Show that these elements are linearly independent

Click For Summary

Discussion Overview

The discussion revolves around proving the linear independence of a set of vectors generated by a linear transformation applied to a vector in a finite-dimensional vector space. Participants explore the use of induction to establish this property, considering various cases and the implications of the inductive hypothesis. The discussion also touches on the representation of the transformation in terms of a basis of the vector space.

Discussion Character

  • Exploratory
  • Mathematical reasoning
  • Debate/contested

Main Points Raised

  • Some participants propose using induction to show that the vectors $\phi^0(v), \phi(v), \ldots, \phi^{m-1}(v)$ are linearly independent for $m \geq 1$.
  • Others argue that the inductive hypothesis should apply to any vector $v$ such that $\phi^{m-1}(v) \neq 0 = \phi^m(v)$.
  • A later reply suggests that if the set becomes dependent upon adding $v$, it leads to a contradiction when applying the linear transformation.
  • Participants discuss the implications of the inductive step, particularly how to show that $\phi^1(v), \ldots, \phi^m(v)$ remain independent.
  • Some participants express uncertainty about the reasoning behind checking the independence of the set when adding $v$.
  • There is a discussion about how to describe the first $m-1$ columns of the matrix representation of the linear transformation with respect to a chosen basis.

Areas of Agreement / Disagreement

Participants generally agree on the use of induction to prove linear independence, but there is some uncertainty regarding specific steps in the reasoning and the implications of the inductive hypothesis. The discussion remains unresolved on certain aspects of the proof and the second question regarding the matrix representation.

Contextual Notes

Some participants note the need for clarity on the assumptions made during the inductive step and the conditions under which the vectors are considered independent. There are also discussions about the representation of vectors in the context of the chosen basis, which may depend on the definitions used.

mathmari
Gold Member
MHB
Messages
4,984
Reaction score
7
Hey! 😊

Let $1\leq n\in \mathbb{N}$, let $\mathbb{K}$ be a field, $V$ a $\mathbb{K}$-vector space with $\dim_{\mathbb{K}}V=n$ and let $\phi:V\rightarrow V$ be linear.

- Let $0\neq v\in V$ and $1\leq m\in \mathbb{N}$, such that $\phi^{m-1}(v)\neq 0=\phi^m(v)$. Show that $\phi^0(v)=v, \phi (v), \ldots ,\phi^{m-1}(v)$ are linearly independent.

Hint : Use induction for $m$ and for the case $m\geq 2$ consider $w=\phi (v)$.

- Let $v,m$ as above. Let $B=(b_1, \ldots , b_n)$ be a basis of $V$, with $b_i=\phi^{i-1}(v)$ for all $1\leq i\leq m$. Describe the first $m-1$ columns of $M_B(\phi)$.

I have done the following:

- Base case: If $m=1$ the we have the element $\phi^0(v)=v$. This element is linearly independent.

Inductive hypothesis: We assume that $\phi^0(v)=v, \phi (v), \ldots ,\phi^{m-1}(v)$ are linearly independent. (IH)

Inductive step: We want to show that $\phi^0(v)=v, \phi (v), \ldots ,\phi^{m-1}(v), \phi^m(v)$ is linearly independent.

\begin{align*}&\lambda_0\phi^0(v)+\lambda_1\phi (v)+ \ldots +\lambda_{m-1}\phi^{m-1}(v)+\lambda_m \phi^m(v)=0 \\ & \Rightarrow \lambda_m \phi^m(v)=-\lambda_0\phi^0(v)-\lambda_1\phi (v)- \ldots -\lambda_{m-1}\phi^{m-1}(v)\end{align*}

We know that the elements of the right side are linearly independent. How can we use that information
here? Could you give me a hint?
 
Last edited by a moderator:
Physics news on Phys.org
mathmari said:
Hey! 😊

Let $1\leq n\in \mathbb{N}$, let $\mathbb{K}$ be a field, $V$ a $\mathbb{K}$-vector space with $\dim_{\mathbb{K}}V=n$ and let $\phi:V\rightarrow V$ be linear.

- Let $0\leq v\in V$ and $1\leq m\in \mathbb{N}$, such that $\phi^{m-1}(v)\neq 0=\phi^m(v)$. Show that $\phi^0(v)=v, \phi (v), \ldots ,\phi^{m-1}(v)$ are linearly independent.

Hint : Use induction for $m$ and for the case $m\geq 2$ consider $w=\phi (v)$.

- Let $v,m$ as above. Let $B=(b_1, \ldots , b_n)$ be a basis of $V$, with $b_i=\phi^{i-1}(v)$ for all $1\leq i\leq m$. Describe the first $m-1$ columns of $M_B(\phi)$.

I have done the following:

- Base case: If $m=1$ the we have the element $\phi^0(v)=v$. This element is linearly independent.

