Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

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

Suppose I have a linear operator defined to be $L={\partial\over\partial x}+{\partial\over\partial y}$ acting on the vector space of polynomials $P(x,y)$ of degree $\leq n$ in $x,y$ respectively.

How do I find its eigenvalues, eigenvectors and the Jordan Canonical matrix representation of $L$?

I guess the JCF of $L$ is a $(n+1)\times (n+1)$ matrix. But how does one convert that sum of partial derivative operators into such a matrix?

Thank you.

share|cite|improve this question
The dimension is wrong. The space of polynomials in $x$ and $y$ of total degree at most $n$ has more than $n+1$ dimensions; if the degree in each of $x$ and $y$ is at most $n$ (so that you can include monomials like $x^ny^n$) then it has even more (it would be an $(n+1)^2\times (n+1)^2$ matrix then). – Arturo Magidin Nov 20 '11 at 1:32
up vote 3 down vote accepted

There are two methods; one is to translate the information into matrix form and work with that. The other is to work with $L$ directly.

For the first method, the first step is to find a basis for your vector space. In this case, you can take the basis of all monomials of degree at most $n$.

Then apply $L$ to the vectors of the basis in order to find the coordinate matrix of $L$; then work with it.

For example, for $n=2$, a basis would be $\beta=\{1,x,x^2,y,xy, y^2\}$ (note the dimension is not $3$, as you conjectured, but in this case $6$; the dimension is equal to the number of monomials of degree at most $n$).

(Note: I'm interpreting your problem as being with polynomials of total degree at most $n$; if you meant that the degree in each of $x$ and $y$ is at most $n$, then for $n=2$ you would need to add $x^2y$, $xy^2$, and $x^2y^2$ to the basis, and include them in the computations below; the main idea is the same, though.)

Then you apply $L$ to each vector in the basis, and express the answer as a linear combination of the basis: $$\begin{align*} L(1) &= \frac{\partial}{\partial x}1 + \frac{\partial}{\partial y}1= 0 + 0\\ &=0\cdot 1 + 0 x + 0x^2 + 0y + 0xy + 0 y^2,\\ L(x) &= =\frac{\partial}{\partial x}x + \frac{\partial}{\partial y}0 = 1 + 0\\ &=1\cdot 1 + 0 x + 0x^2 + 0y + 0xy + 0 y^2,\\ L(x^2) &= \frac{\partial}{\partial x}x^2 + \frac{\partial}{\partial y}x^2 = 2x + 0\\ &= 0\cdot 1 +2x + 0x^2 + 0y + 0xy + 0y^2,\\ L(y) &= \frac{\partial}{\partial x}y + \frac{\partial}{\partial y}y = 0+1\\ &= 1\cdot 1 + 0x + 0x^2 + 0y + 0xy + 0y^2,\\ L(xy) &= \frac{\partial}{\partial x}xy + \frac{\partial}{\partial y}xy= y+x\\ &= 0\cdot 1 + 1 x + 0x^2 + 1y + 0xy + 0y^2,\\ L(y^2) &= \frac{\partial}{\partial x}y^2 + \frac{\partial}{\partial y}y^2 = 0 + 2y\\ &= 0\cdot 1 + 0x + 0x^2 + 2y + 0xy + 0y^2. \end{align*}$$ Now, relative to the basis, the coordinate matrix of $L$ is: $$\left(\begin{array}{cccccc} 0 & 1 & 0 & 1 & 0 & 0\\ 0 & 0 & 2 & 0 & 1 & 0\\ 0 & 0 & 0 & 0 & 0 & 0\\ 0 & 0 & 0 & 0 & 1 & 2\\ 0 & 0 & 0 & 0 & 0 & 0\\ 0 & 0 & 0 & 0 & 0 & 0 \end{array}\right)$$ Now that you have a matrix, it is an easy matter to find the characteristic polynomial (it is $t^6$), the eigenvalues (only $\lambda=0$); the eigenspace (there are three free variables, so the dimension of the eigenspace is $3$); the eigenvectors (spanned by $1$, $x-y$, and $x^2 - 2xy + y^2$), the the Jordan canonical form (check the dimension of the nullspaces of $L^2$ (it's dimension is $5$), and of $L^3$ (which is all of the space). This gives you the Jordan canonical form (three blocks total, largest block of size $3$, so it must be one $3\times 3$ block, one $2\times 2$ block, one $1\times 1$ block).

For the other method you would have to work directly with $L$ and polynomials Solve the equation $L(a+bx+cx^2+dy + exy+fy^2) = \alpha(a+bx+cx^2+dy+exy+fy^2)$ to find that the only solutions have $\alpha=0$ (giving you the eigenvalues), and finding the explicit solutions gives the eigenvectors. Note that $L^2$ is not the zero linear transformation, then that $L^3$ is; this gives all the information you need.

(You will find that in the general case something similar occurs: you cannot have any eigenvalues other than $0$, because $L$ always reduces the maximum degree of $p(x)$ by $1$; that you need $n$ applications of $L$ to get the zero linear transformation; and you should find a pretty simple formula for the dimension of the space (hence the size of the matrix) and a nice way of seeing the Jordan canonical form.)

share|cite|improve this answer
Very well explained, thanks! – glinka Nov 20 '11 at 8:55

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.