yaw.config.ResamplingConfig

class yaw.config.ResamplingConfig(method: str = 'jackknife', crosspatch: bool = True, n_boot: int = 500, global_norm: bool = False, seed: int = 12345)

Bases: BaseConfig

Configuration for error estimation from spatial resampling.

Used for all functions and methods that use spatial patches for error estimation. Use the get_samples() method to generate samples from the spatial patches, which can be reused to ensure consistent error estimates for different data products that use the same patches.

Parameters:

• method (str) – Resampling method to use, see method.

• crosspath (str) – Whether to use cross-patch pair count measurements.

• n_boot (int) – Number of samples to generate for the bootstrap method.

• global_norm (bool) – Whether to normalise paircounts globally or for each sample. Usually not recommended.

• seed (int) – Random seed to use.

Methods

__init__([method, crosspatch, n_boot, ...])
create(**kwargs)	Create a new configuration object.
from_dict(the_dict, **kwargs)	Create a class instance from a dictionary representation of the minimally required data.
get_samples(n_patches)	Generate a list of patch indices that produces samples for the selected resampling method.
modify([method, crosspatch, n_boot, ...])	Create a copy of the current configuration with updated parameter values.
reset()	Reset the internally stored patch indices generated by get_samples().
to_dict()	Serialise the class instance to a dictionary containing a minimal set of required data.

Attributes

crosspatch	Whether to use cross-patch pair count measurements.
global_norm	Whether to normalise paircounts globally or for each sample.
method	Resampling method to use, see method.
n_boot	Number of samples to generate for the bootstrap method.
n_patches	The number of spatial patches for which this configuratin is valid.
seed	Random seed to use.

classmethod create(**kwargs: Any) → T

Create a new configuration object.

By default this is an alias for __init__(). Configuration classes that are hierarchical (i.e. contain configuration objects as attributes) implement this method to provide a single constructer for its own and its subclasses parameters.

crosspatch: bool = True

Whether to use cross-patch pair count measurements.

classmethod from_dict(the_dict: dict[str, Any], **kwargs: dict[str, Any]) → T

Create a class instance from a dictionary representation of the minimally required data.

Parameters:

• the_dict (dict) – Dictionary containing the data.

• **kwargs – Additional data needed to construct the class instance.

get_samples(n_patches: int) → NDArray[np.int64]

Generate a list of patch indices that produces samples for the selected resampling method.

Parameters:

n_patches (int) – Total number of patches for which the samples are generated.

Note

Samples are generated only once for each instance. Later calls to this method will only check if the number of patches agree with the first call and return the initially generated index list. Raises a ValueError otherwise.

The reason is, that the bootstrap method produces random samples, which must be consistent if the resampling is applied to different pair count measurements.

global_norm: bool = False

Whether to normalise paircounts globally or for each sample.

method: str = 'jackknife'

Resampling method to use, see method.

modify(method: str = NotSet, crosspatch: bool = NotSet, n_boot: int = NotSet, global_norm: bool = NotSet, seed: int = NotSet) → ResamplingConfig

Create a copy of the current configuration with updated parameter values.

The method arguments are identical to create(). Values that should not be modified are by default represented by the special value NotSet.

n_boot: int = 500

Number of samples to generate for the bootstrap method.

property n_patches: int | None

The number of spatial patches for which this configuratin is valid.

Available only after generating samples with get_samples().

Returns:

int if samples have been generated, else None.

reset() → None

Reset the internally stored patch indices generated by get_samples().

seed: int = 12345

Random seed to use.

to_dict() → dict[str, Any]

Serialise the class instance to a dictionary containing a minimal set of required data.

Returns:

dict