[main] [research] [srq]mg|pages

SRQ seminar – October 8th, 2018

Notes from a talk in the SRQ series.

INI Seminar 20181010 Toninelli (revised version 20181012)

Interacting dimer model

M matchings. ΠL,teMte the non-interacting measure has an infinite volume limit Πt which is determinantal. Under the limit measure Πt behaves like the massless Gaussian free field. It is true when t=(1,1,1,1) and in an open region of weights.

One can consider φCc and φ=0 and ε2xh(x)φ(εx)\xRightarrowdN(0,dxdyφ(x)φ(y)G(xy)) when t=(1,1,1,1).

Propagator of the non-interacting model

g(x,y)=[π,π]2dp(2π)2eip(xy)μ(p),

with μ(p)=t1+it2eip1+.

G(x)=12π2log|ϕ+(x)|

where ϕ+ is related to the gradient of μ at its two simple zeros p+ and p:

ϕ+(x)=(β+x1α+x2),β+=p2μ(p+),α+=p1μ(p+).

and ϕ+=ϕ. When t=(1,1,1,1) we have p+=(0,0) and p=(π,π).

Interacting model

ΠL,t,λ(M)(eMte)exp(λW(M))

where W(M) is a local function of dimer configuration summed over all translations.

Model 1

Model 2

W(M)=fall faces(I\scriptsize{f}+I\scriptsize{f})

W(M)=feven faces(I\scriptsize{f}+I\scriptsize{f})

(which is linked to the 6-vertex models with Δ=1eλ) where we say (arbitrarily) that even faces are those having black vertices on the upper left and lower right corners.

We ask λR being (very) small and W to be local.

Recall the definition of Kr depending on the type of edge:

Theorem 1. [AIHP '15] t=(1,1,1,1) and |λ|λ0 then

  1. ΠL,t,λΠλ

  2. If e is an edge of type r=1,2,3,4 and b(e)=x, and e is of type r and b(e)=x then

    Πλ(Ie;Ie)=A(λ)2π2Re[KrKrϕ+(xx)2]+
    +(1)x1+x2(x1+x2)Cr,r4π2B(λ)1|ϕ+(xx)|2ν+O(1||2+θ)

    as |xx|, where ν=ν(λ) and A,B are analytic functions of λ.

  3. The height field converges to a log correlated Gaussian field, for example we have, under Πλ,

    Var[h(f)h(f)(log|ff|)1/2]1π2A(λ),as |ff|,

    and in the same limit

    h(f)h(f)(Var())1/2N(0,1).

    Note that the oscillating term disappear in the variance.

Remark 2. In Model 1, we worked out ν,A which turns out to be

ν(λ)=14πλ+O(λ2),A(λ)=

and ν(λ) depends non–trivially on the weight t but recall that ν(0)=1 for any t.

Theorem 3. (J.Stat.Mech.) It holds (analog of Haldane's relation for Luttinger liquids)

ν(λ)=A(λ)

for any |λ|λ0. (Proven for t=(1,1,1,1), but work in progress for more general parameters).

Now we want to show how to represent the correlation functions and partition function as a Grassmann integral.

Kasteleyn theory

GZ2 bipartite, admitting perfect matchings like a box 2m×2n.

ZG,t=|detK|

where the matrix K has row indexed by black and colums indexed by white sites and

K(b,w)={0if bwKrtrif bw and the edgetype is r

On the torus one need to modify the method. One need to define other three matrices Kθ1,θ2 obtained by multiplying the edges which go out of the torus in direction e1 by (1)θ1 and in direction e2 by (1)θ2, then

ZG,t=12θ1,θ2=±1cθ1,θ2detKθ1,θ2

for certain coefficients cθ1,θ2.

From the formula for the partition function one can derive all the formulas for edge edge correlations.

I will pretend from now on that the partition function is given by a single determinant, namely

ZL,t=detK1,1

because K1,1 is invertible. Note that K0,0=0. These matrices are translation invariant so the can be diagonalized in Fourier basis:

fp(wx)=1Leipx

for pD={(p1,p2):pi=2πL(ni+12),0niL1} and then

wK(b,w)fp(w)=fp(b)μ(p).

Therefore from this is easy to deduce that the free energy is

F(t)=limL1L2logZL,t=dp(2π)2logμ(p)

and

K1(wx,by)=1L2pDeip(xy)μ(p)=g(x,y).

We assign Grassmann variables {ψx+,ψx}xTL where ψx+ is on black vertex bx and ψx on the white vertex wx. Define the integral of a non–commuting polynomial of the {ψx+,ψx}xTL

DψxΛψxψx+=1,

and the integral change sign when we exchange two variables (so variables cannot appear more than linearly), moreover when any variable is missing the integral is zero.

For example

Dψeψ1ψ2=Dψ(1+ψ1ψ2).

A few consequences:

  1. we can rewrite the determinant of a matrix as a Grassmann integral. If A is an n×n matrix, then

    detA=Dψex,yAx,yψx+ψy.

    Which recall the Gaussian integral

    (detΣ)1/2=dx1dxn(2π)n/2e(x,Σ1x)/2.
  2. Moreover

    A1(x,y)=Dψex,yAx,yψx+ψyψxψy+Dψex,yAx,yψx+ψy

    again similar to the Gaussian formula for the covariance.

  3. Fermionic Wick's rule. Let us denote (with A1=G)

    EG(f)=Dψex,yGx,y1ψx+ψyf(ψ)Dψex,yGx,y1ψx+ψy

    then

    EG(ψx1ψy1+ψxnψxn+)=det[Gn(x,y)]

    where we introduce the n×n matrix Gn(x,y) as Gn(x,y)i,j=G(xi,yj) for x=(x1,,xn) and y=(y1,,yn) and i,j=1,,n.

For the dimer model we obtain

Z=detK1,1=DψeS(ψ),S(ψ)=x,yK1,1(x,y)ψx+ψy.

The partition function of the interacting model can be also expressed as a fermionic integral. Define

eWΛ(A)=Mw(M)eedgesAeIeM

where AeR. This is equivalent to the replacement teteeAe therefore we consider

eWΛ(A)=Dψex,yeA(bx,wy)K(bx,wy)ψx+ψy=DψeSA(ψ)

Proposition 4. When W(M)=feven faces(I\scriptsize{f}+I\scriptsize{f}) then

ZΛ,λ=Mw(M)e\scriptsize{λfeven faces(I\scriptsize{f}+I\scriptsize{f})}=DψeS(ψ)+αγeγEe

where γ=f,f and Ee=K(e)ψx+ψx if e=(bx,wy) and α=eλ1λ.

If we are interested in the generating function of the interacting model we replace S(ψ)SA(ψ) and EeEeeAe.