time-to-botec

Benchmark sampling in different programming languages
Log | Files | Refs | README

README.md (6014B)


      1 <!--
      2 
      3 @license Apache-2.0
      4 
      5 Copyright (c) 2020 The Stdlib Authors.
      6 
      7 Licensed under the Apache License, Version 2.0 (the "License");
      8 you may not use this file except in compliance with the License.
      9 You may obtain a copy of the License at
     10 
     11    http://www.apache.org/licenses/LICENSE-2.0
     12 
     13 Unless required by applicable law or agreed to in writing, software
     14 distributed under the License is distributed on an "AS IS" BASIS,
     15 WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
     16 See the License for the specific language governing permissions and
     17 limitations under the License.
     18 
     19 -->
     20 
     21 # gasumpw
     22 
     23 > Calculate the sum of absolute values ([_L1_ norm][l1norm]) of strided array elements using pairwise summation.
     24 
     25 <section class="intro">
     26 
     27 The [_L1_ norm][l1norm] is defined as
     28 
     29 <!-- <equation class="equation" label="eq:l1norm" align="center" raw="\|\mathbf{x}\|_1 = \sum_{i=0}^{n-1} \vert x_i \vert" alt="L1 norm definition."> -->
     30 
     31 <div class="equation" align="center" data-raw-text="\|\mathbf{x}\|_1 = \sum_{i=0}^{n-1} \vert x_i \vert" data-equation="eq:l1norm">
     32     <img src="https://cdn.jsdelivr.net/gh/stdlib-js/stdlib@b989712575768de8eab07dd05f635c636b46f278/lib/node_modules/@stdlib/blas/ext/base/gasumpw/docs/img/equation_l1norm.svg" alt="L1 norm definition.">
     33     <br>
     34 </div>
     35 
     36 <!-- </equation> -->
     37 
     38 </section>
     39 
     40 <!-- /.intro -->
     41 
     42 <section class="usage">
     43 
     44 ## Usage
     45 
     46 ```javascript
     47 var gasumpw = require( '@stdlib/blas/ext/base/gasumpw' );
     48 ```
     49 
     50 #### gasumpw( N, x, stride )
     51 
     52 Computes the sum of absolute values ([_L1_ norm][l1norm]) of strided array elements using pairwise summation.
     53 
     54 ```javascript
     55 var x = [ 1.0, -2.0, 2.0 ];
     56 
     57 var v = gasumpw( x.length, x, 1 );
     58 // returns 5.0
     59 ```
     60 
     61 The function has the following parameters:
     62 
     63 -   **N**: number of indexed elements.
     64 -   **x**: input [`Array`][mdn-array] or [`typed array`][mdn-typed-array].
     65 -   **stride**: index increment for `x`.
     66 
     67 The `N` and `stride` parameters determine which elements in `x` are accessed at runtime. For example, to compute the sum of absolute values of every other element in `x`,
     68 
     69 ```javascript
     70 var floor = require( '@stdlib/math/base/special/floor' );
     71 
     72 var x = [ 1.0, 2.0, 2.0, -7.0, -2.0, 3.0, 4.0, 2.0 ];
     73 var N = floor( x.length / 2 );
     74 
     75 var v = gasumpw( N, x, 2 );
     76 // returns 9.0
     77 ```
     78 
     79 Note that indexing is relative to the first index. To introduce an offset, use [`typed array`][mdn-typed-array] views.
     80 
     81 <!-- eslint-disable stdlib/capitalized-comments -->
     82 
     83 ```javascript
     84 var Float64Array = require( '@stdlib/array/float64' );
     85 var floor = require( '@stdlib/math/base/special/floor' );
     86 
     87 var x0 = new Float64Array( [ 2.0, 1.0, 2.0, -2.0, -2.0, 2.0, 3.0, 4.0 ] );
     88 var x1 = new Float64Array( x0.buffer, x0.BYTES_PER_ELEMENT*1 ); // start at 2nd element
     89 
     90 var N = floor( x0.length / 2 );
     91 
     92 var v = gasumpw( N, x1, 2 );
     93 // returns 9.0
     94 ```
     95 
     96 #### gasumpw.ndarray( N, x, stride, offset )
     97 
