allocator

QuantityFactory

sizer = sizer instance-attribute

backend = backend instance-attribute

__init__(sizer, *, backend)

Initialize a QuantityFactory from a GridSizer and a NDSL backend name.

Parameters:

Name	Type	Description	Default
sizer	GridSizer	GridSizer object that determines the array sizes.	required
backend	Backend	NDSL backend name used for performance-optimized allocation.	required

update_data_dimensions(data_dimension_descriptions)

Update the length of data (non-x/y/z) dimensions, unknown data dimensions will be added, existing ones updated.

Parameters:

Name	Type	Description	Default
data_dimension_descriptions	dict[str, int]	Dict of name/length pairs	required

add_data_dimensions(data_dimension_descriptions)

Add new data (non-x/y/z) dimensions via a key-length pair. If the dimension already exists, it will error out.

Parameters:

Name	Type	Description	Default
data_dimension_descriptions	dict[str, int]	Dict of name/length pairs	required

empty(dims, units, dtype=Float, *, allow_mismatch_float_precision=False)

Allocate a Quantity and fill it with uninitialized (undefined) values.

Equivalent to numpy.empty

zeros(dims, units, dtype=Float, *, allow_mismatch_float_precision=False)

Allocate a Quantity and fill it with the value 0.

Equivalent to numpy.zeros

ones(dims, units, dtype=Float, *, allow_mismatch_float_precision=False)

Allocate a Quantity and fill it with the value 1.

Equivalent to numpy.ones

full(dims, units, value, dtype=Float, *, allow_mismatch_float_precision=False)

Allocate a Quantity and fill it with the given value.

Equivalent to numpy.full

from_array(data, dims, units, *, allow_mismatch_float_precision=False)

Create a Quantity from values in the data array.

This copies the values of data into the resulting Quantity. The data array thus must correspond to the correct shape and extent for the given dims.

from_compute_array(data, dims, units, *, allow_mismatch_float_precision=False)

Create a Quantity from values of the compute domain.

This function will allocate the full Quantity (including potential halo points) to zero. The values of data are then copied into the compute domain. That numpy array must correspond to the correct shape and extent of the compute domain for the given dims.

get_quantity_halo_spec(dims, n_halo=None, dtype=Float)

Build memory specifications for the halo update.

Parameters:

Name	Type	Description	Default
dims	Sequence[str]	dimensionality of the data	required
n_halo	int | None	number of halo points to update, defaults to self.n_halo	None
dtype	type	data type of the data	Float