Cross-Platform MOC

Curated map of Zoo entries in the Cross-Platform family.

Entries

Entry	Type	Status
quantum-gate	concept	demonstrated
bacon-shor-code	encoding	demonstrated
color-code-logical-qubit	encoding	demonstrated
erasure-qubit	encoding	demonstrated
floquet-codes	encoding	proposed
qldpc-codes	encoding	proposed
surface-code-logical-qubit	encoding	demonstrated
classical-control	infrastructure	demonstrated
quantum-transduction	infrastructure	demonstrated
nuclear-magnetic-resonance-qubit	qubit	demonstrated

Composition

• concept: 1
• encoding: 6
• infrastructure: 2
• qubit: 1

Conceptual anchors

• codes-moc is the authoritative map when the comparison is about check structure, measurement schedule, or code-family identity; this page takes over when those choices become hardware-wide overhead and systems questions.
• threshold-theorem is the main bridge from surface-code-logical-qubit and color-code-logical-qubit back to hardware performance targets.
• erasure-error-vs-pauli-error separates erasure-qubit from the ordinary stabilizer-code story and explains why flagged loss can change the overhead regime.
• noise-bias-and-asymmetric-error-channels is the complementary lens when the win comes from skewed Pauli channels rather than explicit erasure detection.
• divincenzo-criteria explains why classical-control and quantum-transduction belong here even though they are not qubits.
• classical-hardware-moc owns the measurement-side path from qubit pointer states through amplification and discrimination; this page takes over when that path closes a code-cycle feedback loop or when coherent quantum information must cross a module boundary.

Architecture stack

• quantum-gate is the operation layer: it stays abstract on purpose so the same gate idea can descend either into physical pulse stacks or into encoded logical actions.
• surface-code-logical-qubit and color-code-logical-qubit are the geometry-and-decoder layer: both answer the threshold problem, but with different overhead, layout, and gate-transversality tradeoffs.
• bacon-shor-code, floquet-codes, and qldpc-codes are the check-structure layer: they move complexity respectively into gauge redundancy, time-ordered measurements, or sparse nonlocal connectivity. Their code-level comparison belongs in codes-moc; their control, connectivity, and decoder consequences belong here.
• erasure-qubit is the noise-model layer: it does not replace a code family so much as change the decoder assumptions under which the other logical encodings should be judged.
• classical-control and quantum-transduction are the systems layer: one closes the real-time local feedback loop, the other opens the nonlocal modular-network loop.
• nuclear-magnetic-resonance-qubit is the historical transfer layer: its shaped pulses, refocusing sequences, and early algorithm demonstrations helped establish portable control abstractions, even though its ensemble processor architecture did not scale.

NMR as a portability test

nuclear-magnetic-resonance-qubit appears in this facet because it cleanly separates a platform’s transferable methods from its non-transferable scaling assumptions.

• The portable layer runs toward quantum-gate and classical-control: selective rotations, refocusing, composite pulses, and optimal-control methods survived the NMR processor model and became general hardware tools.
• The non-portable layer stays with molecular-moc: pseudo-pure ensemble preparation, bulk-magnetization readout, and exponentially shrinking signal are properties of the liquid-state architecture, not generic quantum-computing constraints.
• The routing rule is therefore historical and methodological. Use this page to ask what NMR exported to later platforms; use molecular-moc to judge NMR itself against divincenzo-criteria and scalable processor requirements.

This distinction prevents an early algorithm demonstration from being mistaken for a scalable systems architecture: gate sequences transferred, but the state-preparation and measurement model did not.

Family structure

• surface-code-logical-qubit and color-code-logical-qubit are logical overlays on top of many hardware platforms, so they should point outward to code and threshold concepts rather than read like standalone modalities.
• erasure-qubit is a noise-engineering pattern that can be instantiated in multiple families, not a single device recipe.
• classical-control and quantum-transduction are scaling interfaces: one connects algorithms to physical waveforms, the other connects local processors to networked photonic links.
• quantum-gate should stay as the operation-level abstraction that ties these layers together, not become a dumping ground for platform-specific pulse details.
• nuclear-magnetic-resonance-qubit is intentionally a secondary facet here and a primary member of molecular-moc; its cross-platform role is methodological inheritance, not present-day hardware competitiveness.

Local feedback versus coherent inter-module links

classical-control and quantum-transduction sit at different boundaries of a scalable machine. They should not be read as consecutive versions of the same communication problem.

