time-to-botec/js/node_modules/@stdlib/math/tools/unary/docs/repl.txt

120 lines
4.0 KiB
Plaintext
Raw Normal View History

{{alias}}( table )
Returns a function which dispatches to specified functions based on input
argument types.
A `table` resolution object may contain one or more of the following fields:
- scalar: strided look-up table for scalar arguments.
- array: strided look-up table for array-like object arguments.
- ndarray: strided look-up table for ndarray arguments.
Each strided look-up table should be comprised as follows:
[ <dtype>, <fcn>, <dtype>, <fcn>, ... ]
If an argument's data type is *not* found in the argument's corresponding
look-up table and if a 'generic' data type is present in that same table,
the returned dispatch function will resolve the "generic" implementation. In
other words, an implementation associated with a 'generic' data type will be
treated as the default implementation.
If unable to resolve an implementation for a provided argument data type,
the returned function throws an error.
If provided a number, the returned function returns a number.
If provided an ndarray or array-like object, the returned function performs
element-wise computation.
If provided an array-like object, the returned function returns an array-
like object having the same length and data type as the provided input
argument.
If provided an ndarray, the returned function returns an ndarray having the
same shape and data type as the provided input argument.
Parameters
----------
table: Object
Table resolution object.
table.scalar: ArrayLikeObject (optional)
Strided look-up table for scalar arguments. Supported data types:
'number' and 'complex'.
table.array: ArrayLikeObject (optional)
Strided look-up table for array-like object arguments. Implementation
functions must follow strided array interface argument conventions:
fcn( N, x, strideX, y, strideY )
where
- N: number of indexed elements.
- x: input strided array.
- strideX: index increment for `x`.
- y: destination strided array.
- strideY: index increment for `y`.
Supported array data types consist of all supported ndarray data types.
table.ndarray: ArrayLikeObject (optional)
Strided look-up table for ndarray arguments. Implementation functions
must follow strided array ndarray interface argument conventions:
fcn( N, x, strideX, offsetX, y, strideY, offsetY )
where
- N: number of indexed elements.
- x: input strided array (i.e., underlying input ndarray buffer).
- strideX: index increment for `x`.
- offsetX: starting index for `x`.
- y: destination strided array (i.e., underlying output ndarray buffer).
- strideY: index increment for `y`.
- offsetY: starting index for `y`.
Supported data types consist of all supported ndarray data types.
Returns
-------
fcn: Function
Dispatch function.
Examples
--------
> var t = {};
> t.scalar = [ 'number', {{alias:@stdlib/math/base/special/abs}} ];
> t.array = [
... 'float64', {{alias:@stdlib/math/strided/special/dabs}},
... 'float32', {{alias:@stdlib/math/strided/special/sabs}},
... 'generic', {{alias:@stdlib/math/strided/special/abs}}
... ];
> t.ndarray = [
... 'float64', {{alias:@stdlib/math/strided/special/dabs}}.ndarray,
... 'float32', {{alias:@stdlib/math/strided/special/sabs}}.ndarray,
... 'generic', {{alias:@stdlib/math/strided/special/abs}}.ndarray
... ];
> var fcn = {{alias}}( t );
// Provide a number:
> var y = fcn( -1.0 )
1.0
// Provide an array-like object:
> var x = new {{alias:@stdlib/array/float64}}( [ -1.0, -2.0 ] );
> y = fcn( x )
<Float64Array>[ 1.0, 2.0 ]
// Provide an ndarray:
> x = {{alias:@stdlib/ndarray/array}}( [ [ -1.0, -2.0 ], [ -3.0, -4.0 ] ] );
> y = fcn( x )
<ndarray>
> y.get( 0, 1 )
2.0
See Also
--------