Skip to main content

Holographic Reservoir Computing Framework — pure-Python simulation with optional ASIC hardware backend.

Project description

Holographic Reservoir Computing Framework (Simulation + Optional ASIC Hardware)

PyPI License: Apache-2.0 Python

A pure-Python reservoir computing framework built on a bipartite expander graph (the "Veselov layer") with a pluggable hardware backend. The default backend is a deterministic SHA-256 based simulation — no network, no ASIC, no external services. Optional AxeOS ASIC hardware is supported but strictly opt-in.

Install

pip install holographic-reservoir

Optional hardware extras:

pip install "holographic-reservoir[hardware]"

Quickstart (three commands, no hardware)

holographic-reservoir info
holographic-reservoir demo
holographic-reservoir benchmark --size 200 --n 1500

Programmatic use:

from holographic_reservoir import HolographicReservoirModel, mackey_glass

series = mackey_glass(n=1501, rng_seed=0)
u, y = series[:-1].reshape(-1, 1), series[1:]

model = HolographicReservoirModel(size=200, random_seed=0)
model.fit(u[:1000], y[:1000], washout=100)
mse = ((model.predict(u[1000:]) - y[1000:]) ** 2).mean()
print(f"MSE = {mse:.3e}")

Honest scope

  • Hardware backend is optional. The default is a pure-Python simulation using deterministic SHA-256 entropy; it runs anywhere Python 3.10+ runs and needs only NumPy. The AxeOS backend is activated only when you pass --hardware axeos --endpoint IP:PORT and connects to a real miner only in that case.
  • The paper in docs/papers/ is a research preprint, NOT a peer-reviewed IEEE publication. Treat it as an author preprint / technical report. No peer-review status should be inferred from its formatting.
  • Research artefacts (legacy experiment scripts, HTML drafts, hardware telemetry reports) are preserved in archive/ and experiments/ for transparency and are not part of the v1.0 API.

Architecture

holographic_reservoir/
├── backends.py   # HardwareBackend, SimulatedASICBackend, AxeOSBackend
├── model.py      # HolographicReservoirModel + Mackey-Glass generator
├── cli.py        # demo / benchmark / info subcommands
└── core/         # Legacy VeselovLayer expander graph (backcompat shim)

CLI

Command Purpose
holographic-reservoir info Show version and backend availability.
holographic-reservoir demo Offline Mackey-Glass demo, prints MSE.
holographic-reservoir benchmark Mackey-Glass MSE benchmark.

Flags (all subcommands): --hardware {simulation,axeos}, --endpoint IP:PORT, --seed, --size, --n.

Author and credits

Copyright 2026 Francisco Angulo de Lafuente (agnuxo1@gmail.com). Released under the Apache License 2.0.

Contains legacy research code from the CHIMERA / HRC experiments by the same author. Co-authors and collaborators credited in individual research artefacts under docs/papers/ retain their attribution.

License

Apache-2.0. See LICENSE.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

holographic_reservoir-1.0.0.tar.gz (35.9 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

holographic_reservoir-1.0.0-py3-none-any.whl (42.6 kB view details)

Uploaded Python 3

File details

Details for the file holographic_reservoir-1.0.0.tar.gz.

File metadata

  • Download URL: holographic_reservoir-1.0.0.tar.gz
  • Upload date:
  • Size: 35.9 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.13.7

File hashes

Hashes for holographic_reservoir-1.0.0.tar.gz
Algorithm Hash digest
SHA256 0946f03525b4d72c2ac6f9b48137c48fc5feac507b5792ea5bd499153cfd3680
MD5 42d633b428955b747254336691af8c15
BLAKE2b-256 df3d2ed915e9b89b08678cdb65c40a95af325b8f8171cb0f638c8b4926caae01

See more details on using hashes here.

File details

Details for the file holographic_reservoir-1.0.0-py3-none-any.whl.

File metadata

File hashes

Hashes for holographic_reservoir-1.0.0-py3-none-any.whl
Algorithm Hash digest
SHA256 fde3cb5c348e0ba91b0716ac0056d49506a67f93a32cb550281e2d91fe7f51c1
MD5 ebc683cf31a6df173b07a354472f366d
BLAKE2b-256 27d764b41e61b451f22c1acba0da58dc84228067cc4db7c1c68434fb975e2019

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page