SdcExposureScenario
Reliability scenarios · Implementation: mlsysim/engine/scenario_definitions/reliability.py.
Construct this scenario with keyword inputs. The fields below show the types and defaults declared by this class.
Silent-corruption exposure of n_devices, each at event_rate per device-hour, per step_time step.
Notes
Equation: Compose mlsysim.physics.reliability.calc_step_event_probability.
Assumptions: Independent event-bearing steps repeat under a stationary per-device event- rate model.
Excludes: Multiple corruptions inside one step are not separately counted by the step- probability conversion.
References
Derivation: this scenario binds its input fields to the referenced implementation contracts: mlsysim.physics.reliability.calc_step_event_probability. The methods below identify any additional arithmetic.
Examples
>>> from mlsysim import Scenarios
>>> case = Scenarios.SdcExposure10K
>>> round(case.exposure, 4)
5.5556Attributes
| Name | Description |
|---|---|
| event_rate | |
| exposure | Aggregate device exposure in one step. Equation: device count times step duration in |
| interval | Mean duration until an event-bearing step. Equation: steps_between * step_time. |
| n_devices | |
| p_step | Probability of at least one corruption event during a step. Equation: |
| run_duration | |
| run_events | Expected event-bearing steps over the run. Equation: run_duration / interval. Units: |
| step_time | |
| steps_between | Geometric mean number of steps until an event-bearing step. Equation: 1 / p_step. |
n_devices
Type: int.
Declared default: required.
event_rate
Type: float.
Declared default: required.
step_time
Type: Quantity.
Declared default: required.
run_duration
Type: Optional[Quantity].
Declared default: None.
exposure
engine.scenario_types.SdcExposureScenario.exposure
Aggregate device exposure in one step. Equation: device count times step duration in hours. Units: device-hours represented as a scalar.
p_step
engine.scenario_types.SdcExposureScenario.p_step
Probability of at least one corruption event during a step. Equation: mlsysim.physics.reliability.calc_step_event_probability(event_rate, n_devices, step_time). Units: probability under the stated event-rate model.
steps_between
engine.scenario_types.SdcExposureScenario.steps_between
Geometric mean number of steps until an event-bearing step. Equation: 1 / p_step. Units: steps; requires positive probability and independent repeated step trials.
interval
engine.scenario_types.SdcExposureScenario.interval
Mean duration until an event-bearing step. Equation: steps_between * step_time. Units: hours; this uses the step-level geometric model.
run_events
engine.scenario_types.SdcExposureScenario.run_events
Expected event-bearing steps over the run. Equation: run_duration / interval. Units: dimensionless count; multiple events in one step are counted together.