feat: add lapack/base/iladlr
#2854
Conversation
Pranavchiku
added
Feature
| var out; | ||
| var A = new Float64Array( [ 1.0, 2.0, 3.0, 4.0 ] ); | ||
|
|
||
| out = iladlr( 'row-major', 2, 2, A, 2 ); |
There was a problem hiding this comment.
| var out; | |
| var A = new Float64Array( [ 1.0, 2.0, 3.0, 4.0 ] ); | |
| out = iladlr( 'row-major', 2, 2, A, 2 ); | |
| var A = new Float64Array( [ 1.0, 2.0, 3.0, 4.0 ] ); | |
| var out = iladlr( 'row-major', 2, 2, A, 2 ); |
This needs to be fixed here and in other examples.
| ```javascript | ||
| var Float64Array = require( '@stdlib/array/float64' ); | ||
|
|
||
| // Initial an array. |
There was a problem hiding this comment.
| // Initial an array. | |
| // Initial an array: |
Nowhere do we end line comments with only a period. Needs to be fixed here and elsewhere.
| var A1 = new Float64Array( A0.buffer, A0.BYTES_PER_ELEMENT*1 ); // start at 2nd element | ||
|
|
||
| var out = iladlr( 'row-major', 2, 2, A1, 2 ); | ||
| // out => 1 |
There was a problem hiding this comment.
This isn't how we do return annotations for scalar values.
| var A = new Float64Array( [ 1.0, 2.0, 3.0, 4.0 ] ); | ||
| var out = iladlr( 'row-major', 2, 2, A, 2 ); | ||
| console.log( out ); |
There was a problem hiding this comment.
This example isn't doing anything "new". Perhaps, instead, use random/array/bernoulli with a low probability to generate almost empty arrays.
| var A = new Float64Array( [ 4.0, 3.0, 2.0, 4.0 ] ); | ||
|
|
||
| out = iladlr.ndarray( 2, 2, A, -2, -1, 3 ); | ||
| // returns 1 |
There was a problem hiding this comment.
Your examples all do the same thing. Here it is not obvious how the change in stride and offset have any bearing on the output.
In your examples, you should strive to make cause and effect more obvious.
| * var out; | ||
| * var A = new Float64Array( [ 1.0, 2.0, 3.0, 4.0 ] ); | ||
| * | ||
| * out = iladlr( 2, 2, A, 2, 1, 0 ); |
There was a problem hiding this comment.
| * var out; | |
| * var A = new Float64Array( [ 1.0, 2.0, 3.0, 4.0 ] ); | |
| * | |
| * out = iladlr( 2, 2, A, 2, 1, 0 ); | |
| * var A = new Float64Array( [ 1.0, 2.0, 3.0, 4.0 ] ); | |
| * | |
| * var out = iladlr( 2, 2, A, 2, 1, 0 ); |
Here and everywhere.
| if ( A[ offsetA + ( ( M - 1 ) * sa0 ) ] !== 0.0 || A[ offsetA + ( ( M - 1 ) * sa0 ) + ( ( N - 1 ) * sa1 ) ] ) { | ||
| return M - 1; | ||
| } | ||
| // Scan up each column tracking the last zero row seen. |
There was a problem hiding this comment.
| // Scan up each column tracking the last zero row seen. | |
| // Scan up each column tracking the last zero row seen... |
There was a problem hiding this comment.
Also, for row-major matrices, this comment doesn't make much sense. I would argue that row-major matrices deserve their own set of nested loops, as the column-major algorithm is not efficient. Just think about the problem. You want to find a row with at least one non-zero element. That should be trivial to implement in row-major.
There was a problem hiding this comment.
Accordingly, I would refactor and remove the attempt at loop interchange.
| var i; | ||
| var j; | ||
|
|
||
| if ( isRowMajor( [ strideA1, strideA2 ] ) ) { |
There was a problem hiding this comment.
If the goal is loop interchange, you need to also swap dimensions.
kgryte
added
the
Needs Changes
Towards #2464.
Description
This pull request adds JS implementation for
lapack/base/iladlr.Related Issues
NA
Questions
No.
Other
No.
Checklist
@stdlib-js/reviewers