Systems path	Information crossing the boundary	Primary success condition	Route elsewhere when…
Qubit measurement $\to$ decoder $\to$ corrective waveform	A classical record of a syndrome or measurement result	End-to-end latency, deterministic timing, and sustained throughput fit inside the code cycle	the bottleneck is producing or preserving the analog measurement record; use classical-hardware-moc
Local quantum mode $\to$ network-compatible optical mode	An unmeasured quantum state or entanglement resource	Conversion efficiency, added noise, bandwidth, and refrigerator heat load jointly preserve useful entanglement rate	the optical carrier and computation model are already chosen; continue in photonic-moc

The classification hinge is whether measurement is allowed. A decoder loop deliberately turns quantum information into classical bits before acting back on the processor; a transducer must preserve coherence precisely because measuring at the module boundary would destroy the distributed operation. Modular fault tolerance may require both loops, but their error budgets compose differently: control latency limits how quickly local syndromes can be used, while transduction loss and noise limit how often remote checks or entanglement links succeed. Keep both costs explicit before claiming that a code with nonlocal checks is hardware-ready.

Feedback-loop budget: do not hide the boundary in one latency number

A reported decoder latency is only one term in the physical correction loop. For routing purposes, decompose the end-to-end deadline as

$$T_{\mathrm{loop}}=T_{\mathrm{meas}}+T_{\mathrm{capture}}+T_{\mathrm{infer}}+T_{\mathrm{decode}}+T_{\mathrm{route}}+T_{\mathrm{actuate}},$$

with the warning that a real controller may pipeline or overlap some terms. The useful question is not whether one component is fast in isolation, but whether the sustained loop—including its jitter and worst-case tail—meets the syndrome-freshness or feed-forward deadline of the chosen protocol.

Budget term	What it includes	Primary graph route	Failure hidden by quoting only decoder time
$T_{\mathrm{meas}}$	Qubit-to-pointer mapping and the integration window needed to separate states	qubit-readout and, for superconducting dispersive measurement, dispersive-readout-mechanism	The digital pipeline waits on a slow or weak physical measurement
$T_{\mathrm{capture}}$	First-stage gain, digitization, channelization, and movement of samples into programmable logic	classical-hardware-moc; cryogenic-amplification when microwave SNR is limiting	A fast decoder receives late or low-quality bits because the analog front end is the bottleneck
$T_{\mathrm{infer}}+T_{\mathrm{decode}}$	State discrimination, syndrome assembly, and inference of a correction or frame update	classical-control plus the relevant code route in codes-moc	An accelerator benchmark omits acquisition, transfer, or protocol overhead
$T_{\mathrm{route}}$	Deterministic inter-board communication, scheduling, and alignment of the decision with the target cycle	classical-control	Mean latency looks acceptable while skew, congestion, or tail latency misses deterministic deadlines
$T_{\mathrm{actuate}}$	Waveform launch, signal-chain response, and completion of the conditional operation or frame update	classical-control and the destination platform MOC	“Loop closed” means a bit was produced, not that the processor received the intended physical action

This decomposition also separates deadline from throughput. A demonstration can fit one correction inside one cycle yet still fail to sustain repeated rounds across a larger code because acquisition bandwidth, decoder scaling, or inter-board traffic grows with check count. Use threshold-theorem only after the whole loop has been mapped onto the code’s repeated-cycle workload; a sub-microsecond component result is evidence about one layer, not by itself evidence of fault-tolerant operation.

Routing rule: when to enter this family

• Enter here when the main comparison is decoder assumptions, logical-overhead scaling, or systems integration rather than device physics.
• Stay in a source hardware family when the central claim is hardware-shaped noise, then cross over here only after the error model is clear. In practice that means reading kerr-cat-qubit, 0-pi-qubit, or dual-rail-superconducting-qubit through noise-bias-and-asymmetric-error-channels or erasure-error-vs-pauli-error before collapsing them into surface-code-logical-qubit, color-code-logical-qubit, or erasure-qubit.
• Photonic notes should usually enter this family through erasure-qubit: stay in photonic-moc while the question is encoding choice (dual-rail-photonic-qubit, time-bin-photonic-qubit) or resource construction (linear-optical-photonic-qubit, photonic-cluster-state-mbqc-qubit, fusion-based-photonic-qubit), then cross here only once flagged loss or fusion failure becomes a decoder-overhead question.
• Use classical-control and quantum-transduction as the two stack edges: one asks whether local feedback can keep up with the code cycle, the other asks whether the logical architecture survives once the machine becomes modular.

Boundary with the Codes MOC

• Stay in codes-moc while the live question is which stabilizers, gauge operators, or measurement schedule define the code.
• Cross here when the live question becomes what that code demands from a physical machine: local-feedback latency for Bacon-Shor or Floquet cycles, long-range connectivity for qLDPC checks, or decoder assumptions for erasure-aware logical qubits.
• surface-code-logical-qubit, color-code-logical-qubit, and erasure-qubit intentionally appear in both facets. In codes-moc they are code/noise-model families; here they are portable architectural overlays used to compare otherwise unrelated hardware modalities.

Scope boundary

This facet covers architectural glue: platform-local implementation details remain in their physical-platform maps, while this page identifies concepts that travel across platforms.