README.md (5377B)
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 # gapxsumpw 22 23 > Add a constant to each strided array element and compute the sum using pairwise summation. 24 25 <section class="intro"> 26 27 </section> 28 29 <!-- /.intro --> 30 31 <section class="usage"> 32 33 ## Usage 34 35 ```javascript 36 var gapxsumpw = require( '@stdlib/blas/ext/base/gapxsumpw' ); 37 ``` 38 39 #### gapxsumpw( N, alpha, x, stride ) 40 41 Adds a constant to each strided array element and computes the sum using pairwise summation. 42 43 ```javascript 44 var x = [ 1.0, -2.0, 2.0 ]; 45 var N = x.length; 46 47 var v = gapxsumpw( N, 5.0, x, 1 ); 48 // returns 16.0 49 ``` 50 51 The function has the following parameters: 52 53 - **N**: number of indexed elements. 54 - **x**: input [`Array`][mdn-array] or [`typed array`][mdn-typed-array]. 55 - **stride**: index increment for `x`. 56 57 The `N` and `stride` parameters determine which elements in `x` are accessed at runtime. For example, to access every other element in `x`, 58 59 ```javascript 60 var floor = require( '@stdlib/math/base/special/floor' ); 61 62 var x = [ 1.0, 2.0, 2.0, -7.0, -2.0, 3.0, 4.0, 2.0 ]; 63 var N = floor( x.length / 2 ); 64 65 var v = gapxsumpw( N, 5.0, x, 2 ); 66 // returns 25.0 67 ``` 68 69 Note that indexing is relative to the first index. To introduce an offset, use [`typed array`][mdn-typed-array] views. 70 71 <!-- eslint-disable stdlib/capitalized-comments --> 72 73 ```javascript 74 var Float64Array = require( '@stdlib/array/float64' ); 75 var floor = require( '@stdlib/math/base/special/floor' ); 76 77 var x0 = new Float64Array( [ 2.0, 1.0, 2.0, -2.0, -2.0, 2.0, 3.0, 4.0 ] ); 78 var x1 = new Float64Array( x0.buffer, x0.BYTES_PER_ELEMENT*1 ); // start at 2nd element 79 80 var N = floor( x0.length / 2 ); 81 82 var v = gapxsumpw( N, 5.0, x1, 2 ); 83 // returns 25.0 84 ``` 85 86 #### gapxsumpw.ndarray( N, alpha, x, stride, offset ) 87 88 Adds a constant to each strided array element and computes the sum using pairwise summation and alternative indexing semantics. 89 90 ```javascript 91 var x = [ 1.0, -2.0, 2.0 ]; 92 var N = x.length; 93 94 var v = gapxsumpw.ndarray( N, 5.0, x, 1, 0 ); 95 // returns 16.0 96 ``` 97 98 The function has the following additional parameters: 99 100 - **offset**: starting index for `x`. 101 102 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 access every other value in `x` starting from the second value 103 104 ```javascript 105 var floor = require( '@stdlib/math/base/special/floor' ); 106 107 var x = [ 2.0, 1.0, 2.0, -2.0, -2.0, 2.0, 3.0, 4.0 ]; 108 var N = floor( x.length / 2 ); 109 110 var v = gapxsumpw.ndarray( N, 5.0, x, 2, 1 ); 111 // returns 25.0 112 ``` 113 114 </section> 115 116 <!-- /.usage --> 117 118 <section class="notes"> 119 120 ## Notes 121 122 - If `N <= 0`, both functions return `0.0`. 123 - 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. 124 - Depending on the environment, the typed versions ([`dapxsumpw`][@stdlib/blas/ext/base/dapxsumpw], [`sapxsumpw`][@stdlib/blas/ext/base/sapxsumpw], etc.) are likely to be significantly more performant. 125 126 </section> 127 128 <!-- /.notes --> 129 130 <section class="examples"> 131 132 ## Examples 133 134 <!-- eslint no-undef: "error" --> 135 136 ```javascript 137 var randu = require( '@stdlib/random/base/randu' ); 138 var round = require( '@stdlib/math/base/special/round' ); 139 var Float64Array = require( '@stdlib/array/float64' ); 140 var gapxsumpw = require( '@stdlib/blas/ext/base/gapxsumpw' ); 141 142 var x; 143 var i; 144 145 x = new Float64Array( 10 ); 146 for ( i = 0; i < x.length; i++ ) { 147 x[ i ] = round( randu()*100.0 ); 148 } 149 console.log( x ); 150 151 var v = gapxsumpw( x.length, 5.0, x, 1 ); 152 console.log( v ); 153 ``` 154 155 </section> 156 157 <!-- /.examples --> 158 159 * * * 160 161 <section class="references"> 162 163 ## References 164 165 - 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]. 166 167 </section> 168 169 <!-- /.references --> 170 171 <section class="links"> 172 173 [mdn-array]: https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Array 174 175 [mdn-typed-array]: https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/TypedArray 176 177 [@stdlib/blas/ext/base/dapxsumpw]: https://www.npmjs.com/package/@stdlib/blas/tree/main/ext/base/dapxsumpw 178 179 [@stdlib/blas/ext/base/sapxsumpw]: https://www.npmjs.com/package/@stdlib/blas/tree/main/ext/base/sapxsumpw 180 181 [@higham:1993a]: https://doi.org/10.1137/0914050 182 183 </section> 184 185 <!-- /.links -->