Impulsed

class of dascore.proc.impulse
inherits from: PatchProcessor, DascoreBaseModel, pydantic.main.BaseModel
source

Impulsed(
    *args ,
    operation: Any ,
    on_unsettled: str = raise,
    **kwargs ,
)-> None

Run a linear operation in pieces, each read with the reach it measured.

The operation behind a patch method called on Patch.impulse_window, where the tolerances are described; they are given as keywords, such as time=1e-6.

Parameters

Parameter Description
operation The operation to run, as a
PatchProcessor. Its
class must say it is linear and act the same wherever a patch’s
coordinates start (position_free).
on_unsettled What to do when the response along a dimension does not fall under
its tolerance within the longest probe: "raise", the default,
or "whole", which needs that dimension whole as the operation
alone does.

Examples

import dascore as dc
from dascore.proc.impulse import Impulsed
patch = dc.get_example_patch()
out = patch.impulse_window(time=1e-6).pass_filter(time=(1, 20))
same = Impulsed(dc.proc.PassFilter(time=(1, 20)), time=1e-6)(patch)
assert out.equals(same)
Note
  • The reach along each dimension is measured by running the operation’s own kernel on an impulse, with the plan it will be run with: it is the smallest distance beyond which the summed absolute response is under the tolerance times the whole, before and after the impulse separately, so a causal filter reaches one way. The probe grows until the reach sits well inside it. The measurement depends on the plan alone and is kept for it.
  • Data in memory are computed whole, exactly as the operation alone computes them. A lazy or dask patch is computed a chunk at a time, each chunk read with the measured reach around it and that reach discarded, so it differs from the exact result by at most about the tolerance.

Methods

Name Description
check Check the patch for the operation run, then for the window.
chunk_contract Return the measured reach around a chunk; it may be its whole axis.
get_metadata Return the operation’s metadata, and the reach along each axis.
halo Return how far along an axis the kernel reads to fill a sample.
kernel Return the operation’s result on the data’s own backend.
model_copy Copy the model, dropping cached values the update invalidates.
new Create new instance with some attributed updated.
numpy_kernel Return the operation’s result on the data, computed as it computes it.
reconcile Return metadata or a patch holding the final data; default as is.
run Run the operation: check, get_metadata, kernel, reconcile, record.