Nationwide Polamorous Matrimony

    @stdlib/ndarray-dispatch
    TypeScript icon, indicating that this package has built-in type declarations

    0.0.6 • Public • Published

    Dispatch

    NPM version Build Status Coverage Status

    Create an ndarray function interface which performs multiple dispatch.

    Installation

    npm install @stdlib/ndarray-dispatch

    Usage

    var dispatch = require( '@stdlib/ndarray-dispatch' );

    dispatch( fcns, types, data, nargs, nin, nout )

    Returns an ndarray function interface which performs multiple dispatch.

    var unary = require( '@stdlib/ndarray-base-unary' );
    var Float64Array = require( '@stdlib/array-float64' );
    var Float32Array = require( '@stdlib/array-float32' );
    var ndarray = require( '@stdlib/ndarray-ctor' );
    
    function foo( x ) {
        return x * 10.0;
    }
    
    function bar( x ) {
        return x * 5.0;
    }
    
    // Define a list of ndarray functions for applying a unary callback:
    var fcns = [
        unary,
        unary
    ];
    
    // Define a one-dimensional list of input and output array types:
    var types = [
        'float64', 'float64', // input, output
        'float32', 'float32'  // input, output
    ];
    
    // Define a list of callbacks which should be applied based on the provided array types:
    var data = [
        foo,
        bar
    ];
    
    // Define the total number of input arguments:
    var nargs = 2; // input_array + output_array
    
    // Define the number of input ndarrays:
    var nin = 1;
    
    // Define the number of output ndarrays:
    var nout = 1;
    
    // Create an ndarray function interface:
    var fcn = dispatch( fcns, types, data, nargs, nin, nout );
    
    // ...
    
    var xbuf = new Float64Array( [ 1.0, 2.0, 3.0 ] );
    var ybuf = new Float64Array( xbuf.length );
    
    var x = ndarray( 'float64', xbuf, [ 3 ], [ 1 ], 0, 'row-major' );
    var y = ndarray( 'float64', ybuf, [ 3 ], [ 1 ], 0, 'row-major' );
    
    fcn( x, y );
    // ybuf => <Float64Array>[ 10.0, 20.0, 30.0 ]
    
    xbuf = new Float32Array( [ 1.0, 2.0, 3.0 ] );
    ybuf = new Float32Array( xbuf.length );
    
    x = ndarray( 'float32', xbuf, [ 3 ], [ 1 ], 0, 'row-major' );
    y = ndarray( 'float32', ybuf, [ 3 ], [ 1 ], 0, 'row-major' );
    
    fcn( x, y );
    // ybuf => <Float32Array>[ 5.0, 10.0, 15.0 ]

    The function accepts the following arguments:

    • fcns: list of ndarray functions.
    • types: one-dimensional list of ndarray argument data types. The length of types must be the number of ndarray functions multiplied by nin+nout. If fcns is a function, rather than a list, the number of ndarray functions is computed as types.length / (nin+nout).
    • data: ndarray function data (e.g., callbacks). If a list, the length of data must equal the number of ndarray functions. If null, a returned ndarray function interface does not provide a data argument to an invoked ndarray function.
    • nargs: total number of ndarray function interface arguments.
    • nin: number of input ndarrays.
    • nout: number of output ndarrays.

    Notes

    • A returned ndarray function interface has the following signature:

      f( x, y, ... )
      

      where

    • The number of ndarray function interface parameters is derived from nargs, the number of input ndarrays is derived from nin, and the number of output ndarrays is derived from nout.

    • An ndarray function (i.e., a value provided for the fcns argument) should have the following signature:

      f( arrays[, data] )
      

      where

      • arrays: array containing input and output ndarrays.
      • data: ndarray function data (e.g., a callback).
    • For convenience, a single ndarray function may be provided which will be invoked whenever the ndarray argument data types match a sequence of types in types. Providing a single ndarray function is particularly convenient for the case where, regardless of array data types, traversing arrays remains the same, but the ndarray function data differs (e.g., callbacks which differ based on the array data types). For example, the following

      var unary = require( '@stdlib/ndarray-base-unary' );
      
      function foo( x ) {
          return x * 10.0;
      }
      
      function bar( x ) {
          return x * 5.0;
      }
      
      var fcns = [
          unary,
          unary
      ];
      var types = [
          'float64', 'float64',
          'float32', 'float32'
      ];
      var data = [
          foo,
          bar
      ];
      
      var fcn = dispatch( fcns, types, data, 2, 1, 1 );

      is equivalent to

      var unary = require( '@stdlib/ndarray-base-unary' );
      
      function foo( x ) {
          return x * 10.0;
      }
      
      function bar( x ) {
          return x * 5.0;
      }
      
      var types = [
          'float64', 'float64',
          'float32', 'float32'
      ];
      var data = [
          foo,
          bar
      ];
      
      var fcn = dispatch( unary, types, data, 2, 1, 1 );

    Examples

    var unary = require( '@stdlib/ndarray-base-unary' );
    var ndarray = require( '@stdlib/ndarray-ctor' );
    var abs = require( '@stdlib/math-base-special-abs' );
    var Float64Array = require( '@stdlib/array-float64' );
    var dispatch = require( '@stdlib/ndarray-dispatch' );
    
    var types = [ 'float64', 'float64' ];
    
    var data = [
        abs
    ];
    
    var absolute = dispatch( unary, types, data, 2, 1, 1 );
    
    var xbuf = new Float64Array( [ -1.0, -2.0, -3.0, -4.0, -5.0 ] );
    var ybuf = new Float64Array( [ 0.0, 0.0, 0.0, 0.0, 0.0 ] );
    
    var x = ndarray( 'float64', xbuf, [ 5 ], [ 1 ], 0, 'row-major' );
    var y = ndarray( 'float64', ybuf, [ 5 ], [ 1 ], 0, 'row-major' );
    
    absolute( x, y );
    console.log( ybuf );
    // => <Float64Array>[ 1.0, 2.0, 3.0, 4.0, 5.0 ]

    See Also


    Notice

    This package is part of stdlib, a standard library for JavaScript and Node.js, with an emphasis on numerical and scientific computing. The library provides a collection of robust, high performance libraries for mathematics, statistics, streams, utilities, and more.

    For more information on the project, filing bug reports and feature requests, and guidance on how to develop stdlib, see the main project repository.

    Community

    Chat


    License

    See LICENSE.

    Copyright

    Copyright © 2016-2022. The Stdlib Authors.

    Install

    npm i @stdlib/ndarray-dispatch

    Homepage

    stdlib.io

    DownloadsWeekly Downloads

    84

    Version

    0.0.6

    License

    Apache-2.0

    Unpacked Size

    83.5 kB

    Total Files

    10

    Last publish

    Collaborators

    • stdlib-bot
    • kgryte
    • planeshifter
    • rreusser