Consider a 1d Ising model with no external magnetic field $(h=0)$ and adopt a decimation transformation in which every other spin is traced out.

So the Hamiltonian $H$ is given by $$H = -J\sum_{(i,j)} s_i s_j$$ where $(i,j)$ corresponds to the nearest neighbors of $i$.

I am trying to sketch the RG flow of this system, and I am struggling. How do you start a problem like this? I know you coarse grain a system by some spin-block tranformation $\tau$, but I am unsure of how to use it.

The end goal is to derive the RG equation, so the map $K' = R_l[K]$ for this transformation and find the fixed points in the flow.

Any advice would be appreciated!


1 Answer 1


Take the partition function $$ Z = \sum_{\{ S_{j} \}} \exp\left( \beta J \sum_{( i,j )} s_i s_j \right) = \sum_{\{ s_{j} \}} \prod_{(i,j)} \exp\left( \beta J s_i s_j \right) \ , $$ and sum over every other spin (so explicitly evaluate $\{s_{2n+1}\} \in \pm 1$ in the sum, but don't sum over $s_{2n}$).

This should be equal to a similar partition function over the coarse-grained lattice (with a new coupling $J'$ and new spin variables $s_i'$) $$ Z = \sum_{\{ S_{j} \}} \exp\left( \beta J' \sum_{( i,j )} s'_i s'_j \right) = \sum_{\{ s_{j} \}} \prod_{(i,j)} \exp\left( \beta J' s'_i s'_j \right) \ . $$ By comparing term-by-term, you can relate $J$ to $J'$.

  • $\begingroup$ Thank you for your answer @QuantumEyeda! Why did you elect to sum over every other spin? $\endgroup$
    – megamence
    Commented Nov 23, 2020 at 22:14
  • 2
    $\begingroup$ @megamence Summing over every other spin has precisely the same meaning as "every other spin is traced out". $\endgroup$ Commented Nov 23, 2020 at 23:01

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.