr}(A) \\\\ &+ \ldots \\\\ &+ \det A \end{aligned}\]

If a linear map $T : V \to V$ has a basis cosnsiting of eigenvectors of $T$, what does the matrix for $T$ look like with respect to that basis?


\[\begin{bmatrix} \lambda_1 & 0 & \dots & 0 \\\\ 0 & \lambda_2 & \dots & 0 \\\\ \vdots & \vdots & \ddots & \vdots \\\\ 0 & 0 & \dots & \lambda_n \end{bmatrix}\]

Can you state what it means for a linear map $T : V \to V$ to be diagonalisable?


$V$ has a basis consisting of the eigenvectors of $T$.

Can you compare the condition for a matrix $A$ to be diagonalisable or invertible in terms of its eigenvalues?


  • Invertible, all eigenvalues nonzero.
  • Diagonalisable, enough eigenvalues to form a basis for $V$.

If a matrix $A \in \mathbb R^{n \times n}$ has eigenvalues $\lambda _ 1, \ldots, \lambda _ m$ ($m \le n$), then what is true about the corresponding eigenvectors?


They are linearly independent.

If a matrix $A \in \mathbb{R}^{n\times n}$ is diagonalisable, what notation is commonly used for the diagonal matrix consisting of the eigenvalues of $A$ on the diagonal?


\[\Lambda = \begin{bmatrix} \lambda_1 & 0 & \dots & 0 \\\\ 0 & \lambda_2 & \dots & 0 \\\\ \vdots & \vdots & \ddots & \vdots \\\\ 0 & 0 & \dots & \lambda_n \end{bmatrix}\]

Let $A \in \mathbb R^{n\times n}$ be a diagonalisable matrix (i.e. it has enough eigenvectors to form a basis for $V$), $S$ be the matrix consisting of the corresponding eigenvectors as columns, and $\Lambda$ be the matrix with the eigenvalues of $A$ along the diagonal. How can you relate $\Lambda$ and $A$?


\[\Lambda = S^{-1}AS\]

Let $A \in \mathbb R^{n\times n}$ be a diagonalisable matrix (i.e. it has enough eigenvectors to form a basis for $V$), $S$ be the matrix consisting of the corresponding eigenvectors as columns, and $\Lambda$ be the matrix with the eigenvalues of $A$ along the diagonal. What’s a quick proof that

\[\Lambda = S^{-1}AS\]

\[AS=S\Lambda \implies S^{-1}AS = \Lambda\]

What is the definition of $E _ \lambda$, the eigenspace corresponding to $\lambda$?


\[E_\lambda = \ker(T - \lambda I)\]

What is the geometric multiplicity $g _ \lambda$ of an eigenvalue $\lambda$?


The dimension of $E _ \lambda$, the eigenspace of $\lambda$.

What is the algebraic multiplicity $a _ \lambda$ of an eigenvalue $\lambda$?


The multiplicity of $\lambda$ as a root of $\chi _ A(t)$.

What are the two ways that a matrix can fail to be diagonalisable over $\mathbb F$?


  1. $\chi _ A(t)$ doesn’t have $n$ roots over $\mathbb F$.
  2. $g _ \lambda < a _ \lambda$ (does not depend on $\mathbb F$).

What’s the only way a matrix can fail to be diagonalisable over $\mathbb C$?


\[g_\lambda < a_\lambda\]

for some eigenvalue $\lambda$.

When proving that if a matrix/linear transformation $T$ has $m$ distinct eigenvectors $v _ 1, \ldots, v _ m$ (corresponding to $\lambda _ 1, \ldots, \lambda _ m$), then $v _ 1, \ldots, v _ m$ are linearly independent, you argue by contradiction. What set of linearly dependent vectors do you apply the linear transformation $T - \lambda _ k I$ to?


\[v_1, \ldots, v_k\]

where $k \le m$ is the smallest $k$ such that the set is still linearly dependent.

What two facts allow you to prove that if a linear transformation is diagonalisable, then for each eigenvalue $\lambda$, the geometric multiplicity $g _ \lambda$ is equal to the algebraic multiplicity $a _ \lambda$?


\[g_\lambda \le a_\lambda\]

and

\[g_{\lambda_1} + \cdots + g_{\lambda_n} = a_{\lambda_1} + \cdots + a_{\lambda_n} = \dim V\]

What’s the basic proof idea for proving that for any eigenvalue of $T$, the geometric multiplicity $g _ \lambda$ is less than or equal to the algebraic multiplicity $a _ \lambda$?


Consider a basis for $E _ \lambda$, extend to one for $V$, and then consider the transformation matrix of $T$ and $T - x\lambda$.

Proofs

Prove that if $\ker (T - \lambda I) \ne \{ 0 \}$, then $T - \lambda I$ is not invertible.


Todo.

Prove that the characteristic polynomial $\chi _ T(t)$ of a linear transformation $T$ is well-defined and does not depend on the choice of basis for the matrix representing it.


Todo.

Prove that if $A \in \mathbb{R}^{n \times n}$, then $\chi _ A(t)$ has the form

\[\begin{aligned} \chi_A(t) = (-1)^n t^n &+ (-1)^{n-1}\text{Tr}(A) \\\\ &+ \ldots \\\\ &+ \det A \end{aligned}\]

Todo.

Prove that if a matrix $A$ has eigenvalues $\lambda _ 1, \ldots, \lambda _ n$, then

\[\text{Tr}(A) = \sum_{i=1}^n \lambda_i\]

Todo.

Prove that if a matrix $A$ has eigenvalues $\lambda _ 1, \ldots, \lambda _ n$, then

\[\det A = \prod_{i=1}^n \lambda_i\]

Todo.

Prove that if a matrix/linear transformation $T$ has $m$ distinct eigenvectors $v _ 1, \ldots, v _ m$ (corresponding to $\lambda _ 1, \ldots, \lambda _ m$), then $v _ 1, \ldots, v _ m$ are linearly independent.


Todo.

Prove that a matrix $A \in \mathbb R^{n \times n}$ is diagonalisable if and only if there exists an invertible matrix $S$ such that $\Lambda = S^{-1}AS$ where $\Lambda$ is a diagonal.


Todo.

Prove that the geometric multiplicity $g _ \lambda$ of an eigenvalue $\lambda$ is always less than or equal to the algebraic multiplicity, $a _ \lambda$.


Todo.

Prove that if a linear transformation $T$ is diagonalisable, then for each eigenvalue $\lambda$, the geometric multiplicity $g _ \lambda$ is equal to the algebraic multiplicity $a _ \lambda$.


Todo.




Related posts