qldpc/exp_ldpc

QEC utilities for practical realizations of general qLDPC codes

This repository contains (WIP) research code related to practical implementation of general qLDPC codes.

Please look at the scripts in /scripts for usage instructions.

Caveats

There are a few corners of the codebase that are somewhat incomplete, like the Rust decoders, and the documentation is still somewhat sparse

Documentation

Generate documentation with pdoc

pdoc -o docs --html qldpc

Layout

This is organized as a Python library for generating quantum codes, generating syndrome extraction circuits, and setting up QEC experiments. In particular, there is a python package qldpc and some python scripts to do stuff with it in /scripts.

The python package will work without the rust library, but the only installation method is currently through PEP 517 + maturin instead of setuptools. There is an (incomplete) Rust library for decoding.

The circuits are output in the Cirq language and can be sampled using Stim (see /scripts).

Features

Available Codes Generators

• General Hypergraph / Homological Product Codes
• Expander Codes (Hypergraph product codes based on random bipartite graphs)
• Lifted product codes based on Tanner codes on Cayley graphs of PSL(2,q)
• Quasicyclic lifted product codes

Available Decoders (Currently WIP)

• Small Set Flip
• Belief Propagation

Circuit Generation Capabilities

• Syndrome extraction circuit assuming all to all gates based on edge coloring the Tanner graph

Installation

This project uses the python package Maturin for installation.

1. Install Rust ex. following https://rustup.rs/
2. Create a python virtual environment

python -m venv my_python_virtual_environment

3. Activate the virtual environment

source my_python_virtual_environment/bin/activate

4. Install the qldpc package into the virtual environment

git clone https://github.com/qldpc/exp_ldpc.git
cd exp_ldpc
pip install .

You can also build with maturin directly ex. maturin develop --release

Note: the virtual environment will need to be activated any time you want to use the package (step 3)

Alternate Installation (Singularity)

The repository contains a container definition that you can use to build a container with the qldpc package installed

git clone https://github.com/qldpc/exp_ldpc.git
sudo singularity build ldpc.sif exp_ldpc/ldpc.def

Alternate Installation (Nix)

A nix flake is provided: Use nix shell . to spawn a shell where the package is installed in the python interpreter or nix run . -- to start the interpreter directly. Use nix run .#experiment to spawn an interpreter with a few more useful packages. Alternatively, build your own python environment by supplying further python package overlays to the default package (see flake.nix).

Generating a code and syndrome extraction circuit

Try python /scripts/generate_code_circuit.py --help and python /scripts/generate_code_circuit.py 4 3 12 --rounds 1 --save_code code.txt --save_circuit circuit.txt to generate a (3,4) hypergraph product code on 225 qubits with 108 X and Z checks and 9 logicals. Constructing the logicals is a bit tricky and is currently done through Sage.

Code Format

The routines qldpc.write_check_generators and qldpc.read_check_generators read and write the code format specification giving out a scipy sparse matrix that is the X and Z check matrices. The check format is roughly inspired by DIMACS with a header denoting number of qubits, number of X checks, number of Z checks, and then listing the supports of each X and Z check