This issue is for the task of testing the lowerings of ckks.relinearize end-to-end, via lowering the ops to LLVM and running a mul + relinearize on ciphertexts.
#2987 identified a bug in the implementation of basis conversion that this would have caught.
The first step should be scoping the work required to complete this, because I think there are still a few nontrivial gaps. For example, I don't think we have any ops (not to mention lowerings) for generating relinearization keys.
CC @crockeea @bencemali @AlexanderViand
This issue is for the task of testing the lowerings of
ckks.relinearizeend-to-end, via lowering the ops to LLVM and running a mul + relinearize on ciphertexts.#2987 identified a bug in the implementation of basis conversion that this would have caught.
The first step should be scoping the work required to complete this, because I think there are still a few nontrivial gaps. For example, I don't think we have any ops (not to mention lowerings) for generating relinearization keys.
CC @crockeea @bencemali @AlexanderViand