Renormalization Group

Consider the ising model.

H=Ji=1N1sisi+1B~i=1NsiC~.

Let K=JkT, B=B~kT, and c=c~kT. Then, Z=αexp(βH=s1=±1s2=±1sN=±1exp(βH) βH=Ksisi+1Bsic.

Performing a sum over the even terms, doubling the spacing, gives the partial trace, we then want to remap them back to the original lattice spacing to understand the original system. Then, Z(N,K,B)={si}exp(iKsisi+1+B/2(si+si+1)). Keep odd spins si and even spins σi. Z=iexp(k(siσi+σisi+2)+Bσi+B2(si+si+2)). Now we perform the partial trace, σi=±1. So, Z={si}exp((k+B/2)(si+si+2)+B)+exp((kB/2)(si+si+2)B). Then, it must be the case that Z={si}exp(iKsisi+2+B2(si+si+2)+2c). Then, per term, exp((k+B/2)(si+si+1)+B)+exp((k+B2)(si+si+1)B)=exp(ksisi+1+B2(si+si+2)+2c). This must hold for si=±1,si+2=±2. This gives our first equation [1], for si=si+2=1, exp(2K+2B)+exp(2K)=exp(K+B+2c). This gives our second equation [2], for si=si+2=1, exp(2K)+exp(2K2B)=exp(KB+2c). This gives our third [3], for si=si+2=1, exp(B)+exp(B)=exp(K+2c). Solving for K=K(K,B), B=B(K,B), and c=c(K,B). Dividing [1] by [2]: 4exp(2B)=exp(2K)exp(2B)+exp(2K)exp(2K)exp(2B)+exp(2K)=exp(B)(exp(2K+B)+exp(2K+B))exp(B)(exp(2KB)+exp((2KB))=exp(2B)cosh(2K+B)cosh(2KB). The third equation gives, exp(2C)=exp(K)2coshB. From the first equation, exp(B)2cosh(2K+B)=exp(K+B+2C)=exp(2K+B)2coshB. From the second equation, exp(B)2cosh(2KB)=exp(K+B+2C)=exp(2KB)2coshB. So, exp(4K)=cosh(2K+B)cosh(2KB)cosh2B and exp(8C)=exp(4K)24cosh4(B)=16cosh(2K+8)cosh(2KB)cosh2(B). Since we have B(K,B), we have determined K(K,B) and thus C(K,B). Then, K=14lncosh(2K+B)cosh(2KB)cosh2B. Then, B=B+12lncosh(2KB). Then, C=18ln16cosh(2K+B)cosh(2KB)cosh2(B). For B=0, K=12lncosh(2K) So, Z(N,K,B)=exp(NC(K,B))Z(N2,K,B). Then, f(K,B=0)=exp(NC(K,B))=2cosh(2K.

In the thermodynamic limit, FlnZ=Nζ(K) but ζ(K) does not depend on N. So, ζ(K)=12lnf(K)+12ζ(K). So, ζ(K)=2ζ(k)ln(2cosh(2K)).

Author: Christian Cunningham

Created: 2024-05-30 Thu 21:19

Validate