How do I access All rows (resp. columns ) of a Matrix / multidimensional array?

In scilab, you write array_var(:,row_index), colon meaning all columns, at that particular row given by row index will be selected.

can you, forexample, in D, write, array_var[0][] to select all elements (columns, in this case i believe?) of array_var[0] ?

On Thursday, 14 November 2013 at 22:06:28 UTC, seany wrote:
> How do I access All rows (resp. columns ) of a Matrix / multidimensional array?
>
> In scilab, you write array_var(:,row_index), colon meaning all columns, at that particular row given by row index will be selected.
>
> can you, forexample, in D, write, array_var[0][] to select all elements (columns, in this case i believe?) of array_var[0] ?

unittest
{
  int[][] matrix = new int[][](n,m);

  //Getting a collumn is simple.
  int[] collumn0 = matrix[0];

  //Getting a row is trickier
  int[] row0;

  //Loop over all collumns
  foreach(i; matrix) {
    row0 ~= matrix[i][0];
  }
}

On Thursday, 14 November 2013 at 22:18:44 UTC, TheFlyingFiddle wrote:
> unittest
> {
>   int[][] matrix = new int[][](n,m);
>
>   //Getting a collumn is simple.
>   int[] collumn0 = matrix[0];
>
>   //Getting a row is trickier
>   int[] row0;
>
>   //Loop over all collumns
>   foreach(i; matrix) {
>     row0 ~= matrix[i][0];
>   }
> }

Sry the way you get rows is wrong.
It should be

int[] row0;
foreach(collumn; matrix)
    row0 ~= collumn[0];

and not

>   int[] row0;
>
>   //Loop over all collumns
>   foreach(i; matrix) {
>     row0 ~= matrix[i][0];
>   }

On 11/14/2013 02:18 PM, TheFlyingFiddle wrote:
> On Thursday, 14 November 2013 at 22:06:28 UTC, seany wrote:
>> How do I access All rows (resp. columns ) of a Matrix /
>> multidimensional array?
>>
>> In scilab, you write array_var(:,row_index), colon meaning all
>> columns, at that particular row given by row index will be selected.
>>
>> can you, forexample, in D, write, array_var[0][] to select all
>> elements (columns, in this case i believe?) of array_var[0] ?
>
> unittest
> {
>    int[][] matrix = new int[][](n,m);
>
>    //Getting a collumn is simple.
>    int[] collumn0 = matrix[0];
>
>    //Getting a row is trickier
>    int[] row0;
>
>    //Loop over all collumns
>    foreach(i; matrix) {
>      row0 ~= matrix[i][0];
>    }
> }
>
>

transversal() is useful too: :)

  http://dlang.org/phobos/std_range.html#transversal

Ali

On Thursday, 14 November 2013 at 22:26:03 UTC, Ali Çehreli wrote:
> transversal() is useful too: :)
>
>   http://dlang.org/phobos/std_range.html#transversal
>
> Ali

Did not know about this one. Thanks for pointing it out :D

Ali Çehreli:

> transversal() is useful too: :)
>
>   http://dlang.org/phobos/std_range.html#transversal

transversal() is sometimes useful, but if you need a significant amount of matrix processing, slicing and dicing matrices all the time in scientific code, then probably it's better to use a custom matrix library for D. Eventually one "winner" of such libraries should come out for D (as NumPy is in Python world). But to implement such libraries D should improve its management of "$" and "..." more for user-defined types.

Bye,
bearophile

On Fri, Nov 15, 2013 at 12:13:11AM +0100, bearophile wrote:
> Ali Çehreli:
> 
> >transversal() is useful too: :)
> >
> >  http://dlang.org/phobos/std_range.html#transversal
> 
> transversal() is sometimes useful, but if you need a significant amount of matrix processing, slicing and dicing matrices all the time in scientific code, then probably it's better to use a custom matrix library for D. Eventually one "winner" of such libraries should come out for D (as NumPy is in Python world). But to implement such libraries D should improve its management of "$" and "..." more for user-defined types.
[...]

'$' already works for multidimensional arrays, you just have to implement opDollar appropriately.

Slicing syntax isn't, though. Kenji has a pull for it, but it hasn't been merged yet for whatever reason.


T

-- 
Once bitten, twice cry...