Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Prove that if $A\in M_{n}\left(\mathbb{F}\right)$ matrix with a characteristic polynomial that can be written as a product of linear elements (?) ($\left(x-\lambda_{1}\right)^{r_1}\dots\left(x-\lambda_{k}\right)^{r_k}$) that $A$ is similar to an upper triangular matrix.

My attempt: Well first I showed that if $\lambda$ is an eigenvalue of $A$ then it can be written as

$P^{-1}AP=\left(\begin{matrix}\lambda & * & \cdots & *\\ 0 & \ulcorner & & \urcorner\\ \vdots & & B\\ 0 & \llcorner & & \lrcorner \end{matrix}\right)$

for some invertible $P \in M_{n}\left(\mathbb{F}\right)$.

Then by induction on n I showed that for a $2\times 2$ matrix the claim holds using the above lemma. I've assumed correctness for some $n$. And then I tried proving it for $n+1$: I used the lemma for the $\left(n+1\right) \times \left(n+1\right)$ matrix (because we assume that its characteristic polynomial can be reduced).

Then I tried using the induction hypothesis on the inner $B$ of the $\left(n+1\right) \times \left(n+1\right)$ matrix. So I've concluded that there's some matrix $Q\in M_{n}\left(\mathbb{F}\right)$ so that $Q^{-1}BQ$ is upper triangular.

The next step was to use it for my current matrix so I've decided to cunstruct the following matrices:

$ Q^{'}=\left(\begin{matrix}1 & 0 & \cdots & 0\\ 0 & \ulcorner & & \urcorner\\ \vdots & & Q\\ 0 & \llcorner & & \lrcorner \end{matrix}\right) $ and $ Q^{'-1}=\left(\begin{matrix}1 & 0 & \cdots & 0\\ 0 & \ulcorner & & \urcorner\\ \vdots & & Q^{-1}\\ 0 & \llcorner & & \lrcorner \end{matrix}\right) $

So that now I can claim that

$Q^{'-1}\left(\begin{matrix}\lambda & * & \cdots & *\\ 0 & \ulcorner & & \urcorner\\ \vdots & & B\\ 0 & \llcorner & & \lrcorner \end{matrix}\right)Q^{'} = \left(\begin{matrix}\lambda & * & \cdots & *\\ 0 & \ulcorner & & \urcorner\\ \vdots & & Q^{-1}BQ\\ 0 & \llcorner & & \lrcorner \end{matrix}\right) $

which is essentially an upper triangular matrix.

But I'm having hard time determining whether it's correct at all because to be honest a lot of indices confuse me :) and I would really like to know whether my proof is correct and is there a more solid\clear\intuitive way to explain the last conclusion.

Edit 1: I'll try to make my question clearer: We are usually required to be extremely rigorous, so the $Q^{'-1} P^{-1} A P Q^{'}$ part needs to be explained as well but I can't seem to find a better explanation then something like "well it has identity in the upper left corner so that's what happens" but I can't figure a nice way of writing this mathematically without babbling too much.

share|cite|improve this question
It looks good and crystal clear to me. – user1551 Apr 14 '13 at 2:56
@user1551 thank you, I've tried to clarify my question. – Scis Apr 14 '13 at 8:50
up vote 0 down vote accepted

Just so that this question does not remain unanswered, I'll paraphrase the proof given in the question in a less technical way.

A matrix $A$ is similar to an upper triangular matrix if and only if for the linear operator $\phi$ of $V=F^n$ defined by $A$ there exist a chain of $\phi$-stable subspaces $$ \{0\}=V_0\subset V_1\subset V_2\subset\cdots\subset V_n=V $$ where $\dim(V_i)=i$ for all $i$. This is so because if $\def\B{\mathcal B}\B=[b_1,\ldots,b_n]$ is any basis, then $\operatorname{Mat}_\B(\phi)$ is upper triangular if and only if each of the subspaces $\langle b_1,\ldots,b_d\rangle$ is $\phi$-stable, for $d=1,2,\ldots,n$.

Now a proof that if the characteristic polynomial $\chi_A$ splits into a product of linear factors $X-\lambda_i$, then $A$ is similar to an upper triangular matrix (by the way, the converse is obvious). Proceed by induction on $n=\dim V$, the case $n=0$ being obvious. Now let $n>0$, and let $X-\lambda$ be one of the factors of $\chi_A$, then there is some eigenvector $v$ of$~A$ for $\lambda$. Put $V_1=\langle v\rangle$, then since $V_1$ is $\phi$-stable, $\phi$ induces a linear operator $\overline\phi$ of the quotient space $V/V_1$, which is of dimension$~n-1$. The characteristic polynomial of $\overline\phi$ is $\chi_A/(X-\lambda)$, so it also splits into a product of linear factors. Then by induction there is a chain of $\overline\phi$-stable subspaces $$ V_1/V_1=W_0\subset W_1\subset W_2\subset\cdots\subset W_{n-1}=V/V_1 $$ with $\dim W_i=i$ for all $i$. Being a subspace of the quotient space we can write $W_i=V_{i+1}/V_1$ for a uniquely defined subspace $V_{i+1}$ of$~V$, which has dimension $i+1$, and is $\phi$-stable (because $W_i$ is $\overline\phi$-stable). Together with $V_0=\{0\}$ this provides our chain of subspaces that shows that $A$ is similar to an upper triangular matrix, completing the proof.

share|cite|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.