How to diagonalize a hamiltonian matrix that has differential operators as elements? My matrix looks something like: \begin{bmatrix} A \frac{d^{2}}{{d\theta}^{2}}+ B_{1} & a\cos{(b\theta +c)}\\ a\cos{(b\theta +c)} & A \frac{d^{2}}{{d\theta}^{2}}+ B_{2} \end{bmatrix} where A, $B_{1}$, $B_{2}$, a,b and c are all constants. How to do I find the spectrum of eigen values for such systems. Are there any standard methods available.


Let your matrix operate on a vector of functions (of $\theta$, in this case), and look for solutions where it produces a constant multiple of that vector. In other words, solve

\begin{equation} \begin{bmatrix} A \dfrac{\mathrm d^{2}}{{\mathrm d\theta}^{2}} + B_{1} & & a\cos(b\theta +c) \\ a\cos{(b\theta +c)} && A \dfrac{\mathrm d^{2}}{{\mathrm d\theta}^{2}}+ B_{2} \end{bmatrix} \begin{bmatrix} f(\theta)\vphantom{\dfrac{\mathrm d^{2}}{{\mathrm d\theta}^{2}}} \\ g(\theta)\vphantom{\dfrac{\mathrm d^{2}}{{\mathrm d\theta}^{2}}} \end{bmatrix} =\lambda \begin{bmatrix} f(\theta)\vphantom{\dfrac{\mathrm d^{2}}{{\mathrm d\theta}^{2}}} \\ g(\theta)\vphantom{\dfrac{\mathrm d^{2}}{{\mathrm d\theta}^{2}}} \end{bmatrix} \tag{01}\label{eq01} \end{equation} which is a coupled set of second-order differential equations for $f(\theta)$ and $g(\theta)$. To help solve this system and find values of $\lambda$ which produce periodic solutions so that there is no discontinuity in $\theta$, read about Mathieu functions.

  • $\begingroup$ Could someone who knows LaTeX better than I do tell me how to make the matrix brackets and the vector brackets the same height? $\endgroup$ – G. Smith Nov 17 '18 at 4:07
  • $\begingroup$ How can I diagonalise such a matrix? I found something called Fulton-Gouterman transformation. $\endgroup$ – EverydayFoolish Nov 17 '18 at 5:05
  • $\begingroup$ Sorry, I don't know anything about that transformation. But if you can find the operator's eigenvectors and eigenvalues, you can diagonalize it. Unfortunately, I think I'm wrong about the Mathieu functions... I originally didn't notice that the $B$'s had subscripts, and when I took $B_1=B_2=B$ and decoupled the equations by looking at $f+g$ and $f-g$ I got Mathieu functions. But for the more general $B_1\ne B_2$ I get fouth-order PDEs that I don't know how to solve analytically. Are you trying to get an analytic solution or a numerical solution? $\endgroup$ – G. Smith Nov 17 '18 at 5:28
  • $\begingroup$ With Mathematica, I can numerically find the eigenvalues and eigenvector functions using the function NDEigensystem. For example, for $A=B2=B2=a=b=1, c=0$, I get that the first five eigenvalues are -0.2932, -0.2932, 0.405184, -1.28535, and -1.34277. $\endgroup$ – G. Smith Nov 17 '18 at 6:05
  • $\begingroup$ Any idea on how i could decouple them or resource where I can read about solving coupled differential equations $\endgroup$ – EverydayFoolish Nov 17 '18 at 14:43

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.