10 KiB
ind2sub
Convert a linear index to an array of subscripts.
Usage
var ind2sub = require( '@stdlib/ndarray/base/ind2sub' );
ind2sub( shape, strides, offset, order, idx, mode )
Converts a linear index to an array of subscripts.
var shape = [ 2, 2 ];
var order = 'row-major';
var strides = [ 2, 1 ];
var offset = 0;
var subscripts = ind2sub( shape, strides, offset, order, 1, 'throw' );
// returns [ 0, 1 ]
The function supports the following modes
:
throw
: specifies that the function should throw an error when a linear index exceeds array dimensions.wrap
: specifies that the function should wrap around a linear index exceeding array dimensions using modulo arithmetic.clamp
: specifies that the function should set a linear index exceeding array dimensions to either0
(minimum linear index) or the maximum linear index.
var shape = [ 2, 2 ];
var order = 'row-major';
var strides = [ 2, 1 ];
var offset = 0;
var idx = ind2sub( shape, strides, offset, order, -2, 'wrap' );
// returns [ 1, 0 ]
idx = ind2sub( shape, strides, offset, order, 10, 'clamp' );
// returns [ 1, 1 ]
The order
parameter specifies whether an array is row-major
(C-style) or column-major
(Fortran-style).
var shape = [ 2, 2 ];
var order = 'column-major';
var strides = [ 1, 2 ];
var offset = 0;
var idx = ind2sub( shape, strides, offset, order, 2, 'throw' );
// returns [ 0, 1 ]
ind2sub.assign( shape, strides, offset, order, idx, mode, out )
Converts a linear index to an array of subscripts and assigns results to a provided output array.
var shape = [ 2, 2 ];
var order = 'row-major';
var strides = [ 2, 1 ];
var offset = 0;
var out = [ 0, 0 ];
var subscripts = ind2sub.assign( shape, strides, offset, order, 1, 'throw', out );
// returns [ 0, 1 ]
var bool = ( subscripts === out );
// returns true
Notes
-
When provided a stride array containing negative strides, if an
offset
is greater than0
, the function interprets the linear index as an index into the underlying data buffer for the array, thus returning subscripts from the perspective of that buffer. If anoffset
is equal to0
, the function treats the linear index as an index into an array view, thus returning subscripts from the perspective of that view.Dims: 2x2 Buffer: [ 1, 2, 3, 4 ] View = [ a00, a01, a10, a11 ] Strides: 2,1 Offset: 0 View = [ 1, 2, 3, 4 ] Strides: 2,-1 Offset: 1 View = [ 2, 1, 4, 3 ] Strides: -2,1 Offset: 2 View = [ 3, 4, 1, 2 ] Strides: -2,-1 Offset: 3 View = [ 4, 3, 2, 1 ]
var shape = [ 2, 2 ]; var order = 'row-major'; var strides = [ -2, 1 ]; var offset = 2; var mode = 'throw'; // From the perspective of a view... var s = ind2sub( shape, strides, 0, order, 0, mode ); // returns [ 0, 0 ] s = ind2sub( shape, strides, 0, order, 1, mode ); // returns [ 0, 1 ] s = ind2sub( shape, strides, 0, order, 2, mode ); // returns [ 1, 0 ] s = ind2sub( shape, strides, 0, order, 3, mode ); // returns [ 1, 1 ] // From the perspective of an underlying buffer... s = ind2sub( shape, strides, offset, order, 0, mode ); // returns [ 1, 0 ] s = ind2sub( shape, strides, offset, order, 1, mode ); // returns [ 1, 1 ] s = ind2sub( shape, strides, offset, order, 2, mode ); // returns [ 0, 0 ] s = ind2sub( shape, strides, offset, order, 3, mode ); // returns [ 0, 1 ]
In short, from the perspective of a view, view data is always ordered.
