Explore the H-cat qubit encoding, in simulation.
The H-cat encoding (Lee et al., PRX Quantum 5, 030322, 2024) pairs a bosonic cat with a single photon. Both logical states carry an even cat, so the cat's photon-number parity works as a loss flag: one lost photon flips it. Two lost photons leave it even and cause a silent logical error. The animation below shows the one-photon case.
One H-cat qubit, animated
Mode A is a cat (amber); Mode B is a single photon in dual-rail (cyan). Inject a loss and watch.
Each encoding trades one hard problem for another.
- Needs large even cats (amplitude about 2.5 to 2.9); only small cats were simulated, at 3.5 to 8 dB
- Fusion is a destructive hybrid Bell measurement with every output detected
- One lost photon flips cat parity and is flagged
- Two lost photons leave parity even: a silent logical error
- Needs 10+ dB squeezing, hard to reach on chip
- Grid-state preparation is demanding
- Loss degrades the grid as a silent shift error
- Probabilistic fusion, so gates need boosting or multiplexing
- Heavy multiplexing to reach near-deterministic gates
- Loss removes the photon: detectable as a missing click, but the qubit is gone
These are component-level simulations. They do not show that a fault-tolerant H-cat machine is buildable: the encoding needs cats of amplitude about 2.5 to 2.9, far beyond the small cats simulated here.
Simulation tools, in the browser and in Python.
SDK / Run a Circuit
Build a gate-model circuit, run it on an illustrative simulator, and see counts plus model resource estimates.
Resource Analyzer
Explore an abstract circuit model: code distance and cats per logical versus loss. Model values with assumed loss heralds, not a hardware operating point.
Workbench
Open-source Python SDK: small-cat generation from squeezed vacuum, single-mode cat loss curves, and an illustrative resource model.
Fusion Designer
Sketch H-cat qubits and fusion links on a canvas (a drawing tool, no physics). Export and import designs as JSON.
Docs
The H-cat encoding, the component simulations, and how each tool works.
Education
Short lessons on cat qubits, squeezing, breeding, fusion, and fault tolerance.
Component simulations, and what they do not show.
Two independent codebases (QuTiP and Strawberry Fields) agree on small-cat fidelity to about 4e-5. A component simulation.
Surface-code runs with stim and PyMatching use an abstract circuit model with assumed loss heralds. Model values, not a hardware operating point.
Ring fabrication Monte Carlo (component simulation): about 99 percent resonance-spec yield with enough thermal tuner range.
Test chip layout QF-TC1: 5x5 mm, 22 structures, 0 violations against a foundry-neutral stand-in rule deck only. Not fabricated.
pip install dyber and pip install dyberforge both work today, Apache-2.0. SDK on GitHub.
Status: Dyber has no quantum hardware, and the hcat_qpu backend is not available. Circuits run on an illustrative local simulator. Resource numbers are model values from an abstract circuit model with assumed loss heralds, not a hardware operating point. Dyber, Inc.