Inductive hypothesis: We assume that $\phi^0(v)=v, \phi (v), \ldots ,\phi^{m-1}(v)$ are linearly independent. (IH)

Inductive step: We want to show that $\phi^0(v)=v, \phi (v), \ldots ,\phi^{m-1}(v), \phi^m(v)$ is linearly independent.

\begin{align*}&\lambda_0\phi^0(v)+\lambda_1\phi (v)+ \ldots +\lambda_{m-1}\phi^{m-1}(v)+\lambda_m \phi^m(v)=0 \\ & \Rightarrow \lambda_m \phi^m(v)=-\lambda_0\phi^0(v)-\lambda_1\phi (v)- \ldots -\lambda_{m-1}\phi^{m-1}(v)\end{align*}

We know that the elements of the right side are linearly independent. How can we use that information
here? Could you give me a hint?
Hey mathmari!

I believe the inductive hypothesis should be:

For some $m\ge 1$ we assume that for any $v\in V$ such that $\phi^{m-1}(v)\neq 0=\phi^m(v)$ we have that $\phi^0(v)=v, \phi (v), \ldots ,\phi^{m-1}(v)$ are linearly independent (IH).

Then we have to prove in the inductive step that for any $v\in V$ such that $\phi^{m}(v)\neq 0=\phi^{m+1}(v)$ we have that $\phi^0(v)=v, \phi (v), \ldots ,\phi^{m}(v)$ are linearly independent.

Now consider $w=\phi(v)$. 🤔
 
Last edited:
We have $\phi^0(w)=\phi^1(v),\ldots,\phi^{m-1}(w)=\phi^{m}(v)\ne 0,\phi^m(w)=\phi^{m+1}(v)=0$.
According to the inductive hypothesis this means that $\phi^1(v),\ldots, \phi^{m}(v)$ are independent.
What is left to prove, is that if we add $v$ to the set, that it is still independent, yes? :unsure:

So suppose the resulting set in not independent.
Then it must be possible to write $v$ as a linear combination of $\phi^1(v),\ldots, \phi^{m}(v)$.
Apply $\phi$ on both sides. Can we reach a contradiction? 🤔
 
Klaas van Aarsen said:
We have $\phi^0(w)=\phi^1(v),\ldots,\phi^{m-1}(w)=\phi^{m}(v)\ne 0,\phi^m(w)=\phi^{m+1}(v)=0$.
According to the inductive hypothesis this means that $\phi^1(v),\ldots, \phi^{m}(v)$ are independent.

From the inductive hypothesis we have that $\phi^0(v)=v, \phi (v), \ldots ,\phi^{m-1}(v)$ are independent, right? How do we get that $\phi^1(v),\ldots, \phi^{m}(v)$ are independent? :unsure:
Klaas van Aarsen said:
What is left to prove, is that if we add $v$ to the set, that it is still independent, yes? :unsure:

I got stuck right now. Why do we check if the set is still independent when we add the vector $v$ ? Because above we don't have $\phi^0(v)$? :unsure:
 
mathmari said:
From the inductive hypothesis we have that $\phi^0(v)=v, \phi (v), \ldots ,\phi^{m-1}(v)$ are independent, right? How do we get that $\phi^1(v),\ldots, \phi^{m}(v)$ are independent?

Let $w=\phi(v)$ in the inductive step. Then we have $\phi^0(w)=w, \phi (w), \ldots ,\phi^{m-1}(w)\ne 0,\phi^m(w)=0$ yes?
Can we apply the IH from post #2 to that? 🤔

mathmari said:
I got stuck right now. Why do we check if the set is still independent when we add the vector $v$ ? Because above we don't have $\phi^0(v)$?
Let's apply IH from post #2 to $w=\phi(v)$ first and see what comes out of it. Then we'll get back to this. 🤔
 
Last edited:
So do we have the following?

Bas Case: If $m=1$ then we have one element $\phi^0(v)=v$. This element is linearnly independent.

Inductive Hypothesis: For a $m\ge 1$ we assume that for a $v\in V$ such that $\phi^{m-1}(v)\neq 0=\phi^m(v)$ we have that $\phi^0(v)=v, \phi (v), \ldots ,\phi^{m-1}(v)$ are linearly independent. (IH)

