I am trying to do some Monte Carlo simulations for Pfaffian state from Fractional Quantum Hall effect. I am wondering what is the energy functional for $\nu=5/2$ Moore-Read state?

  • 3
    $\begingroup$ What do you mean by the "energy functional"? The energy as a function of the fermionic fields? Also, how do you do Monte Carlo for a Pfaffian system? Genuinely curious, because Pfaffian fermions have a horror of a sign problem--- if you could elaborate, perhaps somebody would know the answer. $\endgroup$
    – Ron Maimon
    Aug 19, 2011 at 8:59

1 Answer 1


You know an explicit expression for the unnormalized Moore-Read wavefunction in (complex) position representation \begin{equation} \psi_{\mathbf{MR}}(\{z_i\})= \mathbf{Pf}\left( \frac1{z_i-z_j} \right) \prod_{i<j} (z_i-z_j)^2 e^{-\sum_k |z_k|^2/4}\quad . \end{equation} The energy of a wavefunction is given by the corresponding average value of the Hamiltonian. \begin{equation} E_\psi=\langle H \rangle = \mathcal{N} \int dz_1\dots dz_n \left|\psi_{\mathbf{MR}}(\{z_i\})\right|^2 H(\{z_i\}) \end{equation} with the normalization constant \begin{equation} \mathcal{N}= \left( \int dz_1\dots dz_n \left|\psi_{\mathbf{MR}}(\{z_i\})\right|^2 \right)^{-1} \quad . \end{equation} In your case I guess the Hamiltonian only consists in a Coulomb term, thus its matrix elements in position space are \begin{equation} H(\{z_i\})=\frac{e^2}{4\pi\epsilon} \sum_{i<j} \frac{1}{|z_i-z_j|} \end{equation} since $||r_i-r_j||=|z_i-z_j|$.

The two integrals involved are computationally costly, but you can benefit from the Metropolis sampling method (http://en.wikipedia.org/wiki/Metropolis%E2%80%93Hastings_algorithm), the most commonly used sampling method in Monte-Carlo algorithms I guess. It is very simple. In the latter you:

  • randomly choose an initial configuration $\{z_i\}$ (i.e. set of position). Its probability is $\pi_0=|\psi\{z_i\}|^2$.

  • generate a new configuration from the first one by randomly moving one of the particles. I note the new configuration probability $\pi_1$.

  • the new configuration is accepted with probability 1 if it has a higher probability than the first one, and with probability $\pi_1$ otherwise.

  • continue until the desired accuracy is reached or maximum number of attempt have be tried out.

Every accepted move contributes to the integral with weight 1/"number of sample configurations used". Explicitly if $I$ is the quantity I want to compute, $N$ the number of configurations $i$ I keep, and $f_i$ the value of the integrand for configuration $i$, \begin{equation} I=\frac1N\sum_{i=1}^N f_i \quad . \end{equation}

Good luck.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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