     98 Computes the sum of absolute values ([_L1_ norm][l1norm]) of strided array elements using pairwise summation and alternative indexing semantics.
     99 
    100 ```javascript
    101 var x = [ 1.0, -2.0, 2.0 ];
    102 
    103 var v = gasumpw.ndarray( x.length, x, 1, 0 );
    104 // returns 5.0
    105 ```
    106 
    107 The function has the following additional parameters:
    108 
    109 -   **offset**: starting index for `x`.
    110 
    111 While [`typed array`][mdn-typed-array] views mandate a view offset based on the underlying `buffer`, the `offset` parameter supports indexing semantics based on a starting index. For example, to calculate the sum of absolute values of every other value in `x` starting from the second value
    112 
    113 ```javascript
    114 var floor = require( '@stdlib/math/base/special/floor' );
    115 
    116 var x = [ 2.0, 1.0, 2.0, -2.0, -2.0, 2.0, 3.0, 4.0 ];
    117 var N = floor( x.length / 2 );
    118 
    119 var v = gasumpw.ndarray( N, x, 2, 1 );
    120 // returns 9.0
    121 ```
    122 
    123 </section>
    124 
    125 <!-- /.usage -->
    126 
    127 <section class="notes">
    128 
    129 ## Notes
    130 
    131 -   If `N <= 0`, both functions return `0.0`.
    132 -   In general, pairwise summation is more numerically stable than ordinary recursive summation (i.e., "simple" summation), with slightly worse performance. While not the most numerically stable summation technique (e.g., compensated summation techniques such as the Kahan–Babuška-Neumaier algorithm are generally more numerically stable), pairwise summation strikes a reasonable balance between numerical stability and performance. If either numerical stability or performance is more desirable for your use case, consider alternative summation techniques.
    133 -   Depending on the environment, the typed versions ([`dasumpw`][@stdlib/blas/ext/base/dasumpw], [`sasumpw`][@stdlib/blas/ext/base/sasumpw], etc.) are likely to be significantly more performant.
    134 
    135 </section>
    136 
    137 <!-- /.notes -->
    138 
    139 <section class="examples">
    140 
    141 ## Examples
    142 
    143 <!-- eslint no-undef: "error" -->
    144 
    145 ```javascript
    146 var randu = require( '@stdlib/random/base/randu' );
    147 var round = require( '@stdlib/math/base/special/round' );
    148 var Float64Array = require( '@stdlib/array/float64' );
    149 var gasumpw = require( '@stdlib/blas/ext/base/gasumpw' );
    150 
    151 var x;
    152 var i;
    153 
    154 x = new Float64Array( 10 );
    155 for ( i = 0; i < x.length; i++ ) {
    156     x[ i ] = round( randu()*100.0 );
    157 }
    158 console.log( x );
    159 
    160 var v = gasumpw( x.length, x, 1 );
    161 console.log( v );
    162 ```
    163 
    164 </section>
    165 
    166 <!-- /.examples -->
    167 
    168 * * *
    169 
    170 <section class="references">
    171 
    172 ## References
    173 
    174 -   Higham, Nicholas J. 1993. "The Accuracy of Floating Point Summation." _SIAM Journal on Scientific Computing_ 14 (4): 783–99. doi:[10.1137/0914050][@higham:1993a].
    175 
    176 </section>
    177 
    178 <!-- /.references -->
    179 
    180 <section class="links">
    181 
    182 [mdn-array]: https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Array
    183 
    184 [mdn-typed-array]: https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/TypedArray
    185 
    186 [@stdlib/blas/ext/base/dasumpw]: https://www.npmjs.com/package/@stdlib/blas/tree/main/ext/base/dasumpw
    187 
    188 [@stdlib/blas/ext/base/sasumpw]: https://www.npmjs.com/package/@stdlib/blas/tree/main/ext/base/sasumpw
    189 
    190 [l1norm]: http://en.wikipedia.org/wiki/Norm_%28mathematics%29
    191 
    192 [@higham:1993a]: https://doi.org/10.1137/0914050
    193 
    194 </section>
    195 
    196 <!-- /.links -->