PolicyComparisonScenario
Agents scenarios · Implementation: mlsysim/engine/scenario_definitions/agents.py.
Construct this scenario with keyword inputs. The fields below show the types and defaults declared by this class.
Two policies, each run on attacked attacked and benign separate benign episodes.
Each policy’s counts: confirmed and unresolved harm and accepted_attacked among attacked episodes; accepted_benign and false_block among benign ones.
Notes
Equation: The result property composes mlsysim.solvers.PolicyComparisonModel.solve with the stored fields and any derived properties described below.
Assumptions: Inputs use the units and validity ranges of mlsysim.solvers.PolicyComparisonModel.solve. The named scenario supplies a fixed configuration rather than observations of a new run.
Excludes: Additional runtime effects or statistical guarantees beyond the cited solver contracts; no measurement is performed by this wrapper.
References
Derivation: The result property composes mlsysim.solvers.PolicyComparisonModel.solve with the stored fields and any derived properties described below. Individual property contracts identify the exact accounting expression or reusable calculation. Registry entries supply the scenario inputs; this class does not turn them into measured outcomes.
Examples
Evaluate a named scenario using its declared inputs.
>>> from mlsysim import Scenarios
>>> scenario = Scenarios.FixtureReleasePolicyComparison
>>> round(float(scenario.result.baseline_lower), 4)
0.2Attributes
| Name | Description |
|---|---|
| attacked | |
| attempt_cost | |
| baseline | |
| benign | |
| candidate | |
| result | Return the modeled PolicyComparison result from |
attacked
Type: int.
Declared default: required.
benign
Type: int.
Declared default: required.
baseline
Type: Dict[str, int].
Declared default: required.
candidate
Type: Dict[str, int].
Declared default: required.
attempt_cost
Type: Optional[float].
Declared default: None.
result
engine.scenario_types.PolicyComparisonScenario.result
Return the modeled PolicyComparison result from mlsysim.solvers.PolicyComparisonModel.solve.
Equation: the scenario supplies its named fields to that solver. Returned quantity units, validity conditions, and omissions follow its calculation contract; this property adds no empirical calibration.