# ssortsh
> Sort a single-precision floating-point strided array using Shellsort.
## Usage
```javascript
var ssortsh = require( '@stdlib/blas/ext/base/ssortsh' );
```
#### ssortsh( N, order, x, stride )
Sorts a single-precision floating-point strided array `x` using Shellsort.
```javascript
var Float32Array = require( '@stdlib/array/float32' );
var x = new Float32Array( [ 1.0, -2.0, 3.0, -4.0 ] );
ssortsh( x.length, 1.0, x, 1 );
// x => [ -4.0, -2.0, 1.0, 3.0 ]
```
The function has the following parameters:
- **N**: number of indexed elements.
- **order**: sort order. If `order < 0.0`, the input strided array is sorted in **decreasing** order. If `order > 0.0`, the input strided array is sorted in **increasing** order. If `order == 0.0`, the input strided array is left unchanged.
- **x**: input [`Float32Array`][@stdlib/array/float32].
- **stride**: index increment.
The `N` and `stride` parameters determine which elements in `x` are accessed at runtime. For example, to sort every other element
```javascript
var Float32Array = require( '@stdlib/array/float32' );
var floor = require( '@stdlib/math/base/special/floor' );
var x = new Float32Array( [ 1.0, -2.0, 3.0, -4.0 ] );
var N = floor( x.length / 2 );
ssortsh( N, -1.0, x, 2 );
// x => [ 3.0, -2.0, 1.0, -4.0 ]
```
Note that indexing is relative to the first index. To introduce an offset, use [`typed array`][mdn-typed-array] views.
```javascript
var Float32Array = require( '@stdlib/array/float32' );
var floor = require( '@stdlib/math/base/special/floor' );
// Initial array...
var x0 = new Float32Array( [ 1.0, 2.0, 3.0, 4.0 ] );
// Create an offset view...
var x1 = new Float32Array( x0.buffer, x0.BYTES_PER_ELEMENT*1 ); // start at 2nd element
var N = floor( x0.length/2 );
// Sort every other element...
ssortsh( N, -1.0, x1, 2 );
// x0 => [ 1.0, 4.0, 3.0, 2.0 ]
```
#### ssortsh.ndarray( N, order, x, stride, offset )
Sorts a single-precision floating-point strided array `x` using Shellsort and alternative indexing semantics.
```javascript
var Float32Array = require( '@stdlib/array/float32' );
var x = new Float32Array( [ 1.0, -2.0, 3.0, -4.0 ] );
ssortsh.ndarray( x.length, 1.0, x, 1, 0 );
// x => [ -4.0, -2.0, 1.0, 3.0 ]
```
The function has the following additional parameters:
- **offset**: starting index.
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 only the last three elements of `x`
```javascript
var Float32Array = require( '@stdlib/array/float32' );
var x = new Float32Array( [ 1.0, -2.0, 3.0, -4.0, 5.0, -6.0 ] );
ssortsh.ndarray( 3, 1.0, x, 1, x.length-3 );
// x => [ 1.0, -2.0, 3.0, -6.0, -4.0, 5.0 ]
```
## Notes
- If `N <= 0` or `order == 0.0`, both functions return `x` unchanged.
- The algorithm distinguishes between `-0` and `+0`. When sorted in increasing order, `-0` is sorted before `+0`. When sorted in decreasing order, `-0` is sorted after `+0`.
- The algorithm sorts `NaN` values to the end. When sorted in increasing order, `NaN` values are sorted last. When sorted in decreasing order, `NaN` values are sorted first.
- The algorithm has space complexity `O(1)` and worst case time complexity `O(N^(4/3))`.
- The algorithm is efficient for **shorter** strided arrays (typically `N <= 50`).
- The algorithm is **unstable**, meaning that the algorithm may change the order of strided array elements which are equal or equivalent (e.g., `NaN` values).
- The input strided array is sorted **in-place** (i.e., the input strided array is **mutated**).
## Examples
```javascript
var round = require( '@stdlib/math/base/special/round' );
var randu = require( '@stdlib/random/base/randu' );
var Float32Array = require( '@stdlib/array/float32' );
var ssortsh = require( '@stdlib/blas/ext/base/ssortsh' );
var rand;
var sign;
var x;
var i;
x = new Float32Array( 10 );
for ( i = 0; i < x.length; i++ ) {
rand = round( randu()*100.0 );
sign = randu();
if ( sign < 0.5 ) {
sign = -1.0;
} else {
sign = 1.0;
}
x[ i ] = sign * rand;
}
console.log( x );
ssortsh( x.length, -1.0, x, -1 );
console.log( x );
```
* * *
## References
- Shell, Donald L. 1959. "A High-Speed Sorting Procedure." _Communications of the ACM_ 2 (7). Association for Computing Machinery: 30–32. doi:[10.1145/368370.368387][@shell:1959a].
- Sedgewick, Robert. 1986. "A new upper bound for Shellsort." _Journal of Algorithms_ 7 (2): 159–73. doi:[10.1016/0196-6774(86)90001-5][@sedgewick:1986a].
- Ciura, Marcin. 2001. "Best Increments for the Average Case of Shellsort." In _Fundamentals of Computation Theory_, 106–17. Springer Berlin Heidelberg. doi:[10.1007/3-540-44669-9_12][@ciura:2001a].
[@stdlib/array/float32]: https://www.npmjs.com/package/@stdlib/array-float32
[mdn-typed-array]: https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/TypedArray
[@shell:1959a]: https://doi.org/10.1145/368370.368387
[@sedgewick:1986a]: https://doi.org/10.1016/0196-6774(86)90001-5
[@ciura:2001a]: https://doi.org/10.1007/3-540-44669-9_12
[@stdlib/blas/ext/base/dsortsh]: https://github.com/stdlib-js/blas/tree/main/ext/base/dsortsh
[@stdlib/blas/ext/base/gsortsh]: https://github.com/stdlib-js/blas/tree/main/ext/base/gsortsh
[@stdlib/blas/ext/base/ssort2sh]: https://github.com/stdlib-js/blas/tree/main/ext/base/ssort2sh