If you have a system of difference equations like this:

$x_{n+1} = a_{11}x_n + a_{12}y_n$

$y_{n+1} = a_{21}x_n + a_{21}y_n$

And you know that there is only one unique eigenvalue (ie: $\lambda_1=\lambda_2=\lambda$), how do you show that the general solution is $x_n=C_1\lambda^n + C_2n\lambda^n$?

My work so far: I've combined the two equations into a single equation $x_{n+2} - (a_{11} + a_{22})x_{n+1} + (a_{22}a_{11} - a_{12}a_{21})x_n = 0$

I know that this type of difference equation typically has the form $C\lambda^n$, but I have no idea how to come up with or show the bit with $Cn\lambda^n$.


Let $A=\begin{bmatrix} a_{11} & a_{12} \\ a_{21} & a_{22} \end{bmatrix}$ have equal (algebraic) eigenvalues: $\lambda_1=\lambda_2=\lambda$. Then either $a_{12}=a_{21}=0$ (in which case the situation is very simple) or the Jordan normal form of $A$ is

$$J=\begin{bmatrix} \lambda & \lambda \\ 0 & \lambda \end{bmatrix}.$$

In the latter case, you can show that

$$J^n=\begin{bmatrix} \lambda^n & n \lambda^n \\ 0 & \lambda^n \end{bmatrix}$$

by induction. Your result follows from there, since $A^n=V J^n V^{-1}$, where $V$ is the generalized eigenvector matrix.

Alternately, you can just guess-and-check, again using induction.


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.