add `Div` and `Mul` for `Complex<{float}>` by folkertdev · Pull Request #162832 · rust-lang/rust

folkertdev · 2026-09-15T22:54:08Z

tracking issue: #154023

There are no fallbacks for the libcalls, so Miri will currently fail. The fallback is blocked on rust-lang/libs-team#876.

folkertdev · 2026-09-16T08:56:02Z

@bors try jobs=test-various,test-aarch64-apple-,-gnu-nopt-,test-x86_64-mingw-,test-aarch64-msvc-*,test-arm-android

add `Div` and `Mul` for `Complex<{float}>` try-job: test-various try-job: test-aarch64-apple-* try-job: *-gnu-nopt-* try-job: test-x86_64-mingw-* try-job: test-aarch64-msvc-* try-job: test-arm-android

rust-bors · 2026-09-16T12:09:13Z

☀️ Try build successful (CI)
Build commit: 3b59d93 (3b59d930efb08e53dc535d0e705b2564a402ce9b)
Base parent: 28e8a8c (28e8a8c81bf3b37909edac6c2a76e56f30cd492f)

folkertdev · 2026-09-16T14:32:00Z

@RalfJung I'm a bit unsure on how to get Miri to support this. Is this a case where cfg(miri) is OK?

We can add a fn complex_mul<F: Float>(mut a: F, mut b: F, mut c: F, mut d: F) -> Complex<F> implementation based on crate::num::imp::Float and only call it when cfg(miri).

if unlikely(z.re.is_nan() && z.im.is_nan()) {
    cfg_select! {
        miri => complex_mul(a, b, c, d), // use an implementation in core
        _ => $mul(a, b, c, d), // call the libcall
    }
} else {
    z
}

I know you dislike using cfg(miri) but I don't see other good options here.

RalfJung · 2026-09-16T14:48:19Z

Typically we just implement the libcalls in Miri.

folkertdev · 2026-09-16T15:01:20Z

Ok, that should work. We can merge this then and I'll work on a Miri implementation in parallel.

r? tgross35 or @beetrees

rustbot · 2026-09-16T15:01:25Z

tgross35 is currently at their maximum review capacity.
They may take a while to respond.

RalfJung · 2026-09-16T15:08:27Z

@bors try jobs=test-x86_64-gnu-aux

add `Div` and `Mul` for `Complex<{float}>` try-job: test-x86_64-gnu-aux

rust-bors · 2026-09-16T17:46:53Z

☀️ Try build successful (CI)
Build commit: f9b288f (f9b288f412d8bf91cb68eaee520f73e5bdb9410e)
Base parent: e15cecc (e15ceccfc6209c15b6c4bc6352f6ec6bfe579eaa)

tgross35 · 2026-09-17T04:13:50Z

+            #[inline]
+            fn mul(self, rhs: Self) -> Self::Output {
+                let Complex { re: a, im: b } = self;
+                let Complex { re: c, im: d } = rhs;
+
+                let ac = a * c;
+                let bd = b * d;
+                let ad = a * d;
+                let bc = b * c;
+
+                let z = Complex::new(ac - bd, ad + bc);
+
+                // Only call the libcall when both components are NaN.
+                if unlikely(z.re.is_nan() && z.im.is_nan()) { $mul(a, b, c, d) } else { z }
+            }


The comment says pretty much the same thing as the code, it would be helpful to explain why. Also mild preference for cold_path over unlikely since it's stable.

I think the optimization may pay off but it's unfortunate it means repeated work. Not too bad for hard floats but this means e.g. 12 additional function calls if you hit a NaN with f16. I wonder if there's a case to be made that these (and possibly other) libcalls should be preserve_most or preserve_all since stashing the registers is likely the most expensive part of a call.

View changes since the review

I wrote something, and now usecold_path. I'm not sure how much it helps given that the branch is small, but it also serves as documentation that this is an unlikely branch.

And, this ABI etc was designed way before f16 entered the picture, I don't think this will matter much in practice.

I did mean more of a comment about the optimization going on here, i.e. why are we doing some things inline rather than going directly to the libcall

tgross35 · 2026-09-17T04:17:52Z

The fallback is blocked on rust-lang/libs-team#876.

Also I don't think this is blocked on anything, the same mul/div algorithms could be implemented in apfloat for CTFE/miri.

@nia-e

Complex conjugate, negation and default tracking issue: rust-lang#154023 Add several more straightforward (I hope) impls and methods to `Complex`. - `Default` when `T: Default`. For numbers this means `Complex::default` is the unit element with respect to complex addition - `Neg` when `T: Neg`. Maps `a + bi` to `-a - bi`, which can underflow on integers (but this is no different from addition being able to overflow/underflow too - `conjugate` maps `a + bi` to `a - bi` - make `Complex::new` a `const fn`, because why not Complex division and multiplication is in progress in rust-lang#162832. r? @nia-e for continuity (or `r? libs` if you're busy)

@nia-e

Complex conjugate, negation and default tracking issue: rust-lang#154023 Add several more straightforward (I hope) impls and methods to `Complex`. - `Default` when `T: Default`. For numbers this means `Complex::default` is the unit element with respect to complex addition - `Neg` when `T: Neg`. Maps `a + bi` to `-a - bi`, which can underflow on integers (but this is no different from addition being able to overflow/underflow too - `conjugate` maps `a + bi` to `a - bi` - make `Complex::new` a `const fn`, because why not Complex division and multiplication is in progress in rust-lang#162832. r? @nia-e for continuity (or `r? libs` if you're busy)

folkertdev

I have a Miri PR ready to go, but to test it I need the instances to work first.

