Sampling

This module determines the sample size and is used to prepare chunk parameters for worker_func of Function implementations.

class Sampling:
    def max_chunks() -> int
    def get_state(offset: int, size: int) -> SamplingState

class SamplingState:
    def chunks(results: Results) -> List[SampleChunk]

The get_state method creates an instance of SamplingState with parameters offset and size for a specific estimating process.

The chunks method returns a list of SampleChunk which depends on the size of the already computed Results. Each SampleChunk is a tuple of the form (ChunkOffset, ChunkLength ).

You can also specify common arguments for all implementations:

• split_into – Chunk size for one worker thread/process. To calculate the estimated value, the sample is split into value/split_into chunks.

Const sampling

This implementation defines a sample of constant size.

from core.module.sampling import Const

sampling = Const(
    size: int,
    split_into: int
)

Epsilon sampling

This implementation defines a sample of dynamic size, starting at step_size and gradually increasing to max_size in step_size increments.

from core.module.sampling import Epsilon

sampling = Epsilon(
    step_size: int,
    epsilon: float,
    max_size: int,
    split_into: int,
    delta: float = 0.5
)

Other core modules

• Space

• Limitation

• Comparator