Ab Initio Modelling of High Temperature Exciton Superfluidity in 2D Heterostructures

Project for calculating parameters for a more realistic modelling of exciton superfluids in van der Waals Heterostructures.

Getting Started

Go to the main_results directory to reproduce the main results of the article. In this directory you will also find more documentation on the computational methods. In the text below, we cover how to install necessary dependencies and for convenience we also list useful commands.

Install dependencies

You need a virtual environment with GPAW to run calculations. If on NiflHeim, use the install_gpaw.sh. Other dependencies are listed in requirements.txt. Run the following command to install the python packages

pip install -r requirements.txt

The BiEx code package used to solve the gap equation is available from the authors upon reasonable request.

---

Usefull commands specific to the dtu gbar hpc environment

Load in intel compiler module

module add intel

Queuing tasks/jobs

Navigate to repo folder on gbar. Then you can run the following:

• Submit job on 8 cores for 4 hours

mq submit script.py -R 8:4h

• Submit job WSe2_bb.py after WSe2_gs.py has completed:

mq submit WSe2_bb.py -d WSe2_gs.py -R 8:4h

• List jobs

mq ls

• Remove job

mq rm -i JOBID

• Help

mq help

• Help on command (e.g. rm command)

mq rm -h

General commands (from startup guide)

• Login with Forward X11 (use dtu user password)

ssh -XY [email protected]

• Log onto a compute node

linuxsh -X

• Create virtual environment: Follow instructions here
• Activate virtual environemtn (you can add this to your .bashrc)

source ~/path-to-your-venv/bin/activate

• Launch interactive jupyter notebook (replace venv with name of your virtual environment)

python -m ipykernel install --user --name=venv
jupyter notebook --no-browser --port 40000 --ip=$HOSTNAME

Open new terminal locally and run (USERNAME: your DTU username, HOSTNAME: servername you see in the other window (n-XX- YY-ZZ), PORT: 40000)

ssh [email protected] -g -L8080:HOSTNAME:PORT -N

Go into your browser and open: http://localhost:8080. When you are creating a new note notebook on the web browser, go to “new” and then choose the name of your virtual environment to make it work. If you already have written the notebook, just go to “kernel” -> “change kernel” and choose your respective environment.

Git Commands

Note: You need to have cloned the repo and logged in with git to run the following commands. Checkout a beginner's guide to git, if you are unfamiliar with it.

• Check which files/folders have been modified.

git status

• View changes for a spceific file. (This command opens up a less editor/program. Type h for help and q to exit).

git diff [filename]

• Add modified file to "staging area" (i.e. tell git, that you intent to commit this modified file)

git add [filename]

• Add all modified files

git add *

• Commit added files

git commit -m "message"

• Push file to remote repository (upload changes)

git push

• Pull latest commits from remote repository (download latest changes)

git pull

VS Code Guides/Commands

Connecting to ssh in VSCode

Add the Remote-SSH-extension to VSCode and connect to the host (ssh -XY [email protected]) using their getting started guide.

Trouble shooting with Remote-SHH

If the remote ssh-connection suddenly fails (e.g. displays password prompt again and again), you can try the following:

1. Open the command pallet (CTRL + SHIFT + P or COMMAND + SHIFT + P (mac) ).
2. Type Remote-SSH: Kill VS Code Server on Host. Then try to connect again

Running Jupyter Notebooks in VSCode

This may only work on a mac (unix system) because the X11-forwarding seems to be unsupported for VSCode on Windows. On the hpc compute node, run the following command:

• (optional) Set a password for the Jupyter Notebook Server:

jupyter notebook password

• Launch a Jupyer server with the following command:

jupyter notebook --no-browser --port 40000 --ip=$HOSTNAME

This outputs a Jupyter server address (e.g. http://n-10-10-1:40000/). Connect to the Jupyter Server as described here (don't include a token). Type the password when prompted to do so.

• Open up a jupyter notebook file (ipynb-file). In the top right corner, choose the remote venv kernel.