ml-kem: use NttPolynomial from module-lattice#210
Merged
Conversation
It seems `module-lattice` has a `Mul` impl for `NttPolynomial` that implements ML-DSA's `MultiplyNTT` which seems incompatible with the one in `ml-kem` that defines its own `MultiplyNTTs` (Algorithm 11) which is used here instead. It perhaps suggests that code should be factored into `ml-dsa`.
tarcieri
added a commit
that referenced
this pull request
Jan 29, 2026
Adds `MultiplyNtt` to `module-lattice`, and `Ntt` and `NttInverse` to `ml-kem`. Previously `module-lattice` defined the `Mul` impl on `NttPolynomial` in terms of Algorithm 45 from FIPS 204, and this is used by the `Mul` impls on `NttVector` and `NttMatrix` as well. For ML-KEM we instead need to plug in Algorithm 11 from FIPS 203 (MultiplyNTTs), so we need a trait that lets each crate define the `Mul` impl on `NttPolynomial`. This adds a `MultiplyNtt` trait to `module-lattice`, currently defined on a `Field` (perhaps not great but it's the generic parameter we have), which allows each construction to plug in their own NTT multiplication algorithm. The existing implementation of Algorithm 45 from FIPS 204 will need to move to the `ml-dsa` crate. Also, following the structure of the `ml-dsa` crate, this defines two same-shaped traits, `Ntt` and `NttInverse`, that can be impl'd on types from `module-lattice` to use the same method syntax that was being used previously before #210. Unfortunately we can't share traits with `ml-dsa`, because the only way we can impl these traits for types from `module-lattice` is if we define them. But they're small and there aren't that many of them.
tarcieri
added a commit
that referenced
this pull request
Jan 29, 2026
Adds `MultiplyNtt` to `module-lattice`, and `Ntt` and `NttInverse` to `ml-kem`. Previously `module-lattice` defined the `Mul` impl on `NttPolynomial` in terms of Algorithm 45 from FIPS 204, and this is used by the `Mul` impls on `NttVector` and `NttMatrix` as well. For ML-KEM we instead need to plug in Algorithm 11 from FIPS 203 (MultiplyNTTs), so we need a trait that lets each crate define the `Mul` impl on `NttPolynomial`. This adds a `MultiplyNtt` trait to `module-lattice`, currently defined on a `Field` (perhaps not great but it's the generic parameter we have), which allows each construction to plug in their own NTT multiplication algorithm. The existing implementation of Algorithm 45 from FIPS 204 will need to move to the `ml-dsa` crate. Also, following the structure of the `ml-dsa` crate, this defines two same-shaped traits, `Ntt` and `NttInverse`, that can be impl'd on types from `module-lattice` to use the same method syntax that was being used previously before #210. Unfortunately we can't share traits with `ml-dsa`, because the only way we can impl these traits for types from `module-lattice` is if we define them. But they're small and there aren't that many of them.
tarcieri
added a commit
that referenced
this pull request
Jan 29, 2026
Adds `MultiplyNtt` to `module-lattice`, and `Ntt` and `NttInverse` to `ml-kem`. Previously `module-lattice` defined the `Mul` impl on `NttPolynomial` in terms of Algorithm 45 from FIPS 204, and this is used by the `Mul` impls on `NttVector` and `NttMatrix` as well. For ML-KEM we instead need to plug in Algorithm 11 from FIPS 203 (MultiplyNTTs), so we need a trait that lets each crate define the `Mul` impl on `NttPolynomial`. This adds a `MultiplyNtt` trait to `module-lattice`, currently defined on a `Field` (perhaps not great but it's the generic parameter we have), which allows each construction to plug in their own NTT multiplication algorithm. The existing implementation of Algorithm 45 from FIPS 204 will need to move to the `ml-dsa` crate. Also, following the structure of the `ml-dsa` crate, this defines two same-shaped traits, `Ntt` and `NttInverse`, that can be impl'd on types from `module-lattice` to use the same method syntax that was being used previously before #210. Unfortunately we can't share traits with `ml-dsa`, because the only way we can impl these traits for types from `module-lattice` is if we define them. But they're small and there aren't that many of them.
