I am interested in diagonalizing the all-to-all quantum spin model

\begin{align} \hat{H} = \frac{1}{2}\sum_{i,j \neq i} \hat{S}_i \cdot \hat{S}_j \end{align}

or, if possible, a more general form involving an arbitrary coupling $\tilde{H}$ between the spin states on sites $i$ and $j$:

\begin{align} \hat{H} = \frac{1}{2} \sum_{i, j\neq i} \tilde{H}_{ij} \end{align}

There are numerous treatments for how to approach this sort of problem using mean field techniques, but for the applications I'm interested in I would like to obtain the spectrum exactly. Is this a tractable problem? Is the answer known? Is it trivial?

I am imagining something along the lines of an adaptation of the Bethe Ansatz to this problem-- for instance, in the case of the first Hamiltonian when the couplings are instead restricted to be nearest-neighbor instead of all to all, the solution is well known.

  • $\begingroup$ Why spin “glass”? Your original Hamiltonian is antiferromagnetic… Don’t you rather mean a quantum analogue of the Sherrington-Kirpatrick model? $\endgroup$
    – LPZ
    Jan 29 at 22:19
  • $\begingroup$ @LPZ you're right, I've removed "glass" from the title. I had in mind the SK model when I wrote it but for my problem the fact that the couplings are fixed nonrandom is actually a core part so I would not expect glassy behavior. $\endgroup$
    – Panopticon
    Jan 29 at 22:29

1 Answer 1


Assuming that each spin has total spin $s$, then you can rewrite the Hamiltonian: $$ H = -\frac J2\hat S{}^2 +\frac {JN}2s(s+1)\\ \hat S = \sum_i \hat s_i $$ You therefore just need to diagonalize $\hat S{}^2$. For this, you'll need to decompose the product space of $N$ $s$ spin spaces into irreducible representations. The spectrum of $\hat S{}^2$ will therefore be of the form $S(S+1)$ with $S$ ranging the possible total spin number, namely $0, 1 ... Ns$. To diagonalise the Hamiltonian, you'll need to use Clebsch-Gordon coefficients.

Hope this helps.


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.