Installation

Requirements

• Python 3.11+

• NumPy ≥ 1.26

• ASE ≥ 3.22

• SciPy ≥ 1.13

• geomeTRIC ≥ 1.0.0

• NetworkX ≥ 3.0

Install from PyPI

pip install mlfsm

Install from Source

To access the most recent features and bug fixes, install from source:

git clone https://github.com/thegomeslab/ML-FSM.git
cd ML-FSM
pip install -e .

Optional: Calculator Dependencies

ML-FSM works with any ASE-compatible calculator including traditional QM and semi-empirical potential energy surfaces. The following potentials have been tested and are supported out of the box via examples/fsm_example.py:

Calculator	Install	Notes
AIMNet2	git clone https://github.com/isayevlab/AIMNet2.git cd AIMNET2 pip install -e .	AIMNet2 pre-trained model.
MACE	pip install mace-torch	Pre-trained models available via mace_off
FAIR UMA	pip install fairchem-core	Meta’s Universal Model for Atoms
xTB	mamba install xtb-python	GFN2-xTB semiempirical method
Q-Chem	See Q-Chem	File-based I/O; requires a Q-Chem license