# Open Data

## Current: /2211.11480

These datasets were used to produce all figures in the linked arXiv paper. If you use them, please cite the paper.

We use the most popular data storage format hdf5 due to possibility of including metadata (unlike numpy), processing the data without loading the entire dataset into memory with a single command, and other features that make the workings with data easier.

Unless noted otherwise, datasets contain time evolutions for XXZ model with ZZ interaction strength Delta=1 and open boundary conditions. All arrays have time as the first column.

Example readout of a dataset is presented in the notebook **example_dataset_readout.ipynb**. It requires installation of some *.h5 processing library. In the example we use python and tables. For a quick glimpse what the datasets contain, on linux use vitables to explore the files and arrays therein.

** /datasets/ **

** ChebyshevPP_XXZ_OBC_L14_L16_W1-5.h5 ** Time evolutions of correlation (arrays "Spin") for system sizes L=14, L=16 and time 0...1e4 with timestep 1. 1000 random uniform [-W, W] disorder realizations, with W = [1, 2, 3, 4, 5]. Calculated using Chebyshev polynomial expansion - can be considered exact.

** ChebyshevPP_XXZ_OBC_L26_L28_L14_W1.h5 ** Time evolutions of correlation (arrays "Spin") for system sizes L=14, L=26, L=28 and time 0...50 with timestep 0.2. 1000 random uniform [-W, W] disorder realizations with W = 1. Calculated using Chebyshev polynomial expansion - can be considered exact.

** Exact_AndersonXXZ_OBC.h5 ** Time evolutions of correlation (arrays "Spin") for system sizes L=50, L=100 and time 0...1e12 with changing timestep. 1000 random uniform [-W, W] disorder realizations with W = 3. Calculated using mapping to noninteracting fermions model and exact diagonalization.

** TDVP_XXZ_OBC.h5 ** Time evolutions of correlation (array "Spin"), von Neuman bipartite entanglement entropy: Configuration entropy part (array "EntConf") and the Number entropy part (array "EntNum"); Energy (array "en" - obviously it should be conserved at all time). We consider system sizes L=[50, 80, 100, 150, 200, 300] and final times up to 5000. Timestep = 1. Different numbers (32 or 1000) of random uniform [-W, W] disorder realizations with W=[8, 10]. Some evolutions have been obtained for many bond dimensions chi=[50,..., 192]. The bond dimension for each dataset is given as the last number in the hdf5 root catalog name, for example UW10L300C192 corresponds to U - uniform disorder, W=10, L=300, chi=192. Calculated using TDVP tensor network algorithm implemented on top of ITensor C++ library.

** /extras/ **

Datasets not used in the paper but some people may find them useful. Here, disorder comes from quasiperiodic distribution W_i = W cos(2 \pi \zeta i + \phi) where i is the index of the lattice site, \zeta = (\sqrt(5) - 1)/2 is the golden ratio and \phi is random variable uniformly distributed in the interval [0, 2\pi].

** TDVP_quasiperiodic_OBC.h5 ** Same as TDVP_XXZ_OBC.h5 above with L=[50, 100, 200] and bond dimensions chi=[128, 192]

** ChebyshevPP_XXZ_quasiperiodic_OBC.h5 ** Same as other Chebyshev* datasets, with L=[14, 16] and W=[0.5, 1, 1.5, 2, 3, 4, 5].

** /figs_PDFs/ ** Contains figures from the paper in high-quality PDF format.

[Parent Directory]