Skip to main content

SharedWriteScenario

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.

Workers each writing writes_per_worker of files files, uniformly and independently.

Notes

Equation: The calculation members below define disjoint, disjoint_approx, collision, even_odds_fleet.

Assumptions: Each worker chooses a uniform write set independently, with no repeated file within a set.

Excludes: Hot-file concentration, synchronized edit behavior, and semantic merge conflicts between distinct files.

References

Derivation: Workers each writing writes_per_worker of files files, uniformly and independently. Each member records the corresponding algebra or named physics function.

Examples

Evaluate the named configuration and display a modeled result.

>>> from mlsysim import Scenarios
>>> case = Scenarios.ThousandFileSharedWrites
>>> value = case.disjoint[case.small_fleet]
>>> print(f"{value:.3g}")
0.904

Attributes

Name Description
collision Probability of at least one shared write by fleet count, complement of no-overlap probability.
disjoint Exact no-overlap probability by fleet count via calc_disjoint_writes.
disjoint_approx Approximate no-overlap probability by fleet count via calc_disjoint_writes_approx.
even_odds_fleet First enumerated fleet count with collision probability above one half, or None.
files
large_fleet
max_fleet
small_fleet
writes_per_worker

files

Type: int.

Declared default: required.

writes_per_worker

Type: int.

Declared default: required.

max_fleet

Type: int.

Declared default: required.

small_fleet

Type: int.

Declared default: required.

large_fleet

Type: int.

Declared default: required.

disjoint

engine.scenario_types.SharedWriteScenario.disjoint

Exact no-overlap probability by fleet count via calc_disjoint_writes.

Exact probability that no two write sets overlap, for each fleet size.

Calculation: mlsysim.physics.coordination.calc_disjoint_writes.

disjoint_approx

engine.scenario_types.SharedWriteScenario.disjoint_approx

Approximate no-overlap probability by fleet count via calc_disjoint_writes_approx.

Calculation: mlsysim.physics.coordination.calc_disjoint_writes_approx.

collision

engine.scenario_types.SharedWriteScenario.collision

Probability of at least one shared write by fleet count, complement of no-overlap probability.

Probability of at least one shared write.

Equation: {n: 1 - p for n, p in disjoint.items()}.

even_odds_fleet

engine.scenario_types.SharedWriteScenario.even_odds_fleet

First enumerated fleet count with collision probability above one half, or None.

The smallest fleet for which a shared write is more likely than not.

Equation: next((n for n, p in disjoint.items() if p < 0.5), None).

Back to top