I'm a beginner in statistical physics and I'm reading some stuff about the Ising model. So this might be a silly question. My question is: why we study the Ising model for high dimension cases, despite that our physical world has only dimension $2$ or $3$?
-
Firstly, I don't think that this is a silly question at all, and secondly, I think that @DanYand 's comment contains the essence of the answer. – Nov 24 '18 at 17:20
1 Answers
Let us start with a quote from a paper by Michael Fisher and David Gaunt in 1964 (Phys. Rev. 133, A224), at a time when it was still necessary to justify such studies:
To elucidate the general problem of dependence on dimensionality and coordination number, it seemed worthwhile to investigate the Ising model and self-avoiding walks for lattices of dimensionality higher than three. [...] Of course the behavior of physical systems in four or more space-like dimensions is not directly relevant to comparison with experiment! We can hope, however, to gain theoretical insight into the general mechanism and nature of phase transitions.
As they say, it turns out that the spatial dimensionality (and more generally the connectivity properties of the underlying graph) plays a major role in the behavior of macroscopic systems. This is certainly the case at a critical point, where the critical exponents are well-known to depend generally on the spatial dimension, but can also be seen away from the critical point. As one example of the latter, consider the asymptotic behavior of the energy-energy correlations above the critical temperature: $$ \langle \epsilon_0\epsilon_{n\vec e_1} \rangle_\beta - \langle \epsilon_0\rangle_\beta \langle \epsilon_{n\vec e_1} \rangle_\beta \sim \begin{cases} n^{-2} e^{-2n/\xi} & (d=2)\\ n^{-2}(\log n)^{-2} e^{-2n/\xi} & (d=3)\\ n^{-(d-1)} e^{-2n/\xi} & (d\geq 4) \end{cases} $$ where, for any $k\in\mathbb{Z}^d$, $\epsilon_k = \sigma_k\sigma_{k+\vec{e}_1}$ and $\xi$ denotes the correlation length. As can be seen, the corrections to the exponential decay exhibit an interesting dependence on the dimension, which it is very natural for a physicist to try to understand.
- 10,371
-
1Actually this question came into my mind while I was reading your recent book on equilibrium statistical physics :) I'm a phd candidate doing theoretical computer science, especially the computational complexity of certain counting problems and the partition functions from statistical physics models. A fascinating phenomenon is the correspondence between phase transitions and the computational easiness/hardness of approximately computing the partition functions. Thank you and your coauthor for writing this wonderful book, which is a pleasure to read and is easy to follow. – Chao Liao Nov 26 '18 at 13:44
-
Dear @ChaoLiao , many thanks for your nice words! I am glad that you enjoy our book :) . – Yvan Velenik Nov 26 '18 at 15:31