Supplementary material for the article "Domain wall fluctuations of the six-vertex model at the ice point"
by Michael Prähofer and Herbert Spohn
(2023), arxiv.org/abs/2305.09502

Domino tilings of the Aztec diamond can be mapped to the six-vertex model of (N{+}1)\times(N{+}1) tiles with domain wall boundary condition (DWBC). It can be further mapped configurationwise to an integer-valued height function \Phi:[0,N{+}2]^2\to\mathbb{Z}  which is constant on \mathbb{\R}^2\setminus(\mathbb{R}{\times}\mathbb{Z}\,\cup\,\mathbb{Z}{\times}\mathbb{R}) and has fixed values in the outside frame of tiles as  \Phi(x,y)=0 for  x\in[0,1)or y\in[0,1) and \Phi(x,y)=\lfloor x\rfloor for y\in(N+1,N+2], \Phi(x,y)=\lfloor y\rfloor for x\in(N+1,N+2]. On this page we determined for some N\in\mathbb{N} the probability (P_{N})_{k,l},that \Phi(k+\tfrac12,l+\tfrac12)=0, k,l=1,\dots,N, which sometimes is called the emptiness probability matrix.
For example P_1=\begin{pmatrix}\tfrac12\end{pmatrix}\in\R^{1\times1}P_2=\begin{pmatrix}\tfrac34&\tfrac14\\\tfrac14&0\end{pmatrix}\in\R^{2\times2}P_3=\frac1{16}\begin{pmatrix}14&8&2\\8&1&0\\2&0&0\end{pmatrix}\in\R^{3\times3}, P_4=\frac1{64}\begin{pmatrix}60&44&20&4\\44&13&1&0\\20&1&0&0\\4&0&0&0\end{pmatrix}\in\R^{4\times4}.
Note that due to symmetries of the model we have (P_N)_{1,k}+(P_N)_{1,N+1-k}=1=(P_N)_{k,1}+(P_N)_{N+1-k,1} for k=1,\dots,N.

In general, the total number of domino tilings of an Aztec diamond of generation N is 2^{N(N+1)/2}. Nevertheless d_N=2^{\lfloor(N{+}1)/2\rfloor\lceil(N{+}1)/2\rceil} is the smallest denominator, which renders  d_N P_Ninteger-valued. The following file contains in each line  \{N,\ d_N,\ d_NP_N\} for N=1,\dots,32, with the matrix d_NP_N\in\mathbb{N}_0^{N\times N} written in the format \{\{\dots,\dots\},\dots,\{\dots,\dots\}\}:

aztec_occupation_exact_1_32.txt (265K)

The exact values were obtained by numerically evaluating the corresponding infinite dimensional Fredholm determinants with Meixner kernel for the Aztec diamond to a high enough precision, to accurately guess the numerators in P_N.

The following files contain the results from Monte-Carlo simulations, the first lines starting with '#' and containing header information like system size and number of samples. Each data line contains the numbers k, l,(P_N)_{k,l} separated by spaces, with the probability (P_N)_{k,l}\in(0,1)
guessed from the Monte-Carlo results. Only entries of P_Nwhich are different from 0 and 1 are recorded.  

aztec_occupation_sparse_00004.txt
aztec_occupation_sparse_00008.txt
aztec_occupation_sparse_00016.txt
aztec_occupation_sparse_00032.txt
aztec_occupation_sparse_00064.txt
aztec_occupation_sparse_00128.txt
aztec_occupation_sparse_00256.txt
aztec_occupation_sparse_00512.txt (1.8M)
aztec_occupation_sparse_01024.txt (4.6M)
aztec_occupation_sparse_02048.txt (12M)
aztec_occupation_sparse_04096.txt (26M)
aztec_occupation_sparse_08192.txt (65M)
aztec_occupation_sparse_16384.txt (124M)

They have been produced with the help of the C program "ArcticCircle.c" and a primitive python script contained in

ArcticCircle.zip

Michael Prähofer, 17.04.2023

  • Keine Stichwörter