[MeKaustubh07]

---

type: pre_commit_static_analysis_report
description: Results of running static analysis checks when committing changes. report:

• task: lint_filenames status: passed
• task: lint_editorconfig status: passed
• task: lint_markdown_pkg_readmes status: na
• task: lint_markdown_docs status: na
• task: lint_markdown status: na
• task: lint_package_json status: na
• task: lint_repl_help status: na
• task: lint_javascript_src status: passed
• task: lint_javascript_cli status: na
• task: lint_javascript_examples status: na
• task: lint_javascript_tests status: passed
• task: lint_javascript_benchmarks status: na
• task: lint_python status: na
• task: lint_r status: na
• task: lint_c_src status: na
• task: lint_c_examples status: na
• task: lint_c_benchmarks status: na
• task: lint_c_tests_fixtures status: na
• task: lint_shell status: na
• task: lint_typescript_declarations status: passed
• task: lint_typescript_tests status: na
• task: lint_license_headers status: passed ---

Resolves #{{TODO: add issue number}}.

Description

Resolves #{{issue number}}.

Description

What is the purpose of this pull request?

This pull request:

• fixes blas/base/dgemm, blas/base/sgemm, and blas/base/ggemm so that, when K is 0, the functions always return β*C, regardless of the value of α and of the memory layout of A and B.

The functions compute C = α*op(A)*op(B) + β*C, where each element of op(A)*op(B) is a dot product of length K. When K is 0, op(A) has no columns and op(B) has no rows, so the product contributes nothing and the result is β*C. Reference BLAS (dgemm.f) behaves this way: it scales C by β and then runs a loop over K, which does not execute when K is 0, so α is never used.

In lib/base.js, C is first scaled by β, after which the function returns early only when α is 0:

// Check whether we can early return...
if ( alpha === 0.0 ) {
	return C;
}

When K is 0 and α is nonzero, execution continues into one of two kernels, chosen based on the memory layouts of A and B:

• the loop-tiling kernel (blocked) iterates for ( k = K; k > 0; ), which does not run when K is 0, so C remains β*C (correct).
• the dot-product kernel (naive) computes C[ ic ] += alpha * ddot( K, ... ). For K = 0, ddot returns 0.0, and when α is NaN or ±Infinity, alpha * 0.0 is NaN, so every element of C becomes NaN (incorrect).

As a result, the same call returns different results depending only on memory layout:

var Float64Array = require( '@stdlib/array/float64' );
var dgemm = require( '@stdlib/blas/base/dgemm' );

var A = new Float64Array( 0 );
var B = new Float64Array( 0 );

var C1 = new Float64Array( [ 1.0, 2.0, 3.0, 4.0 ] );
dgemm( 'row-major', 'no-transpose', 'no-transpose', 2, 2, 0, Infinity, A, 1, B, 2, 0.5, C1, 2 );
// C1 => <Float64Array>[ 0.5, 1.0, 1.5, 2.0 ]

var C2 = new Float64Array( [ 1.0, 2.0, 3.0, 4.0 ] );
dgemm( 'row-major', 'no-transpose', 'transpose', 2, 2, 0, Infinity, A, 1, B, 1, 0.5, C2, 2 );
// C2 => <Float64Array>[ NaN, NaN, NaN, NaN ]

Both calls should return [ 0.5, 1.0, 1.5, 2.0 ].

Change

The early return now also covers K = 0, which, like α = 0, leaves nothing to add after scaling C by β:

// Check whether we can early return...
if ( alpha === 0.0 || K === 0 ) {
	return C;
}

The existing quick return for β = 1 (( alpha === 0.0 || K === 0 ) && beta === 1.0) is unchanged. The same change is applied to dgemm, sgemm, and ggemm (in both lib/base.js and lib/accessors.js), so it applies to both the main and ndarray APIs.

Tests

The existing tests only cover K = 0 together with β = 1. Added, for each package (and for ggemm with accessor arrays), tests that when K is 0 and α is NaN, ±Infinity, or finite, the functions return β*C (for β = 0.5) and a zero-filled matrix (for β = 0) for row-major no-transpose/no-transpose (loop-tiling kernel) and row-major no-transpose/transpose and column-major transpose/no-transpose (dot-product kernel). These tests fail before this change and pass after it.

Related Issues

Does this pull request have any related issues?

This pull request has the following related issues:

• #{{issue number}}

Reference Packages

Did the content of this pull request follow one or more existing packages? Answering is optional.

This pull request followed these packages:

• {{TODO: add package path (e.g., blas/ext/base/gxmy), or none, breaking new ground}}

What carried over from the reference(s)?

• {{TODO: add what carried over (e.g., package structure, naming, test shape, benchmark layout, and/or documentation)}}

What differed, or what did you have to work out yourself?

• {{TODO: add what you worked out without a reference}}

Questions

Any questions for reviewers of this pull request?

No.

Other

Any other information relevant to this pull request? This may include screenshots, references, and/or implementation notes.

No.

Checklist

Please ensure the following tasks are completed before submitting this pull request.

• Read, understood, and followed the contributing guidelines.

AI Assistance

When authoring the changes proposed in this PR, did you use any kind of AI assistance?

• Yes
• No

If you answered "yes" above, how did you use AI assistance?

• Code generation (e.g., when writing an implementation or fixing a bug)
• Test/benchmark generation
• Documentation (including examples)
• Research and understanding

Disclosure

If you answered "yes" to using AI assistance, please provide a short disclosure indicating how you used AI assistance. This helps reviewers determine how much scrutiny to apply when reviewing your contribution. Example disclosures: "This PR was written primarily by Claude Code." or "I consulted ChatGPT to understand the codebase, but the proposed changes were fully authored manually by myself.".

{{TODO: add disclosure if applicable}}

---

@stdlib-js/reviewers

[stdlib-bot]

Coverage Report

Package	Statements	Branches	Functions	Lines
blas/base/dgemm	$\\color{green}843/843$ $\\color{green}+0.00\\%$	$\\color{green}120/120$ $\\color{green}+0.00\\%$	$\\color{green}8/8$ $\\color{green}+0.00\\%$	$\\color{green}843/843$ $\\color{green}+0.00\\%$
blas/base/ggemm	$\\color{red}1335/1344$ $\\color{green}+0.00\\%$	$\\color{red}174/177$ $\\color{green}+0.00\\%$	$\\color{green}14/14$ $\\color{green}+0.00\\%$	$\\color{red}1335/1344$ $\\color{green}+0.00\\%$
blas/base/sgemm	$\\color{green}844/844$ $\\color{green}+0.00\\%$	$\\color{green}120/120$ $\\color{green}+0.00\\%$	$\\color{green}8/8$ $\\color{green}+0.00\\%$	$\\color{green}844/844$ $\\color{green}+0.00\\%$

The above coverage report was generated for the changes in this PR.

[kgryte]

Use pinf and ninf.

[kgryte]

Use pinf and ninf.

[kgryte]

Use pinf and ninf.

[kgryte]

Use pinf and ninf.

[kgryte]

Use pinf and ninf.

[kgryte]

Use pinf and ninf.

[kgryte]

Use pinf and ninf.

[kgryte]

Use pinf and ninf.

[kgryte]

Other than making sure to use the corresponding constants packages, this PR is looking good. Thanks for finding and fixing, @MeKaustubh07!