reachqDocumentation
GitHub
09 / guide

Parameters

Use the stable call-site parameters first; treat refinement toggles as algorithmic experiments, not product modes.

Edit this pageUpdated with the source

Primary parameters

  • random_seed makes seeded construction reproducible.
  • epsilon sets the hopset approximation budget.
  • omega controls the conservative exponent used for construction parameters.
  • parallel_workers only matters when the reachability refinement enables process dispatch.

RefinementConfig

RefinementConfig() enables the implemented refinements by default and keeps parallel disabled by default. Use RefinementConfig.from_dict for a partial, validated override.

from reachq.config import RefinementConfig

refinement = RefinementConfig(
    parallel=True,
    enable_tc_pruning=False,
)

Individual flags are implementation-level controls. They are documented because they are public inputs, but their names do not imply independent theoretical guarantees.