feat: add `fft/rfftf` by gunjjoshi · Pull Request #16089 · stdlib-js/stdlib

gunjjoshi · 2026-10-12T18:31:33Z

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: passed
task: lint_markdown_docs status: na
task: lint_markdown status: na
task: lint_package_json status: passed
task: lint_repl_help status: passed
task: lint_javascript_src status: passed
task: lint_javascript_cli status: na
task: lint_javascript_examples status: passed
task: lint_javascript_tests status: passed
task: lint_javascript_benchmarks status: passed
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: passed
task: lint_license_headers status: passed ---

Resolves stdlib-js/metr-issue-tracker#1385.

Description

What is the purpose of this pull request?

This pull request:

adds fft/rfftf.

Related Issues

Does this pull request have any related issues?

This pull request has the following related issues:

[RFC]: add fft/rfftf metr-issue-tracker#1385.

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.".

Used Claude Code to help with the implementation, tests, and docs. Also used it to understand multiple things and do the necessary research work.

@stdlib-js/reviewers

--- 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: passed - task: lint_markdown_docs status: na - task: lint_markdown status: na - task: lint_package_json status: passed - task: lint_repl_help status: passed - task: lint_javascript_src status: passed - task: lint_javascript_cli status: na - task: lint_javascript_examples status: passed - task: lint_javascript_tests status: passed - task: lint_javascript_benchmarks status: passed - 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: passed - task: lint_license_headers status: passed ---

gunjjoshi · 2026-10-12T18:34:16Z

+
+// VARIABLES //
+
+var idtypes = dtypes( 'real_and_generic' );


Used real_and_generic here instead of real_floating_point_and_generic, so that it can handle integer arrays too.

gunjjoshi · 2026-10-12T18:35:21Z

+/**
+* Computes the forward FFT of a real-valued input.
+*
+* @private
+* @param {ArrayLikeObject<Object>} arrays - array-like object containing ndarrays
+* @returns {void}
+*/
+function gkernel( arrays ) {
+	kernel( reinterpret, grffti, grfftf, Float64Array, arrays );
+}
+
+/**
+* Computes the forward FFT of a real-valued double-precision input.
+*
+* @private
+* @param {ArrayLikeObject<Object>} arrays - array-like object containing ndarrays
+* @returns {void}
+*/
+function dkernel( arrays ) {
+	kernel( reinterpret128, drffti, drfftf, Float64Array, arrays );
+}
+
+/**
+* Computes the forward FFT of a real-valued single-precision input.
+*
+* @private
+* @param {ArrayLikeObject<Object>} arrays - array-like object containing ndarrays
+* @returns {void}
+*/
+function skernel( arrays ) {
+	kernel( reinterpret64, srffti, srfftf, Float32Array, arrays );
+}


Needed these wrappers to wrap our strided kernels, since we have a custom logic here, to avoid an explicit hc2c conversion.

gunjjoshi · 2026-10-12T18:37:14Z

+	// Copy the input into the output buffer at offset 1:
+	gcopy( N, getData( x ), getStride( x, 0 ), getOffset( x ), view, 1, 1 );
+
+	// Initialize the workspace and compute the forward FFT in-place:
+	w = new Workspace( ( 2*N ) + 34 );
+	rffti( N, w, 1, 0 );
+	rfftf( N, view, 1, 1, w, 1, 0 );
+
+	// Move the DC component from the imaginary slot to the real slot of the first complex element and zero out the imaginary part:
+	view[ 0 ] = view[ 1 ];
+	view[ 1 ] = 0.0;
+
+	// For even-length sequences, zero out the imaginary part of the Nyquist component:
+	if ( isEven( N ) ) {
+		view[ ( 2*M ) - 1 ] = 0.0;
+	}
+}


This is the approach that we discussed in #fft > hc2c @ 💬.

stdlib-bot · 2026-10-12T18:40:58Z

Coverage Report

Package	Statements	Branches	Functions	Lines
fft/rfftf	$\\color{red}628/645$ $\\color{green}+97.36\\%$	$\\color{red}46/53$ $\\color{green}+86.79\\%$	$\\color{red}5/6$ $\\color{green}+83.33\\%$	$\\color{red}628/645$ $\\color{green}+97.36\\%$

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

kgryte · 2026-10-13T10:59:21Z

+		return;
+	}
+	// Copy the input into the output buffer at offset 1:
+	gcopy( N, getData( x ), getStride( x, 0 ), getOffset( x ), view, 1, 1 );


I am thinking you should pass in one of dcopy, scopy, or gcopy here. The former can be significantly faster. Should be straightforward to update your kernel wrappers below.

kgryte · 2026-10-13T11:04:51Z

+* var out = new ndarray( 'complex128', new Complex128Array( 3 ), [ 3 ], [ 1 ], 0, 'row-major' );
+*
+* // Perform operation:
+* var v = rfftf( out, x );


This call signature seems backward. Why is out first and x second?

