I have read several times by now that the couplings in the SYK model are drawn randomly from a gaussian distribution. I was wondering what exactly is meant by that. To elaborate, when I compute an observable in the SYK model I start in some way from the Lagrangian. This already includes the couplings, so how do I know they have "previously" been determined randomly? What is the difference to just fixing the couplings all to be let's say 0.4, because I like that particular number and they just randomly turned out to all be 0.4?


The SYK model is always considered in the large-N limit (N = number of Majorana modes), and the probability that the distribution of couplings in a particular realization differs appreciably from the probability distribution you used to specify the model goes to zero as N -> infinity. This is the Law of Large Numbers at work.

Your question does make sense for other disordered models in which we don't take the large-N limit. Then we indeed have to distinguish between the disorder-averaged properties of the model, averaged over all disorder realizations with a given probability distribution, and the properties of a particular disorder realization which might be atypical.

  • $\begingroup$ So should I average the hypothetical observable I calculated over all realizations of SYK models with the appropriate weight for the given realization of the couplings? $\endgroup$ – QuantumStudent Dec 10 '18 at 14:32
  • $\begingroup$ @QuantumStudent Like I said, in the large N limit it won't matter. "Almost all" (i.e. all except a set of measure zero) realizations will give the same expectation value. $\endgroup$ – Dominic Else Dec 10 '18 at 14:34

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.