from dascore.lazy.contract import Halo
halo = Halo({0: (2, 3)})
box = (slice(10, 20), slice(0, 5))
assert halo.needs(box, (100, 5)) == (slice(8, 23), slice(0, 5))
assert halo.needs(box, (21, 5)) == (slice(8, 21), slice(0, 5))Halo
Halo(
widths: collections.abc.Mapping[int, tuple[int, int]] ,
)-> None
An output chunk needs the input chunk and some samples around it.
The output has the input’s shape, and each sample is computed from the input samples within a fixed distance of its own position. A chunk is grown by that distance, stopping at the ends of the array, so the kernel treats the array’s own ends as it does eagerly and what it makes of a cut end is trimmed off.
Parameters
| Parameter | Description |
|---|---|
| widths |
For each axis the kernel reaches along, how many samples before and after an output sample it reads. |
Examples
Methods
| Name | Description |
|---|---|
| apply | Return the output box, computed from the input it needs. |
| needs | Return the box grown by the widths, inside the array. |