I've been told recently that one can use the Ising model can find solutions to certain NP-hard problems, such as Clique, although it doesn't do so in polynomial time.
Googling gets a few Arxiv results but where might I find an introduction to this interesting relationship?
Finding the minimal energy configuration in the Ising model is NP-hard, and many problems can be directly reduced to it. For a list, including Karp's original 21 NP-hard problem see e.g. Ising formulations of many NP problems. The Ising model is also equivalent to unconstrained quadratic binary optimization which also can be used to model a large number of other NP-complete problems, see e.g. A unified modeling and solution framework for combinatorial optimization problems.