Inductive Step: We want to show that for $v\in V$ with $\phi^{m}(v)\neq 0=\phi^{m+1}(v)$ it holds that $\phi^0(v)=v, \phi (v), \ldots ,\phi^{m}(v)$ are linearly independent.
We consider $w=\phi(v)$.
Then we have that $\phi^0(w)=w, \phi (w), \ldots ,\phi^{m-1}(w)\ne 0,\phi^m(w)=0$.
From (IH) we have that $\phi^0(w), \phi (w), \ldots ,\phi^{m-1}(w)$ are lineanrlyindependent. Since $\phi^0(w)=\phi^1(v),\ldots,\phi^{m-1}(w)=\phi^{m}(v)$, it follows that $\phi^1(v),\ldots,\phi^{m}(v)$ are linearly independent.
We check if the set remains linearly independent even if we add $v$ to the set.
We assume that this set is linearly dependent
Then we can write $v$ as a linear combination of $\phi^1(v),\ldots, \phi^{m}(v)$.
We apply $\phi$ at both sides.
\begin{align*}&v=\lambda_1\phi^1(v)+\ldots+\lambda_{m-1}\phi^{m-1}(v)+\lambda_m \phi^{m}(v) \\ & \Rightarrow \phi (v)=\phi \left (\lambda_1\phi^1(v)+\ldots+\lambda_{m-1}\phi^{m-1}(v)+\lambda_m \phi^{m}(v)\right ) \\ & \Rightarrow \phi (v)=\lambda_1\phi^2(v)+\ldots+\lambda_{m-1}\phi^{m}(v)+\lambda_m \phi^{m+1}(v) \\ & \Rightarrow \phi (v)=\lambda_1\phi^2(v)+\ldots+\lambda_{m-1}\phi^{m}(v)+\lambda_m \cdot 0 \\ & \Rightarrow \phi (v)=\lambda_1\phi^2(v)+\ldots+\lambda_{m-1}\phi^{m}(v) \\ & \Rightarrow -\phi (v)+\lambda_1\phi^2(v)+\ldots+\lambda_{m-1}\phi^{m}(v)=0 \\ & \Rightarrow -\phi^1 (v)+\lambda_1\phi^2(v)+\ldots+\lambda_{m-1}\phi^{m}(v)=0\end{align*}
This means that $\phi^1(v),\ldots,\phi^{m}(v)$ are linearly dependent, since not all coefficients are zero.
A contradiction. Therefore $\phi^0(v)=v, \phi (v), \ldots ,\phi^{m}(v)$ are linearly independent.Is everything correct? :unsure:Could you give me also a hint for the second question? :unsure:
It must hold that $\phi (b_i)=b_i\Rightarrow \phi^i(v)=b_i$, or not?
 
Last edited by a moderator:
mathmari said:
Is everything correct?

Could you give me also a hint for the second question?
Yep. (Nod)

The columns of the matrix of $\phi$ with respect to $B$ are the images of the basis vectors with respect to $B$ aren't they?
The first basis vector is $b_1$, which is represented by $\begin{pmatrix}1\\0\\\vdots\\0\end{pmatrix}$.
And its image (assuming $m\ge 2$) is $b_2$, which is represented by $\begin{pmatrix}0\\1\\0\\\vdots\\0\end{pmatrix}$, which is also the first column of the matrix then, isn't it? 🤔
 
Klaas van Aarsen said:
The columns of the matrix of $\phi$ with respect to $B$ are the images of the basis vectors with respect to $B$ aren't they?
The first basis vector is $b_1$, which is represented by $\begin{pmatrix}1\\0\\\vdots\\0\end{pmatrix}$.
And its image (assuming $m\ge 2$) is $b_2$, which is represented by $\begin{pmatrix}0\\1\\0\\\vdots\\0\end{pmatrix}$, which is also the first column of the matrix then, isn't it? 🤔

So is the $i$-th column of the matrix the vector $e_{i+1}$ ? :unsure:
 
mathmari said:
So is the $i$-th column of the matrix the vector $e_{i+1}$ ?
Yep. Up to column $m-1$ that is. We don't know what the remaining columns are. 🤔
 
  • #10
Klaas van Aarsen said:
Yep. Up to column $m-1$ that is. We don't know what the remaining columns are. 🤔

Do we know that this holds till the column $m-1$ because we only know that the $\phi^i$'s for $0\leq i\leq m-1$ are linearly independent? :unsure:
 
  • #11
mathmari said:
Do we know that this holds till the column $m-1$ because we only know that the $\phi^i$'s for $0\leq i\leq m-1$ are linearly independent?
It's not about linear independence is it?
Just that we only know $\phi^i(b_1)=\phi^i(v)$ up to $i=m-1$. 🤔
 
  • #12
Klaas van Aarsen said:
It's not about linear independence is it?
Just that we only know $\phi^i(b_1)=\phi^i(v)$ up to $i=m-1$. 🤔

How do we get $\phi^i(b_1)=\phi^i(v)$ ?
We have that $b_i=\phi^{i-1}(v) \Rightarrow \phi (b_i)=\phi^{i}(v) $, or not? :unsure:
 
  • #13
mathmari said:
How do we get $\phi^i(b_1)=\phi^i(v)$ ?
We have that $b_i=\phi^{i-1}(v) \Rightarrow \phi (b_i)=\phi^{i}(v) $, or not?
I should have written $\phi^i(b_1)=\phi^{i-1}(v)$. 😶
 

Similar threads

  • · Replies 24 ·
Replies
24
Views
2K
  • · Replies 5 ·
Replies
5
Views
2K
  • · Replies 1 ·
Replies
1
Views
2K
  • · Replies 10 ·
Replies
10
Views
2K
  • · Replies 23 ·
Replies
23
Views
2K
  • · Replies 23 ·
Replies
23
Views
2K
  • · Replies 26 ·
Replies
26
Views
1K
  • · Replies 4 ·
Replies
4
Views
1K
  • · Replies 9 ·
Replies
9
Views
2K
  • · Replies 39 ·
2
Replies
39
Views
3K