fastscapelib.flow.numba.flow_kernel.NumbaFlowKernel¶

class fastscapelib.flow.numba.flow_kernel.NumbaFlowKernel(kernel, func, apply_dir, outputs, n_threads, generated_code, generated_spec, node_data_create, node_data_init, node_data_getter, node_data_setter, node_data_free)¶

Immutable Proxy object representing a numba jit-compiled flow kernel function.

It is returned by create_flow_kernel() and required by apply_kernel() alongside the NumbaFlowKernelData mapping object.

kernel¶

An internal object used to pass and execute the flow kernel either from C++ or from numba jit-compiled code.

Type:: object

func¶

The numba-compiled kernel function, which accepts one argument (the kernel’s node data as an instance of a numba jit-compiled class).

Type:: callable

apply_dir¶

The direction and order in which the flow kernel will be applied along the graph.

Type:: FlowGraphTraversalDir

outputs¶

The names of the kernel output variables.

Type:: tuple

n_threads¶

Number of threads to use for applying the kernel function in parallel along the flow graph (if equal to 1, the kernel is applied sequentially).

Type:: int

generated_code¶

A dictionary with the source code generated for each of the functions below.

Type:: dict

generated_spec¶

The numba types of the attributes of the jit-compiled classes used by the kernel.

Type:: tuple

node_data_create¶

Internal numba compiled function used to create the kernel’s node data object (one instance per each thread).

Type:: callable

node_data_init¶

Internal numba jit-compiled function used to initialize node data scalar values.

Type:: callable or None

node_data_getter¶

Internal numba jit-compiled function used to get grid data values at the current node, its receivers and its donors.

Type:: callable

node_data_setter¶

Internal numba jit-compiled function used to set grid data values from the kernel’s data at the current node, its receivers and its donors.

Type:: callable

node_data_free¶

Used internally to de-allocate memory used by the kernel’s node data.

Type:: object

__init__(kernel, func, apply_dir, outputs, n_threads, generated_code, generated_spec, node_data_create, node_data_init, node_data_getter, node_data_setter, node_data_free)¶

Methods

__init__(kernel, func, apply_dir, outputs, ...)

Attributes

`kernel`
`func`
`apply_dir`
`outputs`
`n_threads`
`generated_code`
`generated_spec`
`node_data_create`
`node_data_init`
`node_data_getter`
`node_data_setter`
`node_data_free`