Fast-randomized Iteration (FRI) Coupled Cluster

⚠️ WARNING: PROJECT IS UNDER CONSTRUCTION AND SUBJECT TO BREAKING CHANGES ⚠️

Installation

Requirements

• C++20 Compiler
• cmake
• Python >= 3.6
  • PySCF
  • pytest
  • matplotlib
  • h5py

Tested Configurations

Compiler	Success
g++10	✔️

From Source

git clone [email protected]:jamesETsmith/fri-cc.git --recursive
CXX=<your desired c++ compiler> cmake -B build
cmake --build build --parallel --target install
python -m pip install -e .

For cmake<3.14

git clone [email protected]:jamesETsmith/fri-cc.git --recursive
mkdir build && cd build
CXX=<your desired c++ compiler> cmake ..
make install -j
cd ..
python -m pip install -e .

⚠️ NOTE: fri-cc uses git submodules

Broadly speaking, the build/install process has the following steps:

1. Compile the shared library libfricc
2. Using pybind11, create Python wrappers for parts of libfricc called py_rccsd
3. Package and install the compiled Python modules.

Testing

There are two test suites, the first in C++, can be turned on by using an extra cmake flag -DCPP_TEST=ON:

mkdir build && cd build
CXX=<your desired c++ compiler> cmake .. -DCPP_TEST=ON
make install -j
make test

The second is a Python-based test suite, which you can run after following the instructions in the From Source section.

Benchmarking

Benchmarking is done from Python and scripts are contained in fricc/benchmark.