1-Qubit Ordering Strategies

Ordering strategies control which qubit position each MUX contributes to a 1-qubit calibration step. Box classification and shared-module grouping remain the responsibility of OneQubitScheduler.

Strategy Contract

MuxOrderingStrategy defines two operations:

• order_qids_in_mux() orders the selected qubits within one MUX for a MUX-group schedule.
• generate_synchronized_steps() creates chip-wide steps from a set of MUXes and qubits.

Both operations receive OrderingContext, which contains the chip ID, qubit-grid size, MUX-grid size, and qubit-to-MUX map. Strategies return qubit IDs only; they do not inspect wiring resources, start Prefect tasks, or execute hardware operations.

Built-in Strategies

DefaultOrderingStrategy uses numeric qubit order within each MUX. Its synchronized form takes the same local offset from every selected MUX at each step:

step 0: 4N
step 1: 4N + 1
step 2: 4N + 2
step 3: 4N + 3

DefaultSynchronizedStrategy has the same ordering behavior and a distinct metadata name for synchronized scheduling.

CheckerboardOrderingStrategy alternates local offsets by MUX parity:

MUX parity	Offset order
Even	0, 1, 2, 3
Odd	2, 3, 0, 1

For MUXes 0 through 3, the synchronized steps begin as follows:

step 0: 0, 6, 8, 14
step 1: 1, 7, 9, 15
step 2: 2, 4, 10, 12
step 3: 3, 5, 11, 13

This interleaving avoids selecting the same physical row position from every neighboring MUX at once. It is a scheduling policy based on the fixed four-qubits-per-MUX layout, not a replacement for measured-frequency validation.

How the Scheduler Uses a Strategy

For generate(), the caller may pass any MuxOrderingStrategy. The scheduler applies it independently to each MUX after box classification.

from qdash.workflow.engine import OneQubitScheduler
from qdash.workflow.engine.scheduler.one_qubit_plugins import (
    CheckerboardOrderingStrategy,
)

scheduler = OneQubitScheduler(chip_id="64Qv3")
schedule = scheduler.generate_from_mux(
    mux_ids=[0, 1, 2, 3],
    ordering_strategy=CheckerboardOrderingStrategy(),
)

For generate_synchronized(), callers select the built-in policy with use_checkerboard. The scheduler creates A steps, then B steps, then one or more MIXED step sequences. Empty local-offset steps are removed, so a partial target set can produce fewer than four steps per sequence.

Add a Strategy

Subclass MuxOrderingStrategy and implement all three abstract methods:

1. Return only qubits supplied to order_qids_in_mux().
2. Return synchronized steps whose union is exactly the supplied qubit set, without duplicates.
3. Return stable metadata that identifies the policy and any parameters that affect ordering.

Keep the strategy deterministic unless nondeterminism is an explicit, tested part of the policy. Test full and partial MUX selections, even and odd MUX IDs, and synchronized output. If the policy depends on new wiring data, add that responsibility to the scheduler context deliberately rather than loading files inside the strategy.

Source and Tests

• workflow/engine/scheduler/one_qubit_plugins.py contains the contract and built-ins.
• workflow/engine/scheduler/one_qubit_scheduler.py composes strategies with wiring constraints.
• tests/qdash/workflow/engine/scheduler/test_one_qubit_plugins.py tests ordering behavior.
• tests/qdash/workflow/engine/scheduler/test_one_qubit_scheduler.py tests composition into final schedules.

See 1-Qubit Calibration Scheduler for stage semantics and result forms.