Bug: mod_switch_to_next throws transparent ciphertext exception when using CKKS

[s0l0ist]

Hello all,

An exception is thrown (erroneously) when attempting to perform a modulus switch on a ciphertext when using the overload accepting a destination parameter: evaluator.mod_switch_to_next

This appears to occur when using the CKKS scheme.

Example C++ code:

int main()
{
    std::cout << "Running main..." << std::endl;
    EncryptionParameters parms(scheme_type::CKKS);
    size_t poly_modulus_degree = 4096;
    parms.set_poly_modulus_degree(poly_modulus_degree);
    parms.set_coeff_modulus(CoeffModulus::BFVDefault(poly_modulus_degree));
    //parms.set_plain_modulus(786433);
    auto context = SEALContext::Create(parms);
    KeyGenerator keygen(context);
    PublicKey public_key = keygen.public_key();
    SecretKey secret_key = keygen.secret_key();
    CKKSEncoder ckks_encoder(context);
    Encryptor encryptor(context, public_key);
    Evaluator evaluator(context);
    Decryptor decryptor(context, secret_key);

    size_t slot_count = ckks_encoder.slot_count();
    vector<double> vect(slot_count, 3.3);

    Plaintext plain;
    ckks_encoder.encode(vect, pow(2.0, 24), plain);

    Ciphertext cipher, cipher_dest;
    encryptor.encrypt(plain, cipher);

    try {
        std::cout << "Attempting to switch modulus..." << std::endl;
        evaluator.mod_switch_to_next(cipher, cipher_dest); // <---- throws "result ciphertext is transparent"
        // evaluator.mod_switch_to_next_inplace(cipher); // <---- works

        // This code never gets executed unless the `inplace` method is used instead
        Plaintext plain_result;
        decryptor.decrypt(cipher_dest, plain_result);

        vector<double> vect_result;
        ckks_encoder.decode(plain_result, vect_result);

        std::cout << "Result [0]: " << vect_result[0] << std::endl;
    }
    catch (exception& e) {
        std::cout << "An exception occurred: " << e.what() << std::endl;
    }
    return 0;
}

Build settings:

cmake \
  -DSEAL_USE_CXX17=ON \
  -DSEAL_USE_INTRIN=OFF \
  -DSEAL_USE_ZLIB=ON \
  -DSEAL_USE_MSGSL=ON \
  -DSEAL_BUILD_EXAMPLES=OFF \
  -DSEAL_BUILD_TESTS=OFF \
  -DBUILD_SHARED_LIBS=OFF \
  -DCMAKE_BUILD_TYPE=Release \
  .

[kimlaine]

Thanks for reporting this immediately. There was a mistake in a variadic lambda function resolving argument to "wrong" type -- oops. While looking into this, I noticed another bug in util::IterTuple that I fixed as well. I'll put out 3.5.3 right away with these fixes.