PyHDC 2.2.1

A Python library for Hyperdimensional Computing

PyHDC

A Python library for Hyperdimensional Computing (HDC) and Vector Symbolic Architectures (VSA).

Full documentation: https://pyhdc.readthedocs.io/en/latest/

---

Overview

PyHDC provides a unified interface for working with high-dimensional binary and continuous vectors used in HDC/VSA-based computing. It supports multiple encoding schemes, pluggable pseudorandom generators, and both NumPy and PyTorch backends.

Installation

pip install PyHDC

Quick Start

import pyhdc as hdc

# Create an encoding
enc = hdc.MAP_C(dimension=10_000)

# Generate hypervectors
v1 = enc.generate()
v2 = enc.generate()

# Core operations
bundled = v1.bundle(v2)        # superposition
bound   = v1.bind(v2)          # association
sim     = v1.similarity(v2)    # similarity score

Features

• 14 encoding schemes: MAP-C, MAP-I, MAP-I Bits, MAP-B, HRR, HRR (no norm), HRR (const norm), FHRR, VTB, MBAT, BSC, BSDC-CDT, BSDC-S, BSDC-SEG
• 7 generator families: LCG, Fibonacci/Galois LFSR, DLFSR, LCA, PCG, Xorshift, Shifted Counter
• NumPy and PyTorch backends with optional GPU support
• Composable operations: bind, bundle, unbind, similarity, thinning
• Custom encodings: extend Encoding with your own EncodingSpec
• Type-annotated throughout

License

MIT LICENSE.