tSDRG

Tree tensor strong disorder renormalization group.

Installation

One may choose from one of these methods below

I. Docker container

One may build the Docker image with

docker build --no-cache --force-rm -t tsdrg .

and launch the container interactively

docker run --rm -it tsdrg

See the documentation here if you want to run the container in background, or to mount the system volume into container.

Note: The size of this image is about 22.8GB, mainly because of Intel oneAPI.

II. Singularity container

Currently, to keep the consistency, we use Docker image to generate Singularity container.

To build the singularity image file (.sif), it is required to have a Dockerhub account for hosting the Docker image. One could therefore build .sif extension with

singularity build tsdrg.sif docker://<username>/<imagename>

One can then interact with the image in several ways. For instance,

singularity shell tsdrg.sif

See here for more information about working Docker with Singularity.

Note: By default, Singularity bind mounts /home/$USER, /tmp, and $PWD automatically into your container at runtime. This is different from Docker.

III. Build from scratch

Install Intel oneAPI

Intel Base toolkit and HPC toolkit are required for compiling Uni10 later.
Install Uni10

Current dependency of this module uses Uni10 v.2.0.0, which has been attached in this repo, and one can unzip them by using
```
tar -xzvf uni10-2.0.0.tar.gz
```
One can therefore compile the source code by following the instruction here. It is recommended to use MKL and Intel compiler for compiling.

Set up the environment variables

export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/uni10/lib/
export C_INCLUDE_PATH=$C_INCLUDE_PATH:/usr/local/uni10/include/
export CPLUS_INCLUDE_PATH=$CPLUS_INCLUDE_PATH:/usr/local/uni10/include/

Running a script

The main script is put in Main/. One can compile it with Main/makefile by targeting to the source code, such as

make code=Heisenberg_spin1.cpp

Executable will be named as job.exe, one could pass the parameters as

./job.exe {L} {chi} {Pdis} {Jdis} {Dim} {BC} {seed1} {seed2}

where

L: system size
chi: bond dimension, number of kept states in the triangular tensor (projector)
Pdis: the distribution of random variable
Jdis: disorder strength of the coupling J
Dim: Dimerization constant
BC: boundary condition, PBC or OBC.
seed1: random seed number in order to repeat data
seed2: random seed number in order to repeat data

Analyzing the data

Data will be saved into Main/data in .csv extension, and with hierarchical folder structure to the input parameters. For instance,

- data/
    - PBC/
        - Jdis010/
            - Dim010/
                - L32_P10_m8_1/
                    - J_list.csv
                    - L32_P10_m8_1_corr2.csv
                    - ZL.csv
                    - w_loc.csv
                    - L32_P10_m8_1_corr1.csv
                    - L32_P10_m8_1_string.csv
                    - energy.csv

tanlin2013/tSDRG