iqm.qrisp_iqm.passes.routing.core.parameter_selection.compute_parameters

iqm.qrisp_iqm.passes.routing.core.parameter_selection.compute_parameters#

iqm.qrisp_iqm.passes.routing.core.parameter_selection.compute_parameters(effort: int, num_2qb_gates: int, cpu_count: int = 128) → dict[str, int]#

Derive transpiler parameters from a single effort level.

Parameters:

  • effort (int) – Classical compute investment level (≥ 1). - effort=1 : fast — one parallel wave, C=1-2 - effort=3 : moderate — 1-2 layout waves, C=2-6 - effort=5 : thorough — several layout waves, C=4-10 Wall-clock time scales roughly linearly with effort.

  • cpu_count (int, optional) – Number of available CPU cores (auto-detected by default).

  • num_2qb_gates (int, optional) – Number of 2-qubit gates in the circuit. Controls the budget split between layout attempts (ILA) and routing diversity (C). Shallow circuits (< ~100 gates) invest more in ILA; deep circuits invest more in C. When None, a balanced 50/50 split is used.

Returns:

Keys: init_layout_attempts, C, layout_iterations.

Return type:

dict

Examples

>>> params = compute_parameters(1, cpu_count=16)
>>> params['init_layout_attempts'] >= 16
True
>>> params['C'] >= 1
True
>>> params['layout_iterations']
3

>>> # Shallow circuit → heavy ILA, light C
>>> p = compute_parameters(5, cpu_count=16, num_2qb_gates=30)
>>> p['init_layout_attempts'] >= 48
True
>>> p['C'] <= 5
True

>>> # Deep circuit → light ILA, heavy C
>>> p = compute_parameters(5, cpu_count=16, num_2qb_gates=1000)
>>> p['init_layout_attempts'] <= 32
True
>>> p['C'] >= 8
True
Contents