Scenarios¶

The policyflux.scenarios package contains reusable comparative workflows.

Modules:

comparative_systems
country_comparison
lobbying_sweep
party_discipline_sweep
veto_player_sweep

What scenarios solve¶

Scenario modules package repeatable experiment patterns so you can:

vary one institutional or behavioral factor systematically,
preserve comparability across runs,
produce interpretable summary outputs for analysis.

Recommended scenario workflow¶

Choose a baseline (preset + fixed model scale).
Define a sweep variable range.
Run deterministic validation.
Run multi-seed or Monte Carlo confirmation.
Export and compare summary metrics.

Manual sweep template¶

from policyflux import build_engine, create_parliamentary_config

supports = [0.45, 0.55, 0.65, 0.75]

for support in supports:
    config = create_parliamentary_config(
        num_actors=100,
        policy_dim=2,
        iterations=200,
        seed=42,
    ).with_flat(public_support=support)

    engine = build_engine(config)
    engine.run()
    print(f"public_support={support:.2f} -> pass_rate={engine.pass_rate:.1%}")

Multi-seed scenario robustness¶

from policyflux import build_engine, create_presidential_config

def evaluate(public_support: float) -> float:
    rates = []
    for seed in range(1, 16):
        config = create_presidential_config(
            num_actors=120,
            policy_dim=2,
            iterations=220,
            seed=seed,
        ).with_flat(public_support=public_support)
        engine = build_engine(config)
        engine.run()
        rates.append(engine.pass_rate)
    return sum(rates) / len(rates)

print(evaluate(0.55), evaluate(0.65))

Interpreting scenario outputs¶

Prioritize:

direction of change (increase/decrease),
effect magnitude,
stability across seeds,
consistency across institutional presets.

For parameter design details, see Configuration and Presets.