ldpc

library2.4.1pypypi✓ verified 80d ago

A Python library for constructing, encoding, and decoding Low Density Parity Check (LDPC) codes. Current version 2.4.1, active development with periodic releases. Supports a variety of constructions (e.g., random, protograph, QCLDPC) and decoders (belief propagation, min-sum).

data ai-ml

pip install ldpc

package ↗docs ↗

INSTALL

IMPORT

SIG · LDPC

ldpc

datapythonv2.4.1

Install

15.4s avg

Import

—

Disk

398MB

Pass rate

4/ 10

Env Coverage4 / 10

glibc

3.9–3.13

musl

3.9–3.13

Install & Compatibility

Where this runs

tested against v2.4.1 · pip install

no network on importno background threads

Install × environment matrix

Each cell = how many times install + import succeeded across repeated harness runs. Partial = flaky.
glibc = Debian/Ubuntu slim · musl = Alpine Linux

musl

glibc

py 3.10

✕ build_error

✓ 15.3s

py 3.11

✕ build_error

✓ 14.7s

py 3.12

✕ build_error

✓ 15.6s

py 3.13

✕ build_error

✓ 15.9s

py 3.9

✕ build_error

398MB installed

● package 398MB

Code

Verified usage

Verified import paths — ran on the pinned version, not inferred.

BeliefFindDecoder

✓ from ldpc import BeliefFindDecoder

✗ from ldpc import LdpcCode

BpDecoder

✓ from ldpc import BpDecoder

✗ from ldpc import RandomLdpcCode

UnionFindDecoder

✓ from ldpc import UnionFindDecoder

✗ from ldpc import ProtographLdpcCode

Basic example: create a random LDPC code, encode, add noise, and decode.

import numpy as np from ldpc import RandomLdpcCode, LdpcEncoder, LdpcDecoder # Create a random LDPC code with rate 0.5, block length 100 code = RandomLdpcCode(block_len=100, rate=0.5) # Encode: generate random message msg = np.random.randint(0, 2, code.k) codeword = code.encode(msg) # Introduce noise (BSC) noise = np.random.binomial(1, 0.1, codeword.shape) received = (codeword + noise) % 2 # Decode decoder = code.decoder() decoded = decoder.decode(received, max_iter=50) print('Decoded correctly:', np.array_equal(decoded, codeword))

Debug

Known issues

breakingIn version 2.0+, the internal package structure was refactored. Custom parity-check matrices (H) must be passed as numpy arrays or scipy sparse matrices, not as lists.

fix

Ensure H is a 2D numpy array. Use np.array(H_list) if you have a list of lists.

affects: >=2.0.0

gotchaThe decoder's decode() method expects a 1D array of received symbols (0/1 for BSC, or soft bits for AWGN). For soft decoding, set decoder_type='bp' and provide log-likelihood ratios.

fix

For BSC: received = (codeword + noise) % 2. For AWGN: received = -2*noisey_llr + 1 (or similar).

affects: all

deprecatedThe function `ldpc.decode` (top-level) is deprecated; use `LdpcDecoder.decode` instead.

fix

Replace `ldpc.decode(...)` with `decoder.decode(...)` where decoder = code.decoder().

affects: >=2.0.0

Upgrade

Version history

2.4.1latest on PyPI · released Dec 8, 2025

Audit

Dependencies

numpyrequiredCore dependency for array operations.

scipyrequiredUsed for sparse matrix operations and linear algebra.

numbaoptionalUsed for JIT compilation of decoding loops (optional but recommended for performance).

Agent activity

4 hits · last 30 days

node

Resources

packagehttps://pypi.org/project/ldpc/ ↗

docshttps://software.roffe.eu/ldpc ↗