We can move it into CMSE later, right now there isn't much point to it I think because many other things do not support const.

View changes since this review

@nia-e

Rollup merge of #162865 - folkertdev:complex-conjugate, r=nia-e Complex conjugate, negation and default tracking issue: #154023 Add several more straightforward (I hope) impls and methods to `Complex`. - `Default` when `T: Default`. For numbers this means `Complex::default` is the unit element with respect to complex addition - `Neg` when `T: Neg`. Maps `a + bi` to `-a - bi`, which can underflow on integers (but this is no different from addition being able to overflow/underflow too - `conjugate` maps `a + bi` to `a - bi` - make `Complex::new` a `const fn`, because why not Complex division and multiplication is in progress in #162832. r? @nia-e for continuity (or `r? libs` if you're busy)

tgross35

r=me with a rebase and optional comment update. Make sure to update the tracking issue API summary, probably going to get easy to lose track

View changes since this review

tgross35 · 2026-09-17T19:24:38Z

+            #[inline]
+            fn mul(self, rhs: Self) -> Self::Output {
+                let Complex { re: a, im: b } = self;
+                let Complex { re: c, im: d } = rhs;
+
+                let ac = a * c;
+                let bd = b * d;
+                let ad = a * d;
+                let bc = b * c;
+
+                let z = Complex::new(ac - bd, ad + bc);
+
+                // Only call the libcall when both components are NaN.
+                if unlikely(z.re.is_nan() && z.im.is_nan()) { $mul(a, b, c, d) } else { z }
+            }


I did mean more of a comment about the optimization going on here, i.e. why are we doing some things inline rather than going directly to the libcall

rustbot · 2026-09-17T20:13:07Z

This PR was rebased onto a different main commit. Here's a range-diff highlighting what actually changed.

Rebasing is a normal part of keeping PRs up to date, so no action is needed—this note is just to help reviewers.

folkertdev added the F-complex_numbers `#![feature(complex_numbers)]` label Sep 15, 2026

rustbot added S-waiting-on-author Status: This is awaiting some action (such as code changes or more information) from the author. T-libs Relevant to the library team, which will review and decide on the PR/issue. labels Sep 15, 2026

This comment has been minimized.

Sign in to view

rustbot assigned tgross35 Sep 16, 2026

folkertdev marked this pull request as ready for review September 16, 2026 15:01

rustbot added S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. and removed S-waiting-on-author Status: This is awaiting some action (such as code changes or more information) from the author. labels Sep 16, 2026

This comment has been minimized.

Sign in to view

rust-bors Bot pushed a commit that referenced this pull request Sep 16, 2026

Auto merge of #162832 - folkertdev:complex-mul-div, r=<try>

f9b288f

add `Div` and `Mul` for `Complex<{float}>` try-job: test-x86_64-gnu-aux

folkertdev mentioned this pull request Sep 16, 2026

Complex conjugate, negation and default #162865

Merged

tgross35 reviewed Sep 17, 2026

View reviewed changes

folkertdev commented Sep 17, 2026

View reviewed changes

folkertdev force-pushed the complex-mul-div branch from ac91627 to 381f005 Compare September 17, 2026 18:36

This comment has been minimized.

Sign in to view

tgross35 approved these changes Sep 17, 2026

View reviewed changes

This comment has been minimized.

Sign in to view

tgross35 mentioned this pull request Sep 17, 2026

rustdoc-gui tests sporadically failing #93784

Open

add Div and Mul for Complex<{float}>

ffaef7f

folkertdev force-pushed the complex-mul-div branch from 381f005 to ffaef7f Compare September 17, 2026 20:13

folkertdev mentioned this pull request Sep 17, 2026

implement complex div/mul libcalls rust-lang/miri#5319

Draft

Uh oh!

Conversation

folkertdev commented Sep 15, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

folkertdev commented Sep 16, 2026

Uh oh!

This comment has been minimized.

rust-bors Bot commented Sep 16, 2026

Uh oh!

folkertdev commented Sep 16, 2026

Uh oh!

RalfJung commented Sep 16, 2026

Uh oh!

folkertdev commented Sep 16, 2026

Uh oh!

rustbot commented Sep 16, 2026

Uh oh!

RalfJung commented Sep 16, 2026

Uh oh!

This comment has been minimized.

rust-bors Bot commented Sep 16, 2026

Uh oh!

tgross35 Sep 17, 2026 • edited by rustbot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

folkertdev Sep 17, 2026

Choose a reason for hiding this comment

Uh oh!

tgross35 Sep 17, 2026

Choose a reason for hiding this comment

Uh oh!

tgross35 commented Sep 17, 2026

Uh oh!

folkertdev left a comment • edited by rustbot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

This comment has been minimized.

tgross35 left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

tgross35 Sep 17, 2026

Choose a reason for hiding this comment

Uh oh!

This comment has been minimized.

rustbot commented Sep 17, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

folkertdev commented Sep 15, 2026 •

edited

Loading

tgross35 Sep 17, 2026 •

edited by rustbot

Loading

folkertdev left a comment •

edited by rustbot

Loading

tgross35 left a comment •

edited

Loading