Okay. I see the issue. The main thing is that you are thinking that you need to provide a single kernel which combines both initialization AND computation. That isn't the case.

And importantly, you are missing one of the very advantages of the workspace array. Notice above that you are creating a new workspace for each subarray. That is unnecessary and you forgot the guidance when porting FTTPACK. Namely, for constant length N, you can reuse the workspace array for each computed FFT.

Which means, in main.js, assign.js, or similar, what you do is you allocate a single workspace array and initialize once. That workspace array should then become an auxiliary array which is passed down with the subarray view to the ndarray wrapper, using nullary-*.

So, in short, you need to refactor this implementation.

kgryte · 2026-10-13T11:16:52Z

+	if ( n !== N ) {
+		if ( n < N ) {
+			// Create a view of the first n elements:
+			x = sliceTo( x, [ n ], false, false );
+		} else {
+			// Create a new array of length n and copy the first N elements:
+			y = baseZeros( getDType( x ), [ n ], getOrder( x ) );
+			gcopy( N, getData( x ), getStride( x, 0 ), getOffset( x ), getData( y ), 1, 0 ); // eslint-disable-line max-len
+			x = y;
+		}
+	}


We need to discuss this, as this implementation is not correct.

You are assuming a one-dimensional input ndarray. This is not correct. A user should be able to specify a dim (default: -1), meaning that a user can batch an FFT along one or more tensor fibers.

kgryte · 2026-10-13T11:17:32Z

+		} else {
+			s = 1.0 / n;
+		}
+		gscal( 2*M, s, view, 1, 0 );


This isn't going to work, as you are assuming a 1D input ndarray. We need to add blas/scal.

kgryte

Left initial comments.

gunjjoshi requested a review from a team October 12, 2026 18:31

gunjjoshi added the METR Pull request associated with the METR project. label Oct 12, 2026

Merge branch 'develop' into fft-rfftf

6d8ab46

gunjjoshi commented Oct 12, 2026

View reviewed changes

stdlib-bot added the Needs Review A pull request which needs code review. label Oct 12, 2026

kgryte added the Feature Issue or pull request for adding a new feature. label Oct 13, 2026

kgryte reviewed Oct 13, 2026

View reviewed changes

kgryte requested changes Oct 13, 2026

View reviewed changes

kgryte added Needs Changes Pull request which needs changes before being merged. and removed Needs Review A pull request which needs code review. labels Oct 13, 2026

Navigation Menu

Repository navigation

feat: add `fft/rfftf` - #16089

feat: add `fft/rfftf`#16089
gunjjoshi wants to merge 2 commits into
stdlib-js:developfrom
gunjjoshi:fft-rfftf

gunjjoshi commented Oct 12, 2026

Uh oh!

gunjjoshi Oct 12, 2026

Uh oh!

gunjjoshi Oct 12, 2026

Uh oh!

gunjjoshi Oct 12, 2026

Uh oh!

stdlib-bot commented Oct 12, 2026 •

edited

Loading

Uh oh!

kgryte Oct 13, 2026

Uh oh!

kgryte Oct 13, 2026

Uh oh!

kgryte Oct 13, 2026

Uh oh!

kgryte Oct 13, 2026

Uh oh!

kgryte Oct 13, 2026

Uh oh!

kgryte left a comment

Uh oh!

This branch has not been deployed

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Conversation

gunjjoshi commented Oct 12, 2026

Description

Related Issues

Questions

Other

Checklist

AI Assistance

Disclosure

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

stdlib-bot commented Oct 12, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Coverage Report

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

kgryte left a comment

Choose a reason for hiding this comment

Uh oh!

This branch has not been deployed

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

stdlib-bot commented Oct 12, 2026 •

edited

Loading