BosonicFlow-GKP

This repository contains a five-part PennyLane demo series on GKP-based logical quantum error correction in photonic systems.

Abstract

This five-part demo series introduces GKP-based quantum error correction in photonic systems from a software perspective using PennyLane. Unlike ordinary qubit-level error correction, where syndrome extraction and recovery act directly on qubits, the GKP framing starts from bosonic oscillator noise and studies how correction appears at the logical-circuit layer as reduced effective logical noise. The series progresses from single-qubit logical coherence under effective noise to channel comparisons, multi-qubit Bell/GHZ behavior, and an interactive playground.

Quickstart

Install project dependencies from the repository root (notebooks + desktop app):

• pennylane and numpy for simulation
• matplotlib for plots
• ipywidgets for interactive controls in S05
• PySide6 for the desktop app

pip install -r requirements.txt

Where this sits in the photonic stack

Photonic quantum computing spans multiple layers, from continuous-variable physics to high-level quantum algorithms. This project intentionally operates between hardware physics and algorithms, at the logical-qubit abstraction layer.

Physical photonic hardware
  └── Continuous-variable states
      (squeezing, phase space, homodyne detection)
        └── GKP encoding and correction
            (bosonic, non-Gaussian processes)
              └── Logical qubit with effective noise   <- this repo
                  └── Quantum algorithms

Series roadmap (demos/)

S01 - Logical noise fundamentals

Equation: $\rho \rightarrow \mathcal{E}(\rho)$ with a depolarizing channel, measured via $\langle X \rangle$.

Notebook: demos/gkp_qec_demo_01.ipynb

S02 - Error correction as noise suppression

Equation: $p_{\text{corrected}} = \alpha , p_{\text{raw}}$, with a toy mapping $p_{\text{raw}}(\sigma) = 1 - \exp(-(\sigma/\sigma_0)^2)$.

Notebook: demos/gkp_qec_demo_02.ipynb

S03 - Logical noise model exploration

Equation example: $\mathcal{E}_{\text{bitflip}}(\rho) = (1-p),\rho + p,X\rho X$ alongside other channels.

Notebook: demos/gkp_qec_demo_03.ipynb

S04 - Multi-qubit logical systems

Equation focus: Bell and GHZ correlations such as $\langle X \otimes X \rangle$ and $\langle X \otimes X \otimes X \rangle$.

Notebook: demos/gkp_qec_demo_04.ipynb

S05 - Interactive logical noise playground

Equation focus: $p_{\text{corrected}} = \alpha , p_{\text{raw}}$ and the displacement-to-logical mapping above.

Notebook: demos/gkp_qec_demo_05.ipynb

Note: the notebook uses direct numeric inputs (no sliders). For a live GUI, use the desktop app below.

Desktop app (offline playground)

The desktop app mirrors S05 with live plots, data export, and multiple noise models.

Build scripts (offline desktop):

• macOS build script
• Windows build script (PowerShell)
• Linux build script

Build instructions: pyqt_gui/README.md

CI workflow: .github/workflows/build-desktop-app.yml

Implementation notes

• Simulations use PennyLane default.mixed devices for noisy evolution.
• Noise channels are inserted explicitly in the circuits.
• Each demo is self-contained and follows a continuous narrative.

What BosonicFlow is

• A conceptual framework for photonic logical noise and correction
• Focused on logical qubits, logical operations, and logical noise models
• Architecture-aware, with a photonic-first interpretation
• Designed to support education, documentation, and onboarding
• Structured as a progressive demo series (S01 to S05)

What BosonicFlow is not

• Not a new quantum software framework
• Not a continuous-variable (CV) gate tutorial
• Not a replacement for PennyLane or Strawberry Fields
• Not a fault-tolerance benchmark or hardware performance study
• Not a physical simulation of GKP state preparation or decoding