tarcieri
added a commit
that referenced
this pull request
Jan 29, 2026
Adds `MultiplyNtt` to `module-lattice`, and `Ntt` and `NttInverse` to `ml-kem`. Previously `module-lattice` defined the `Mul` impl on `NttPolynomial` in terms of Algorithm 45 from FIPS 204, and this is used by the `Mul` impls on `NttVector` and `NttMatrix` as well. For ML-KEM we instead need to plug in Algorithm 11 from FIPS 203 (MultiplyNTTs), so we need a trait that lets each crate define the `Mul` impl on `NttPolynomial`. This adds a `MultiplyNtt` trait to `module-lattice`, currently defined on a `Field` (perhaps not great but it's the generic parameter we have), which allows each construction to plug in their own NTT multiplication algorithm. The existing implementation of Algorithm 45 from FIPS 204 will need to move to the `ml-dsa` crate. Also, following the structure of the `ml-dsa` crate, this defines two same-shaped traits, `Ntt` and `NttInverse`, that can be impl'd on types from `module-lattice` to use the same method syntax that was being used previously before #210. Unfortunately we can't share traits with `ml-dsa`, because the only way we can impl these traits for types from `module-lattice` is if we define them. But they're small and there aren't that many of them.
tarcieri
added a commit
that referenced
this pull request
Jan 29, 2026
Adds `MultiplyNtt` to `module-lattice`, and `Ntt` and `NttInverse` to `ml-kem`. Previously `module-lattice` defined the `Mul` impl on `NttPolynomial` in terms of Algorithm 45 from FIPS 204, and this is used by the `Mul` impls on `NttVector` and `NttMatrix` as well. For ML-KEM we instead need to plug in Algorithm 11 from FIPS 203 (MultiplyNTTs), so we need a trait that lets each crate define the `Mul` impl on `NttPolynomial`. This adds a `MultiplyNtt` trait to `module-lattice`, currently defined on a `Field` (perhaps not great but it's the generic parameter we have), which allows each construction to plug in their own NTT multiplication algorithm. The existing implementation of Algorithm 45 from FIPS 204 will need to move to the `ml-dsa` crate. Also, following the structure of the `ml-dsa` crate, this defines two same-shaped traits, `Ntt` and `NttInverse`, that can be impl'd on types from `module-lattice` to use the same method syntax that was being used previously before #210. Unfortunately we can't share traits with `ml-dsa`, because the only way we can impl these traits for types from `module-lattice` is if we define them. But they're small and there aren't that many of them.
Merged
tarcieri
added a commit
that referenced
this pull request
Apr 28, 2026
## Added - `Seed` support e.g. `DecapsulationKey::from_seed` (#133, #138) - PKCS#8 support (#135) - `KeyInit`, `KeySizeUser`, and `KeyExport` impls for decapsulation keys (#156, #228) - Parameter set modules: `ml_kem_512`, `mk_kem_768`, `mk_kem_1024` (#162) - `DecapsulationKey::from_expanded` deprecated compatibility support (#163) - `TryKeyInit` and `KeyExport` impls for encapsulation keys (#188) - Validations against Wycheproof test vectors (#213, #214, #215, #217) - Implement `kem::Kem` trait (#223) - Support for `kem::FromSeed` trait (#255) ## Changed - Edition changed to 2024 and MSRV bumped to 1.85 (#118) - Relax MSRV policy and allow MSRV bumps in patch releases - Upgrade `hybrid-array` dependency to 0.4 (#129) - Extract `module-lattice` crate (#199, #202, #204, #209, #210, #211, #212, #218, #219, #220) - Replace `EncodedSizeUser` with `ExpandedKeyEncoding` (#226) - Bump `getrandom` to v0.4 (#245) - Bump `rand_core` to v0.10 (#245) - Migrate from `subtle` to `ctutils` (#277) - Bump `sha3` dependency to v0.11 (#282) - Bump `kem` dependency to v0.3 (#283) - Bump `pkcs8` dependency to v0.11 (#291) ## Fixed - Validate encryption/encapsulation keys (#179) - Validate expanded decapsulation key hash (#207) ## Removed - `Kem` struct and `KemCore` trait - replaced by `kem::Kem` (#223)
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
It seems
module-latticehas aMulimpl forNttPolynomialthat implements ML-DSA'sMultiplyNTTwhich seems incompatible with the one inml-kemthat defines its ownMultiplyNTTs(Algorithm 11) which is used here instead.It perhaps suggests that code should be factored into
ml-dsa.