OCP_screening

Codebase for screening catalysts using the Open Catalyst Project. Largely based off of the Open Catalyst Project tutorials available on Meta AI Research's fair-chem repository. Includes scripts for determining relevant bulk structures, cleaving surfaces, setting adsorption sites, configuring adslabs, relaxation, and post-processing.

Setup (UCSB CSC Pod Cluster)

The first thing we need to do is get you set up on the UCSB cluster. Here is the UCSB CNSI CSC's overview on HPC (High Perfomance Computing):

1. Request an account on the UCSB Center for Scientific Computing. It will take a couple of days to get it set up.

2. Referring to the document above, set up the client that suits your local operating system. You will access the campus cluster (Pod) via ssh, and each operating system (Windows, MacOS, Linux) has different ways of using ssh.

   • I use wsl on my Windows system (you can also use Putty) and the default terminal on my Mac.

3. You'll get an email that has your login information (with a temporary password). ssh in to your account by typing in:

   ssh [email protected]

   It will ask for your password. Just type in the temporary password in your email, and press Enter. This will log you into Pod's login node, which is where you set up your environments, scripts, and send your jobs to other nodes in the cluster. Then, follow the instructions on the email to change your temporary password to the one that you want!

   Review the information on CSC's guidlines to set up your file transfer system, and VPN if you plan on logging in to the cluster off-campus.

4. If you are logged in properly, you should see something like:

   ----------------------------
   
   Welcome to Pod
   For basic documentation to get started please see
   http://csc.cnsi.ucsb.edu/docs/pod-cluster

5. Now, let's get ready to install stuff! We will use anaconda to manage our packages and environments. Specifically, I like to use Miniconda, which is a minimal version of Anaconda. Follow the installation guides from the Miniconda link above. You can use the "Quick Command Line Install" part at the bottom of the page. The Pod cluster is a Linux system.

   mkdir -p ~/miniconda3
   wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh -O ~/miniconda3/miniconda.sh
   bash ~/miniconda3/miniconda.sh -b -u -p ~/miniconda3
   rm -rf ~/miniconda3/miniconda.sh

   It will take a while to install. Sit back and relax!

   After the installation is complete, remember to initialize the newly-installed Miniconda.

   ~/miniconda3/bin/conda init bash
   ~/miniconda3/bin/conda init zsh

6. If everything went okay, you should be able to see a 'miniconda3' folder in your home directory.
   This is a good time to go over basic linux commands.

   list files/folders	change directory	pring working directory
   ls	cd {directory_where_you_want_to_move_to}	pwd

   (The brackets {} are a notation for placeholders which you can replace with your relevant string. Don't include the brackets!)

   (FYI: the 'home directory' for you account can be accessed by cd ~. This is the same directory as /home/{your_user_name}/)

   CSC's HPC overview, as well as resources such as this cheat sheet will guide you through more essential Linux CLI(command line interface) commands.

7. Now, let's install some packages. An easy way to install (almost) everything you need to run ocp calculations is by following the guidelines here. However, the default environment uses cuda 11.8, which is not an available module on Pod. Therefore, we will set it up so that it uses cuda 12.1. I have set up my local environment as such, and the environment file is available on Pod as: fairchem_Pod.yml. Copy the .yml file into your home directory:

   cp /home/jaewon_lee/fairchem_Pod.yml ~

8. The .yml file acts like the seed of a tree. We can use it to reproduce the environment that I have on my system:

   conda env create -f fairchem_Pod.yml

   (If you run into problems, reach out to [email protected])

9. Now you should have the proper fair-chem environment on your login node! Let's check this by:

   conda activate fair-chem

   If you're in the vanilla bash settings (if you haven't changed anything about how the Linux CLI appears), you should see the following change:

   (base) [your_user_name@pod-login1 ~]$ conda activate fair-chem ⟶ (fair-chem) [your_user_name@pod-login1 ~]$

10. Congrats! You're all set to run ocp calculations!