Examples
var discreteUniform = require( '@stdlib/random/base/discrete-uniform' );
var shape2strides = require( '@stdlib/ndarray/base/shape2strides' );
var strides2offset = require( '@stdlib/ndarray/base/strides2offset' );
var numel = require( '@stdlib/ndarray/base/numel' );
var randu = require( '@stdlib/random/base/randu' );
var abs = require( '@stdlib/math/base/special/abs' );
var ind2sub = require( '@stdlib/ndarray/base/ind2sub' );
// Specify array characteristics:
var shape = [ 3, 3, 3 ];
var order = 'row-major';
// Compute array meta data:
var ndims = shape.length;
var strides = shape2strides( shape, order );
var len = numel( shape );
// Determine stride indices to be used for formatting how views are displayed...
var s0;
var s1;
if ( order === 'column-major' ) {
s0 = ndims - 1;
s1 = s0 - 1;
} else { // row-major
s0 = 0;
s1 = s0 + 1;
}
// Initialize a linear array...
var arr = [];
var i;
for ( i = 0; i < len; i++ ) {
arr.push( 0 );
}
// Generate random views and display the mapping of elements in the linear array to view subscripts...
var offset;
var row;
var j;
var s;
for ( i = 0; i < 20; i++ ) {
// Randomly flip the sign of one of the strides...
j = discreteUniform( 0, ndims-1 );
strides[ j ] *= ( randu() < 0.5 ) ? -1 : 1;
offset = strides2offset( shape, strides );
// Print view meta data...
console.log( '' );
console.log( 'Dimensions: %s.', shape.join( 'x' ) );
console.log( 'Strides: %s.', strides.join( ',' ) );
console.log( 'View (subscripts):' );
// Print the mapping of elements in the linear array to view subscripts...
row = ' ';
for ( j = 0; j < len; j++ ) {
s = ind2sub( shape, strides, offset, order, j, 'throw' );
row += '(' + s.join( ',' ) + ')';
if ( ndims === 1 && j === len-1 ) {
console.log( row );
} else if ( ndims === 2 && (j+1)%abs( strides[ s0 ] ) === 0 ) {
console.log( row );
row = ' ';
} else if ( ndims > 2 && (j+1)%abs( strides[ s1 ] ) === 0 ) {
console.log( row );
if ( (j+1)%abs( strides[ s0 ] ) === 0 ) {
console.log( '' );
}
row = ' ';
} else {
row += ', ';
}
}
}
C APIs
Usage
#include "stdlib/ndarray/base/ind2sub.h"
stdlib_ndarray_ind2sub( ndims, *shape, *strides, offset, order, idx, mode, *out )
Computes the minimum and maximum linear indices in an underlying data buffer accessible to an array view.
#include "stdlib/ndarray/index_modes.h"
#include "stdlib/ndarray/orders.h"
#include <stdint.h>
int64_t ndims = 2;
int64_t shape[] = { 3, 3 };
int64_t strides[] = { -3, 1 };
int64_t offset = 6;
int64_t out[ 2 ];
int8_t status = stdlib_ndarray_ind2sub( ndims, shape, strides, offset, STDLIB_NDARRAY_ROW_MAJOR, 7, STDLIB_NDARRAY_INDEX_ERROR, out );
if ( status == -1 ) {
// Handle error...
}
The function accepts the following arguments:
- ndims:
[in] int64_t
number of dimensions. - shape:
[in] int64_t*
array shape (dimensions). - strides:
[in] int64_t*
array strides. - offset:
[in] int64_t
index offset. - order:
[in] enum STDLIB_NDARRAY_ORDER
specifies whether an array is row-major (C-style) or column-major (Fortran-style). - idx:
[in] int64_t
linear index in an array view. - mode:
[in] enum STDLIB_NDARRAY_INDEX_MODE
specifies how to handle a linear index which exceeds array dimensions. - out:
[out] int64_t*
output array.
int8_t stdlib_ndarray_ind2sub( int64_t ndims, int64_t *shape, int64_t *strides, int64_t offset, enum STDLIB_NDARRAY_ORDER order, int64_t idx, enum STDLIB_NDARRAY_INDEX_MODE mode, int64_t *out );
Examples
#include "stdlib/ndarray/base/ind2sub.h"
#include "stdlib/ndarray/index_modes.h"
#include "stdlib/ndarray/orders.h"
#include <stdint.h>
#include <stdio.h>
#include <inttypes.h>
int main() {
int64_t ndims = 2;
int64_t shape[] = { 3, 3 };
int64_t strides[] = { -3, 1 };
int64_t offset = 6;
int64_t out[ 2 ];
stdlib_ndarray_ind2sub( ndims, shape, strides, offset, STDLIB_NDARRAY_ROW_MAJOR, 7, STDLIB_NDARRAY_INDEX_ERROR, out );
int i;
printf( "subscripts = { " );
for ( i = 0; i < ndims; i++ ) {
printf( "%"PRId64"", out[ i ] );
if ( i < ndims-1 ) {
printf( ", " );
}
}
printf( " }\n" );
}