Hello,
I noticed there are currently no explicit KeyGen test vectors for ML-DSA that use a seed, which differs from how ML-KEM test vectors are structured.
Right now, in the ML-DSA sign_seed test groups, the vectors provide a privateSeed. To run these signature tests, our harness has to implicitly execute the deterministic Key Generation algorithm to expand this seed into the private key (sk), and then use that sk to run the signing tests. The public key (pk) generated during this step is effectively ignored.
While this implicitly tests the KeyGen logic (since an incorrectly generated key would result in mismatched signatures downstream), having explicit KeyGen test vectors would allow for direct, isolated validation of both the generated pk and sk against expected hex outputs.
Are there any plans to add explicit KeyGen test vectors for ML-DSA in the future?
Thank you!
Hello,
I noticed there are currently no explicit KeyGen test vectors for ML-DSA that use a seed, which differs from how ML-KEM test vectors are structured.
Right now, in the ML-DSA sign_seed test groups, the vectors provide a privateSeed. To run these signature tests, our harness has to implicitly execute the deterministic Key Generation algorithm to expand this seed into the private key (sk), and then use that sk to run the signing tests. The public key (pk) generated during this step is effectively ignored.
While this implicitly tests the KeyGen logic (since an incorrectly generated key would result in mismatched signatures downstream), having explicit KeyGen test vectors would allow for direct, isolated validation of both the generated pk and sk against expected hex outputs.
Are there any plans to add explicit KeyGen test vectors for ML-DSA in the future?
Thank you!