2
$\begingroup$

Let $A$ be a $4\times 4$ matrix over $\mathbb C$ such that $\operatorname{rank} A=2$ and $A^3=A^2\neq 0$

Suppose $A$ is not diagonalizable.

Then which one is true

  1. $A^2=A\neq 0$

  2. $\exists$ a vector $v$ such that $Av\neq 0$ but $A^2v= 0$

  3. the characteristic polynomial of $A$ is $x^4-x^3$

My try:

If $A^2=A$ then minimal polynomial of $A$ has two distinct roots and hence $A$ is diagonalisable hence false. Since $A^3=A^2\neq 0$ so the characteristic roots of $A$ are $0,0,1$ Since rank of $A$ is $2$ $1$ must appear twice so the characteristic polynomial of $A=x^4-(0+0+1+1)x^3+(0.0.1+0.1.1+1.1.0+1.0.1)x^2+(0.0+0.1+1.1+...)x+(0.0.1.1)$ hence 3 is false

2 I am not sure Please find the error in my solution and don't refer it to some previous solution

$\endgroup$
3
$\begingroup$

The minimal polynomial $m(\lambda)$ for $A$ must divide $\lambda^{2}(\lambda-1)$ because $$ 0=A^{3}-A^{2}=A^{2}(A-I). $$ The factor of $\lambda-1$ must be present in $m(\lambda)$ because $(A-I)A^{2}=0$ and $A^{2}\ne 0$, which means $A-I$ has a non-trivial null space. A factor of $\lambda$ must be present in $m(\lambda)$ because, otherwise, $A=I$ is diagonalizable. If $m(\lambda)=\lambda(\lambda-1)$, then $A$ is diagonalizable. Therefore, $m(\lambda)=\lambda^{2}(\lambda-1)$.

The presence of $\lambda^{2}$ in the minimal polynomial means that there is a vector $v$ such that $A^{2}(A-I)w=0$ but $A(A-I)w \ne 0$. Therefore, $v=(A-I)w$ satisfies $Av \ne 0$ and $A^{2}v=0$.

The characteristic polynomial $p$ must be divisible by $m$, and $p$ cannot have factors other than $\lambda$ and $\lambda-1$. And $p$ cannot have a second factor of $\lambda-1$ unless the null space of $A-I$ is two-dimensional, which would make the rank of $A$ at least $3$ because the range of $A$ would include the null space of $A-I$ and the vector $Av$ described above, a vector which is independent of the null space of $A-I$. So $p$ has to be $\lambda^{3}(\lambda-1)=\lambda^{4}-\lambda^{3}$.

| cite | improve this answer | |
$\endgroup$
2
$\begingroup$

I believe both (2) and (3) are true. I will use $(N(A),C(A)$ to stand for the null space of $A$ and column space of $A$ respectively.

We can establish the following: $$ (a)\ \dim(N(A)) = \dim(c(A) = 2\\ (b)\ \dim(N(A^2)) = 3, \dim(C(A^2)) = 1\\ (c)\ A \text{ is similar to } \pmatrix{0&1&0&0\\0&0&0&0\\0&0&0&0\\0&0&0&1} $$

proof: of (a) is immediate from the fact $rank(A) = 2$ and nullity theorem $\dim(C(A)) + dim(N(A)) = \text{ number of columns of } A$

(b) that $c(A^2) \subset c(A)$ and $\dim(C(A)) = 2$ gives us $\dim(C(A^2)) = 0, 1, or\ 2.$ $A^2 \neq 0$ rules out the possibility $0.$ suppose $dim(C(A^2)) = 2.$ we will derive a contradiction to the fact that $A$ is not diagonalizable. the fact $A^3 = A^2$ implies that any nonzero $A^2x,$ which is in $C(A^2),$ is an eigenvector of $A$ corresponding to the eigenvalue $1.$ now we have two linearly independent eigenvectors corresponding to eigenvalue $1$ and we already have two eigenvectors corresponding to the eigenvalue in $N(A).$ that makes $A$ diagonalizable and the contradiction. we are done with proving (b).

(c) from the proof of (b) we know that there is $u_2 \neq 0 \in N(A^2) \setminus N(A).$ if we set $u_1 = Au_2,$ then $0 \neq u_1 \in N(A)$ and now find the second $u_3$ so that $\{u_2, u_3\} \text{ is a basis for } N(A)$ now choose an eigenvector $u_4$ corresponding to the eigenvalue $1$ of $A$ so that $\{u_1, u_2, u_3, u_4\}$ is a basis for $R^4.$

let me collect all the information on the u's: $$Au_2 = u_1, Au_1 = 0, Au_3 = 0, Au_4 = u_4 $$ with respect to the basis $\{u_1, u_2, u_3, u_4\}$ the matrix $A$ has the representation given in (c).

| cite | improve this answer | |
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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