Installation¶

T3 was tested on Linux (Ubuntu 22.04 LTS and later).

Prerequisites¶

1. Install system packages¶

You need git, a C/C++ compiler, and either curl or wget.

Ubuntu / Debian:

bash sudo apt install git gcc g++ make curl
Fedora / Red Hat:

bash sudo dnf install git gcc gcc-c++ make curl
macOS (with Homebrew):

bash xcode-select --install

2. Install a conda package manager¶

We recommend Miniforge (which includes mamba for faster environment solving):

curl -L -O "https://github.com/conda-forge/miniforge/releases/latest/download/Miniforge3-$(uname)-$(uname -m).sh"
bash Miniforge3-$(uname)-$(uname -m).sh

Follow the prompts and restart your terminal when done.

Install T3 and dependencies¶

T3 requires four co-installed components:

Component	Conda env	Why T3 needs it
T3 itself	`t3_env`	the orchestrator
ARC	`t3_env`	QM job submission and post-processing
RMG-Py	`rmg_env`	mechanism generation and Arkane PDep SA — T3 spawns these as subprocesses
RMG-database	(filesystem only)	thermo / kinetics / transport data used by both RMG and ARC

Why two conda environments?

T3 itself never imports rmg-py; it only shells out to RMG and Arkane. Keeping rmg_env separate from t3_env lets the two have very different dependency trees without conflicting, and lets you upgrade T3 without rebuilding RMG. Both environments must exist for T3 to run end-to-end.

Two installation options are available.

Option 1: Automated installation¶

Clone T3 and install everything with one command:

git clone https://github.com/ReactionMechanismGenerator/T3.git
cd T3
make install

This clones ARC, RMG-Py, and RMG-database into the parent folder if they are not already there, creates both t3_env and rmg_env, compiles RMG-Py and ARC, and installs a custom updated version of PyRDL. It may take a while (RMG-Py's Cython compile is the slow step).

Option 2: Manual installation¶

Clone repositories¶

Clone all four repositories (e.g., under ~/Code/):

cd ~/Code
git clone https://github.com/ReactionMechanismGenerator/T3.git
git clone https://github.com/ReactionMechanismGenerator/ARC.git
git clone https://github.com/ReactionMechanismGenerator/RMG-Py.git
git clone https://github.com/ReactionMechanismGenerator/RMG-database.git

Create the T3 conda environment¶

cd ~/Code/T3
mamba env create -f environment.yml
conda activate t3_env

Install PyRDL (required by ARC)¶

cd ~/Code/ARC
bash devtools/install_pyrdl.sh t3_env

Compile ARC extensions¶

cd ~/Code/ARC
make compile

Create the RMG conda environment and compile RMG-Py¶

T3 spawns RMG and Arkane as subprocesses inside rmg_env, so this step is required even though T3 itself never imports rmg-py.

cd ~/Code/RMG-Py
mamba env create -f environment.yml
conda activate rmg_env
make
conda deactivate

Set environment variables¶

Add the following to your ~/.bashrc (or ~/.zshrc on macOS):

export PYTHONPATH=$PYTHONPATH:~/Code/T3:~/Code/ARC:~/Code/RMG-Py
export RMG_DB_PATH=~/Code/RMG-database

Then reload your shell:

source ~/.bashrc

Verify the installation¶

Activate the environment and run the test suite:

conda activate t3_env
cd ~/Code/T3
pytest tests/ -x -q

All tests should pass. If you encounter import errors, verify that PYTHONPATH and RMG_DB_PATH are set correctly.

Optional: convenient aliases¶

Add these to your ~/.bashrc (adjust paths accordingly):

export t3_path=$HOME'/Code/T3'
alias t3e='conda activate t3_env'
alias t3='python $t3_path/T3.py input.yml'
alias t3code='cd $t3_path'

After sourcing ~/.bashrc or restarting the terminal:

t3e activates the environment
t3code changes to the T3 repository directory
t3 runs T3 with input.yml in the current folder

Updating T3¶

Pull the latest changes from the main branch:

cd ~/Code/T3
git pull origin main

You may also want to update ARC, RMG-Py, and RMG-database:

cd ~/Code/ARC && git pull origin main
cd ~/Code/RMG-Py && git pull origin main
cd ~/Code/RMG-database && git pull origin main

Recompile ARC and RMG-Py with the latest changes:

cd ~/Code/ARC
make compile

cd ~/Code/RMG-Py
conda activate rmg_env
make
conda deactivate