Wrong values for multiplication and mod of DCRTPoly

Library Questions
1

Hello, maybe I misunderstand how the library works. But this code gives me a non zero polynomial:

int pp = 17;
DCRTPoly gg(dgg, paramsPK, Format::EVALUATION);
gg.SetFormat(Format::COEFFICIENT);
auto ggg = gg.Times(pp);
auto gggg = ggg.ToNativePoly();
gggg.SwitchModulus(pp, 1, 0, 0);

It gives me this poly:

COEF: [7 7 0 0 7 0 7 7 0 0 0 0 0 0 7 0 0 7 7 0 7 7 0 7 0 0 7 7 0 7 7 7 0 0 7 0 ...] modulus: 17

Could someone please help me on how to make this multiplication and mod?

2

Please provide a complete example (minimum working example). It is not clear what paramPK is set to. If the starting modulus in paramsPK is too small (so that multiplying by 17 wraps around the modulus used there), then you can easily get non-zero values.

Another question, what are you actually trying to do?

3

Here is how I create parameters:

CCParams<CryptoContextBFVRNS> parameters;
parameters.SetPlaintextModulus(65537);
parameters.SetMultiplicativeDepth(2);
CryptoContext<DCRTPoly> cryptoContext = GenCryptoContext(parameters);
const auto cryptoParams = cryptoContext->GetCryptoParameters();
const std::shared_ptr<ParmType> paramsPK = cryptoParams->GetParamsPK();
std::cout << "ParamsPK:" << paramsPK->GetParams()[0]->GetModulus() << std::endl;
DugType dug(1);
int pp = 17;
DCRTPoly gg(dug, paramsPK, Format::EVALUATION);
gg.SetFormat(Format::COEFFICIENT);
auto ggg = gg.Times(pp);
auto gggg = ggg.ToNativePoly();
gggg.SwitchModulus(pp, 1, 0, 0);
std::cout << "gggg: " << gggg << std::endl;

I tried to change to use discrete uniform, but I guess I am missing on how is the DCRT polynomial representation, since it is a lot of big values. I am only trying to learn the library but got confused with both polynomial representations I guess.

This is the output trimmed:

ParamsPK:1152921504606830593
gggg: COEF: [5 11 1 3 1 6 7 4 10 8 4 0 15 14 7 12 13 14 3 11 15 5 8 9 1 6 6 5 15...] modulus: 17

How can I multiply the full polynomial by a value? I want to multiply a small poly (sampled by a gaussian) by a small number, I tried sampling with ternary as well and gives me wrong values. Only binary and other pp`s worked for now. For ternary, I don’t know why the values are 5, they should be 4 if the modulus operation is not balanced.

The main problem is that the sampler is centered on 0 but the operations (at least Times) are not using balanced modulo, do we have a workaround for this? If you have a polynomial (-x2 + x%101)*3, it should be (-3x2 + 3x)%101 and not (98x**2 + 3x)%101

5

Here is a properly working example. There were a couple of errors. First, dug (random number mod) was used instead of dgg. Second, ToNativePoly() was called, which is used for a different purpose. Now the result is all 0’s.

#include "openfhe.h"
#include "openfhecore.h"

using namespace lbcrypto;

int main() {
	CCParams<CryptoContextBFVRNS> parameters;
	parameters.SetPlaintextModulus(65537);
	parameters.SetMultiplicativeDepth(2);
	CryptoContext<DCRTPoly> cryptoContext = GenCryptoContext(parameters);
	const auto cryptoParams = cryptoContext->GetCryptoParameters();
	const auto paramsPK = cryptoParams->GetParamsPK();
	std::cout << "ParamsPK:" << paramsPK->GetParams()[0]->GetModulus() << std::endl;
	// this code generates random numbers between 0 and modulus - 1
	// DCRTPoly::DugType dug;
	DCRTPoly::DggType dgg(3.19);
	int pp = 17;
	DCRTPoly gg(dgg, paramsPK, Format::EVALUATION);
	gg.SetFormat(Format::COEFFICIENT);
	std::cout << "gg: " << gg << std::endl;
	auto ggg = gg.Times(pp);
	std::cout << "ggg: " << ggg << std::endl;
	auto gggg = ggg.GetAllElements()[0];
	std::cout << "gggg: " << gggg << std::endl;
	gggg.SwitchModulus(pp, 1, 0, 0);
	std::cout << "gggg: " << gggg << std::endl;
    return 0;
}