Fix BFVpp

include/bfv++.hpp

@@ -10,11 +10,6 @@
 
 namespace TFHEpp {
 template <class P>
-inline void RemoveSign(UnsignedPolynomial<P> &res, const Polynomial<P> &a)
-{
-    for (int i = 0; i < P::n; i++) res[i] = (a[i] + 1) / 2;
-}
-template <class P>
 void TRLWEMultWithoutRelinerization(TRLWE3<P> &res, const TRLWE<P> &a,
                                     const TRLWE<P> &b)
 {
@@ -34,18 +29,18 @@ void TRLWEMultWithoutRelinerization(TRLWE3<P> &res, const TRLWE<P> &a,
     TwistFFTrescale<P>(res[0], fftc);
 
     PolyMulRescaleUnsigned<P>(res[1], aa[1], bb[1]);
-    // PolyMulRescaleUnsigned<P>(res[2], aa[0], bb[0]);
+    PolyMulRescaleUnsigned<P>(res[2], aa[0], bb[0]);
 
-    for (int i = 0; i < P::n; i++) {
-        uint64_t ri = 0;
-        for (int j = 0; j <= i; j++)
-            ri += static_cast<uint64_t>(P::plain_modulus) *
-                  static_cast<uint64_t>(a[0][j]) * b[0][i - j];
-        for (int j = i + 1; j < P::n; j++)
-            ri -= P::plain_modulus * static_cast<uint64_t>(a[0][j]) *
-                  b[0][P::n + i - j];
-        res[2][i] = (ri + (1ULL << 31)) >> 32;
-    }
+    // for (int i = 0; i < P::n; i++) {
+    //     uint64_t ri = 0;
+    //     for (int j = 0; j <= i; j++)
+    //         ri += static_cast<uint64_t>(P::plain_modulus) *
+    //               static_cast<uint64_t>(a[0][j]) * b[0][i - j];
+    //     for (int j = i + 1; j < P::n; j++)
+    //         ri -= P::plain_modulus * static_cast<uint64_t>(a[0][j]) *
+    //               b[0][P::n + i - j];
+    //     res[2][i] = (ri + (1ULL << 31)) >> 32;
+    // }
 }
 
 template <class P>

include/mulfft.hpp

@@ -144,6 +144,7 @@ inline void TwistIFFT(PolynomialInFD<P> &res, const Polynomial<P> &a)
     else
         static_assert(false_v<typename P::T>, "Undefined TwistIFFT!");
 }
+
 template <uint32_t N>
 inline void MulInFD(std::array<double, N> &res, const std::array<double, N> &b)
 {
@@ -308,7 +309,7 @@ inline void PolyMulRescaleUnsigned(Polynomial<P> &res,
     PolynomialInFD<P> ffta, fftb;
     TwistIFFT<P>(ffta, a);
     TwistIFFT<P>(fftb, b);
-    MulInFD<P::n>(ffta, ffta, fftb);
+    MulInFD<P::n>(ffta, fftb);
     TwistFFTrescale<P>(res, ffta);
     // }
     // else
@@ -333,6 +334,38 @@ inline void PolyMulNaive(Polynomial<P> &res, const Polynomial<P> &a,
     }
 }
 
+template <class P>
+inline void PolyMulRescale(Polynomial<P> &res, const Polynomial<P> &a,
+                           const Polynomial<P> &b)
+{
+    if constexpr (std::is_same_v<typename P::T, uint32_t>) {
+        UnsignedPolynomial<P> aa, bb;
+        RemoveSign<P>(aa, a);
+        RemoveSign<P>(bb, b);
+        PolyMulRescaleUnsigned<P>(res, aa, bb);
+    }
+    else
+        static_assert(false_v<typename P::T>, "Undefined PolyMul!");
+}
+
+template <class P>
+inline void PolyMulNaieveRescale(Polynomial<P> &res, const Polynomial<P> &a,
+                                 const Polynomial<P> &b)
+{
+    Polynomial<P> aa, bb;
+    for (int i = 0; i < P::n; i++) aa[i] = (a[i] + 1) / 2;
+    for (int i = 0; i < P::n; i++) bb[i] = (b[i] + 1) / 2;
+    for (int i = 0; i < P::n; i++) {
+        __int128_t ri = 0;
+        for (int j = 0; j <= i; j++)
+            ri += static_cast<__int128_t>(aa[j]) * bb[i - j];
+        for (int j = i + 1; j < P::n; j++)
+            ri -= static_cast<__int128_t>(aa[j]) * bb[P::n + i - j];
+        // res[i] = static_cast<typename P::T>((ri) >> (std::numeric_limits<typename P::T>::digits - 3));
+        res[i] = static_cast<typename P::T>((ri) >> 29);
+    }
+}
+
 template <class P>
 std::unique_ptr<std::array<PolynomialInFD<P>, 2 * P::n>> XaittGen()
 {

