16
$\begingroup$

I've searched a lot for a simple explanation of this. Given a Jordan block $J_k(\lambda)$, its $n$-th power is:

$$ J_k(\lambda)^n = \begin{bmatrix} \lambda^n & \binom{n}{1}\lambda^{n-1} & \binom{n}{2}\lambda^{n-2} & \cdots & \cdots & \binom{n}{k-1}\lambda^{n-k+1} \\ & \lambda^n & \binom{n}{1}\lambda^{n-1} & \cdots & \cdots & \binom{n}{k-2}\lambda^{n-k+2} \\ & & \ddots & \ddots & \vdots & \vdots\\ & & & \ddots & \ddots & \vdots\\ & & & & \lambda^n & \binom{n}{1}\lambda^{n-1}\\ & & & & & \lambda^n \end{bmatrix}$$

Why does the $n$th power involve the binomial coefficient?

$\endgroup$
  • 1
    $\begingroup$ Have you tried a few simple cases to see what happens? $\endgroup$ – Bruno Joyal Mar 10 '13 at 18:36
21
$\begingroup$

Let $N$ denote the nilpotent matrix whose superdiagonal contains ones and all other entries are zero. Then $N^k=0$. Hence, by the binomial theorem: $$J_k(\lambda)^n=(\lambda I+N)^n=\sum_{r=0}^\color{red}{n} \binom{n}{r}\lambda^{n-r} N^r=\sum_{r=0}^\color{red}{\min(n,k-1)} \binom{n}{r}\lambda^{n-r} N^r.$$

$\endgroup$
  • 8
    $\begingroup$ It's necessary to mention that $\lambda I$ and $N$ commute. $\endgroup$ – Git Gud Mar 10 '13 at 18:46
  • $\begingroup$ Does $N$ becomes $0$ (in the sum) as soon as $r = k$, right? $\endgroup$ – user34295 Mar 10 '13 at 20:02
  • $\begingroup$ Found it, $N$ should be the canonical nilpotent matrix, that is of order $n$ and degree $n$... correct? $\endgroup$ – user34295 Mar 10 '13 at 20:03
  • 3
    $\begingroup$ @Gremo $N$ is actually identical to $J_k(0)$. It is $k\times k$, i.e. its size conforms to the size of $J_k(\lambda)$. The first exponent that makes $N^r=0$ is $r=k$. For $1\le r<k$, the power $N^r$ is the matrix that has all ones in the $r$-th superdiagonal and zero elsewhere. $\endgroup$ – user1551 Mar 10 '13 at 20:12
23
$\begingroup$

user1551 already gave the "simple explanation", so let me just mention the general result for evaluating matrix functions when the argument is a Jordan block. Given the $n\times n$ Jordan block with eigenvalue $\lambda$,

$$\mathbf J=\begin{pmatrix}\lambda&1&&\\&\lambda&\ddots&\\&&\ddots&1\\&&&\lambda\end{pmatrix}$$

the matrix $f(\mathbf J)$ looks like this:

$$f(\mathbf J)=\begin{pmatrix}f(\lambda)&f^\prime(\lambda)&\cdots&\frac{f^{(n-1)}(\lambda)}{(n-1)!}\\&f(\lambda)&\ddots&\vdots\\&&\ddots&f^\prime(\lambda)\\&&&f(\lambda)\end{pmatrix}$$

This general result is proven in a number of references; e.g. this.

For the power function $f(x)=x^k$, we have the general result (easily proven inductively):

$$\frac1{j!}\frac{\mathrm d^j}{\mathrm dx^j}x^k=\binom{k}{j}x^{k-j}$$

Making the necessary replacements gives the formula for the power of a Jordan block.

$\endgroup$

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.