feat: add math/base/special/exp2f

[aayush0325]

Progresses #649

Description

What is the purpose of this pull request?

This pull request:

• adds math/base/special/exp2f

Related Issues

Does this pull request have any related issues?

This pull request:

• progresses [RFC]: Add C implementations to base special math functions (tracking issue) #649

Questions

Any questions for reviewers of this pull request?

What should i do about benchmark/c/cephes ? that doesn't seem to compile locally and on the CI tests as well.

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.

---

@stdlib-js/reviewers

[aayush0325]

/stdlib update-copyright-years

[stdlib-bot]

Coverage Report

Package	Statements	Branches	Functions	Lines
math/base/special/exp2f	$\color{green}333/333$ $\color{green}+100.00\%$	$\color{green}21/21$ $\color{green}+100.00\%$	$\color{green}4/4$ $\color{green}+100.00\%$	$\color{green}333/333$ $\color{green}+100.00\%$

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

[Planeshifter]

Thanks for your work on this PR! Looks great overall, but some changes will be needed for the JavaScript implementation.

In main.js, we need to use the helper function

var float64ToFloat32 = require( '@stdlib/number/float64/base/to-float32' );

in order to ensure that all operations are done in a manner consistent with single-precision arithmetic. Recall that in JavaScript, all numbers are double-precision by default, so to emulate single-precision we must, in addition to using the single-precision variants of math functions, explicitly cast intermediate results and constants to single-precision floats.

[aayush0325]

good morning @Planeshifter! I've wrapped the intermediate results and final answers in float64ToFloat32(). If there's anything else that needs to be changed here please let me know.

[aayush0325]

Also, as mentioned in the questions section of the PR, what should i do about benchmark/c/cephes since it doesn't seem to compile locally and it fails on CI as well. Is there something that I'm missing from my side? (see this commit)

[gunjjoshi]

/stdlib update-copyright-years

[gunjjoshi]

@aayush0325 These values for domain, number of trials, peak and rms are for the double-precision implementation, i.e., exp2. You'll need to check the single-precision implementation and update these.

[gunjjoshi]

For instance, the domain for exp2f will be restricted to -127,+127, and not -1022,+1024.

[aayush0325]

https://www.netlib.org/cephes/singldoc.html#exp2 i hope this is what you're looking for!

[gunjjoshi]

Suggested change

	return float64ToFloat32( 0.0 );
	return 0.0;

[gunjjoshi]

0.0 can be returned directly here.

[gunjjoshi]

@aayush0325 After L110, you'll need to add a conditional statement to return 1.0 if x is 0.0. You might want to refer to the single-precision implementation (exp2f) from cephes, as it is a bit different from porting exp2 directly to single-precision.

[aayush0325]

https://github.com/jeremybarnes/cephes/blob/master/single/exp2f.c

using this as a reference here

[kgryte]

@aayush0325 You still interested in pushing this PR forward?

[aayush0325]

/stdlib merge

[aayush0325]

good morning @gunjjoshi! sorry for the delay, I've linked to all the references for your ease, kindly let me know if i have to make any other changes

[kgryte]

@anandkaranubc Would you mind taking over this PR?

[anandkaranubc]

/stdlib merge

[anandkaranubc]

@kgryte Does this comment (originally present in exp2) mean following https://github.com/freebsd/freebsd-src/blob/main/lib/msun/src/s_exp2f.c implementation instead?

[kgryte]

Yes, I believe so. The original implementation looks to be based on Cephes, and, generally, we have preferred to use libm over Cephes given that libm is more actively maintained.