feat: add math/base/special/boxcoxf

[navyansh007]

Resolves #10149

Description

What is the purpose of this pull request?

This pull request:

• Adds @stdlib/math/base/special/boxcoxf, a single-precision floating-point one-parameter Box-Cox transformation
• Includes C implementation, JavaScript implementation, native addon bindings, tests, benchmarks, and documentation

Related Issues

Does this pull request have any related issues?

This pull request has the following related issues:

• [RFC]: Add C implementations to base special math functions (tracking issue) #649
• [RFC]: Add implementation for @stdlib/math/base/special/boxcoxf #10149

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.".

---

@stdlib-js/reviewers

[stdlib-bot]

Coverage Report

Package	Statements	Branches	Functions	Lines
math/base/special/boxcoxf	$\color{green}205/205$ $\color{green}+0.00%$	$\color{green}13/13$ $\color{green}+0.00%$	$\color{green}2/2$ $\color{green}+0.00%$	$\color{green}205/205$ $\color{green}+0.00%$

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

[gunjjoshi]

@navyansh007 Any reasons to use double-precision functions here instead of single-precision ones?

Suggested change

	var abs = require( '@stdlib/math/base/special/abs' );
	var isnan = require( '@stdlib/math/base/assert/is-nan' );
	var isPositiveZero = require( '@stdlib/math/base/assert/is-positive-zero' );
	var lnf = require( '@stdlib/math/base/special/lnf' );
	var expm1 = require( '@stdlib/math/base/special/expm1' );
	var absf = require( '@stdlib/math/base/special/absf' );
	var isnanf = require( '@stdlib/math/base/assert/is-nanf' );
	var isPositiveZerof = require( '@stdlib/math/base/assert/is-positive-zerof' );
	var lnf = require( '@stdlib/math/base/special/lnf' );
	var expm1f = require( '@stdlib/math/base/special/expm1f' );

[navyansh007]

Thanks @gunjjoshi. My mistake- I copied the imports directly from math/base/special/boxcox/lib/main.js without switching them to the single-precision equivalents. I'll update abs, isnan, and isPositiveZero to absf, isnanf, and isPositiveZerof.
However, expm1f doesn't currently exist in stdlib, so I'm keeping the double-precision expm1 and casting the result to float32- this matches the approach used in the C implementation as well.

[gunjjoshi]

We'll need to wait until expm1f is implemented. Once we have that, we can come back to this PR.

[gunjjoshi]

Suggested change

	* Computes a one-parameter Box-Cox transformation for single-precision floating-point number.
	* Computes a one-parameter Box-Cox transformation for a single-precision floating-point number.

[gunjjoshi]

@navyansh007 We are missing the "Method" description here, which is present in boxcox

[navyansh007]

I intentionally left out the Method section because the one in boxcox is specifically a numerical analysis for double-precision (float64). It derives the threshold 1.0e-19 from the float64 epsilon and the range of ln for double-precision numbers. Copying it as-is didn't seem correct to me for boxcoxf, since float32 has a larger epsilon and a different log range, which is why the threshold here is 1.0e-10 instead. I'd be happy to add a Method section with the corresponding float32 analysis. Would you be able to point me in the right direction on how to best frame that, or should I follow the same derivation approach as boxcox but with single-precision constants?

[gunjjoshi]

Yes, we will definitely need to frame it for single-precision. You should be able to derive it using single-precision constants. You can do something like:

In [24]: var FLOAT32_EXPONENT_BIAS = require( '@stdlib/constants/float32/exponent-bias' );

In [25]: var FLOAT32_NUM_SIGNIFICAND_BITS = require( '@stdlib/constants/float32/num-significand-bits' );

In [26]: var powf = require('@stdlib/math/base/special/powf')
Out[26]: [Function: powf]

In [27]: var lnf = require( '@stdlib/math/base/special/lnf' );

In [28]: var ln_max = lnf( 2 - powf( 2, -FLOAT32_NUM_SIGNIFICAND_BITS ) ) + ( FLOAT32_EXPONENT_BIAS ) * lnf( 2 )
Out[28]: 88.7228392958641

In [29]: var ln_min = ( 1 - FLOAT32_EXPONENT_BIAS - FLOAT32_NUM_SIGNIFICAND_BITS ) * lnf( 2 )
Out[29]: -103.27893018722534

And something similar for double-precision.

[gunjjoshi]

Suggested change

	"description": "Compute a one-parameter Box-Cox transformation for single-precision floating-point number.",
	"description": "Compute a one-parameter Box-Cox transformation for a single-precision floating-point number.",

[gunjjoshi]

Suggested change

	> Compute a one-parameter [Box-Cox transformation][box-cox-transformation] for single-precision floating-point number.
	> Compute a one-parameter [Box-Cox transformation][box-cox-transformation] for a single-precision floating-point number.

Let's use 'a' here and everywhere else in this PR too.

[gunjjoshi]

Suggested change

	float r;
	double t;
	double t;
	float r;