include/utils.hpp

@@ -221,4 +221,10 @@ inline void Automorphism(Polynomial<P> &res, const Polynomial<P> &poly,
     }
 }
 
+template <class P>
+inline void RemoveSign(UnsignedPolynomial<P> &res, const Polynomial<P> &a)
+{
+    for (int i = 0; i < P::n; i++) res[i] = (a[i] + 1) / 2;
+}
+
 }  // namespace TFHEpp

src/bfv.cpp

@@ -0,0 +1,4 @@
+// #include<bfv++.hpp>
+
+// thread_local intel::hexl::NTT nttlvl1(TFHEpp::lvl1param::n, TFHEpp::lvl1param::q);
+// // thread_local intel::hexl::NTT nttlvl2(TFHEpp::lvl2param::n, TFHEpp::lvl2param::q);

test/polymul.cpp

@@ -72,5 +72,25 @@ int main()
     }
     cout << "FFT Passed" << endl;
 
+    std::cout << "PolyMulRescale Test" << std::endl;
+    for (int test = 0; test < num_test; test++) {
+        std::random_device seed_gen;
+        std::default_random_engine engine(seed_gen());
+        std::uniform_int_distribution<typename TFHEpp::lvl1param::T> message(
+            0, (1ULL << 32) - 1);
+
+        TFHEpp::Polynomial<TFHEpp::lvl1param> p0, p1, pres, ptrue;
+        for (typename TFHEpp::lvl1param::T &i : p0) i = message(engine);
+        for (typename TFHEpp::lvl1param::T &i : p1) i = message(engine);
+
+        TFHEpp::PolyMulRescale<TFHEpp::lvl1param>(pres, p0, p1);
+        TFHEpp::PolyMulNaieveRescale<TFHEpp::lvl1param>(ptrue, p0, p1);
+
+        for (int i = 0; i < TFHEpp::lvl1param::n; i++) {
+            // std::cout<<pres[i]<<":"<<ptrue[i]<<std::endl;
+            assert(abs(static_cast<int>(pres[i] - ptrue[i])) <= 2);
+        }
+    }
+    std::cout << "PolyMulRescale Passed" << std::endl; 
     return 0;
 }

test/trlwemult.cpp

@@ -76,7 +76,7 @@ int main()
         for (int i = 0; i < P::n; i++) ptrue[i] %= P::plain_modulus;
 
         // for (int i = 0; i < P::n; i++)
-        //     std::cout<<pres[i]<<":"<<ptrue[i]<<std::endl;
+            // std::cout<<pres[i]<<":"<<ptrue[i]<<std::endl;
         for (int i = 0; i < P::n; i++) assert(pres[i] == ptrue[i]);
     }
     std::cout << "Passed" << std::endl;

thirdparties/spqlios/fft_processor_spqlios.cpp

@@ -163,7 +163,7 @@ void FFT_Processor_Spqlios::execute_direct_torus32_rescale(uint32_t *res, const
         );
     }
     fft(tables_direct, real_inout_direct);
-    for (int32_t i = 0; i < N; i++) res[i] = uint32_t(std::round(real_inout_direct[i]/(Δ/4)));
+    for (int32_t i = 0; i < N; i++) res[i] = static_cast<uint32_t>(int64_t(real_inout_direct[i]/(Δ/4)));
 }
 
 void FFT_Processor_Spqlios::execute_direct_torus64(uint64_t* res, const double* a) {