Potts Model in statistical physics

Click For Summary
The discussion focuses on the 3-state Potts model in statistical physics, detailing its Hamiltonian and the calculations needed for homework problems. The original Hamiltonian encourages neighboring spins to align, while the simplified Hamiltonian involves an external field. Participants are tasked with computing the average of the difference between the two Hamiltonians and applying variational mean field theory to derive a mean field equation. The average calculations involve understanding the relationship between spin states and their probabilities, leading to a specific mean field equation. Clarifications are sought on the derivation of the average formula, indicating a need for deeper comprehension of the underlying statistical mechanics.
CAF123
Gold Member
Messages
2,918
Reaction score
87

Homework Statement


The 3 state Potts model is defined by $$-\beta \mathcal H = J \sum_{r,r'} (3 \delta_{\sigma(r), \sigma(r')} - 1) + h\sum_r \delta_{\sigma(r),1},$$ with J > 0 to encourage neighbouring Potts spins to have same value and h orienting field. The spin like variables can take values 1,2 or 3. The spins are on a d dim hypercubic lattice so that each spin has z=2d nearest neighbours.

Consider the easier hamiltonian $$-\beta \mathcal H_o = H \sum_r \delta_{\sigma(r),1}$$ where H is an external field.

a) Compute the quantity ##\langle -\beta(\mathcal H- \mathcal H_o)\rangle## wrt easy hamiltonian. Use the fact that, wrt the easy hamiltonian, $$\langle \delta_{\sigma, \sigma'} = \langle \delta_{\sigma,1}\rangle^2 + \langle \delta_{\sigma,2}\rangle^2 + \langle \delta_{\sigma,3}\rangle^2 = \langle \delta_{\sigma,1}\rangle^2 + \frac{(1-\langle \delta_{\sigma,1}\rangle)^2}{2}$$

b) Use variational mean field theory to find the best lower bound for the original partition function using the easy hamiltonian above. Show that the resulting mean field equation is $$m = \frac{e^{h+3Jzm} - 1}{e^{h+3Jzm} + 2}$$

Homework Equations


All in section 1, and ##\sum_{\sigma} e^{-\beta \mathcal H_o}## is partition function associated with system governed by easy hamiltonian

The Attempt at a Solution


I found the partition function associated with the easy hamiltonian and the object for part a) is: $$\langle (h-H) \sum_r \delta_{\sigma(r),1} + J \sum_{\langle r,r'\rangle} (3 \delta_{\sigma(r), \sigma(r')} - 1) \rangle_{0,H}$$ There are N nodes with a spin and at each node the average of ##\delta_{\sigma(r),1}## is (1+0+0)/3 = 1/3, so for N spins, the first term average is N/3. For the second term, I get a 2dN multiplied by average of ##\langle \delta_{\sigma(r),\sigma(r')}\rangle##. I can use the suggested formula and I get this to also evaluate to 1/3.

I am just wondering how this equation they gave comes about. I can write $$\langle \delta_{\sigma,\sigma'}\rangle = \sum_{\sigma, \sigma'} \delta_{\sigma, \sigma'}e^{-\beta \mathcal H_o}$$ which is written like $$\langle \delta_{1,\sigma'}\rangle \langle \delta_{\sigma,1}\rangle + \dots$$ I guess?

Thanks!
 
Physics news on Phys.org
Can anyone help me understand the formula in a)? part b) should be fine thereafter. Thanks
 

Similar threads

Replies
5
Views
3K
  • · Replies 0 ·
Replies
0
Views
2K
  • · Replies 7 ·
Replies
7
Views
2K
Replies
3
Views
2K
Replies
7
Views
2K
  • · Replies 8 ·
Replies
8
Views
2K
  • · Replies 6 ·
Replies
6
Views
2K
  • · Replies 12 ·
Replies
12
Views
3K
  • · Replies 4 ·
Replies
4
Views
3K
Replies
1
Views
2K