README.md

UnaryStrided1dDispatch

Constructor for applying a strided function to an input ndarray.

Usage

var UnaryStrided1dDispatch = require( '@stdlib/ndarray/base/unary-strided1d-dispatch' );

UnaryStrided1dDispatch( table, idtypes, odtypes, policies[, options] )

Returns an interface for applying a strided function to an input ndarray.

var base = require( '@stdlib/stats/base/ndarray/cumax' );

var table = {
    'default': base
};

var dtypes = [ 'float64', 'float32', 'generic' ];
var policies = {
    'output': 'same',
    'casting': 'none'
};

var unary = new UnaryStrided1dDispatch( table, [ dtypes ], dtypes, policies );

The constructor has the following parameters:

• table: strided function dispatch table. Must have the following properties:

  • default: default strided function which should be invoked when provided ndarrays have data types which do not have a corresponding specialized implementation.

  A dispatch table may have the following additional properties:

  • types: one-dimensional list of ndarray data types describing specialized input and output ndarray argument signatures. Only the input and output ndarray argument data types should be specified. Additional ndarray argument data types should be omitted and are not considered during dispatch. The length of types must equal the number of strided functions specified by fcns multiplied by two (i.e., for every pair of input-output ndarray data types, there must be a corresponding strided function in fcns).
  • fcns: list of strided functions which are specific to specialized input and output ndarray argument signatures.

• idtypes: list containing lists of supported input data types for each input ndarray argument.

• odtypes: list of supported output data types.

• policies: dispatch policies. Must have the following properties:

  • output: output data type policy.
  • casting: input ndarray casting policy.

• options: function options (optional).

The constructor supports the following options:

• strictTraversalOrder: boolean specifying whether the order of element traversal must match the memory layout order of an input ndarray. Default: false.

UnaryStrided1dDispatch.prototype.apply( x[, ...args][, options] )

Applies a strided function to a provided input ndarray.

var ndarray = require( '@stdlib/ndarray/base/ctor' );
var ndarray2array = require( '@stdlib/ndarray/to-array' );
var base = require( '@stdlib/stats/base/ndarray/cumax' );

var table = {
    'default': base
};

var dtypes = [ 'float64', 'float32', 'generic' ];
var policies = {
    'output': 'same',
    'casting': 'none'
};

var unary = new UnaryStrided1dDispatch( table, [ dtypes ], dtypes, policies );

var xbuf = [ -1.0, 2.0, -3.0 ];
var x = new ndarray( 'generic', xbuf, [ xbuf.length ], [ 1 ], 0, 'row-major' );

var y = unary.apply( x );
// returns <ndarray>

var arr = ndarray2array( y );
// returns [ -1.0, 2.0, 2.0 ]

The method has the following parameters:

• x: input ndarray.
• ...args: additional input ndarray arguments (optional).
• options: function options (optional).

The method accepts the following options:

• dims: list of dimensions over which to perform an operation.
• dtype: output ndarray data type. Setting this option overrides the output data type policy.

By default, the method returns an ndarray having a data type determined by the output data type policy. To override the default behavior, set the dtype option.

var ndarray = require( '@stdlib/ndarray/base/ctor' );
var base = require( '@stdlib/stats/base/ndarray/cumax' );
var getDType = require( '@stdlib/ndarray/dtype' );

var table = {
    'default': base
};

var dtypes = [ 'float64', 'float32', 'generic' ];
var policies = {
    'output': 'same',
    'casting': 'none'
};

var unary = new UnaryStrided1dDispatch( table, [ dtypes ], dtypes, policies );

var xbuf = [ -1.0, 2.0, -3.0 ];
var x = new ndarray( 'generic', xbuf, [ xbuf.length ], [ 1 ], 0, 'row-major' );

var y = unary.apply( x, {
    'dtype': 'float64'
});
// returns <ndarray>

var dt = getDType( y );
// returns 'float64'

UnaryStrided1dDispatch.prototype.assign( x[, ...args], out[, options] )

Applies a strided function a provided input ndarray and assigns results to a provided output ndarray.

var base = require( '@stdlib/stats/base/ndarray/cumax' );
var ndarray2array = require( '@stdlib/ndarray/to-array' );
var dtypes = require( '@stdlib/ndarray/dtypes' );
var ndarray = require( '@stdlib/ndarray/base/ctor' );

var idt = dtypes( 'real_and_generic' );
var odt = idt;
var policies = {
    'output': 'same',
    'casting': 'none'
};

var table = {
    'default': base
};
var unary = new UnaryStrided1dDispatch( table, [ idt ], odt, policies );

var xbuf = [ -1.0, 2.0, -3.0 ];
var x = new ndarray( 'generic', xbuf, [ xbuf.length ], [ 1 ], 0, 'row-major' );

var ybuf = [ 0.0, 0.0, 0.0 ];
var y = new ndarray( 'generic', ybuf, [ ybuf.length ], [ 1 ], 0, 'row-major' );

var out = unary.assign( x, y );
// returns <ndarray>

var arr = ndarray2array( y );
// returns [ -1.0, 2.0, 2.0 ]

var bool = ( out === y );
// returns true

The method has the following parameters:

• x: input ndarray.
• args: additional input ndarray arguments (optional).
• out: output ndarray.
• options: function options (optional).

The method accepts the following options:

• dims: list of dimensions over which to perform an operation.

Notes

• A strided function should have the following signature:

  f( arrays )
  

  where

  • arrays: array containing an input and an output ndarray, followed by any additional ndarray arguments.

• The output data type policy only applies to the apply method. For the assign method, the output ndarray is allowed to have any supported output data type.

Examples

var dcumax = require( '@stdlib/stats/base/ndarray/dcumax' );
var scumax = require( '@stdlib/stats/base/ndarray/scumax' );
var base = require( '@stdlib/stats/base/ndarray/cumax' );
var discreteUniform = require( '@stdlib/random/array/discrete-uniform' );
var dtypes = require( '@stdlib/ndarray/dtypes' );
var dtype = require( '@stdlib/ndarray/dtype' );
var ndarray2array = require( '@stdlib/ndarray/to-array' );
var ndarray = require( '@stdlib/ndarray/ctor' );
var UnaryStrided1dDispatch = require( '@stdlib/ndarray/base/unary-strided1d-dispatch' );

// Define the supported input and output data types:
var idt = dtypes( 'real_and_generic' );
var odt = dtypes( 'real_and_generic' );

// Define dispatch policies:
var policies = {
    'output': 'same',
    'casting': 'none'
};

// Define a dispatch table:
var table = {
    'types': [
        'float64', 'float64', // input, output
        'float32', 'float32'  // input, output
    ],
    'fcns': [
        dcumax,
        scumax
    ],
    'default': base
};

// Create an interface for performing a reduction:
var cumax = new UnaryStrided1dDispatch( table, [ idt ], odt, policies );

// Generate an array of random numbers:
var xbuf = discreteUniform( 25, -10, 10, {
    'dtype': 'generic'
});

// Wrap in an ndarray:
var x = new ndarray( 'generic', xbuf, [ 5, 5 ], [ 5, 1 ], 0, 'row-major' );
console.log( ndarray2array( x ) );

// Perform operation:
var y = cumax.apply( x, {
    'dims': [ 0 ]
});

// Resolve the output array data type:
var dt = dtype( y );
console.log( dt );

// Print the results:
console.log( ndarray2array( y ) );