Introduction

ITensor is a c++ library for implementing tensor network calculations easily and efficiently. The home page of ITensor. It's easy to use ITensor in MacOS and Linux including the Windows subsystem for linux (WSL). This tutorial is mainly focus on the WSL alone with Ubuntu 18.04LTS which can be installed in the Microsoft Store. An uptodate Windows 10 is needed.

Prepare Ubuntu

1. Turn on wsl:

• Right click on PowerShell and select Run as Administrator.

• Paste the following code and hit return.

  Enable-WindowsOptionalFeature -Online -FeatureName Microsoft-Windows-Subsystem-Linux

• Reboot system.

2. Search Ubuntu in the Microsoft Store and install it.
3. Launch bash by the new created shortcut and creat your user name and password.
4. Go to wsltty repository download and install the lastest release version of wsltty. Use wsltty instead of default terminal for a better experience. If no shortcut created, creat it by hand.

• Right click on the desktop, select New>Shortcut

• Paste following code into the input box then Next

  %LOCALAPPDATA%\wsltty\bin\mintty.exe --wsl -o Locale=C -o Charset=UTF-8 /bin/wslbridge -C~ -t /bin/bash

• Give this shortcut a name then Finish

5. Change default PPA in Ubuntu: replace all the context in /etc/apt/sources.list with lines generated @mirrors.zju.edu.cn. open wsltty, run the following commands step by step:

• sudo nano /etc/apt/sources.list
• delete all contents
• go to mirrors.zju.edu.cn, in 配置生成器 section, seletc Ubuntu>18.04 then generate configuration
• paste lines generated by mirrors.zju.edu.cn (right click to paste)
• CTRL+X to exit nano and type Y to save
• sudo apt-get update
• sudo apt-get install g++ make libopenblas-dev

Install ITensor

• Download the tar.gz type file at ITensor home page. If the download path is D:\Downloads.

• Move the downloaded file to your home directory

  cd ~/

  mv /mnt/d/Downloads/ITensor-*.tar.gz ~/

  tar -xvf ITensor-*.tar.gz

  mv ITensor-* ITensor

  cd ITensor

  path and file name can be autocompleted by hitting tab, the above commands includes * which is a random string, you should use tab to autocomplete it.

• Modify make option file

  cp options.mk.sample options.mk

  nano options.mk

  comment the following two line (add #)

  PLATFORM=macos
  BLAS_LAPACK_LIBFLAGS=-framework Accelerate
  

  uncomment three line (delete #)

  #PLATFORM=openblas
  #BLAS_LAPACK_LIBFLAGS=-lpthread -L/usr/local/opt/openblas/lib -lopenblas
  #BLAS_LAPACK_INCLUDEFLAGS=-I/usr/local/opt/openblas/include -fpermissive -DHAVE_LAPACK_CONFIG_H -DLAPACK_COMPLEX_STRUCTURE
  

  change

  OPTIMIZATIONS=-O2 -DNDEBUG -Wall
  

  to

  OPTIMIZATIONS=-O3 -DNDEBUG -Wall
  

  CTRL+X Y return to save the modified file.

• Compile ITensor

  make

  wait for compile process complete.

Learn&Use ITensor

Use

A simple DMRG code for a Ising chain can be found here. Download and unzip all three files (argv.txt, dmrgising.cc, and Makefile) to a place in Windows, e.g., D:\DMRG\

cd /mnt/d/DMRG

make

export OPENBLAS_NUM_THREADS=1 (set threads for openblas, 1 is most efficient in most cast for itensor)

./dmrgising argv.txt 100 0.6

./dmrgising is the program itself, argv.txt contains settings for dmrg, 100 is the chain length N, 0.6 is the the magnetic field strength hz.

When writting your own code in a file owncode.cc, always put Makefile with it in the same directory and change

APP=dmrgising

to

APP=owncode

then you can type

make

to compile your code into binary file and run it.

Learn

The folllowing tutorials should be learned step by step expect for the last one if you do not care about the implementations of ITensor.

• Basic C++ grammar: take a look at this tutorial.
• For a tensor fundamentals, see The ITensor Book.
• The ITensor Tutorials gives intuitive sense for beginners.
• Code Formulas is a great reference when you write your own code based on ITensor.
• The details of ITensor library can be found here, it's useful when you decide to take a look at the source code of ITensor.