Added extra argument to a couple of matrix-library functions.

git-svn-id: https://svn.code.sf.net/p/kaldi/code/trunk@1363 5e6a8d80-dfce-4ca6-a32a-6e07a63d50c8

src/cudamatrix/cu-vector-inl.h

@@ -310,7 +310,7 @@ void CuVector<Real>::AddRowSumMat(Real alpha, const CuMatrix<Real> &mat, Real be
   #endif
   {
     Vector<Real> tmp(mat.NumCols());
-    tmp.AddRowSumMat(mat.Mat());
+    tmp.AddRowSumMat(1.0, mat.Mat());
     if(beta != 1.0) vec_.Scale(beta);
     vec_.AddVec(alpha, tmp);
   }
@@ -360,7 +360,7 @@ void CuVector<Real>::AddColSumMat(Real alpha, const CuMatrix<Real> &mat, Real be
   #endif
   {
     Vector<Real> tmp(mat.NumRows());
-    tmp.AddColSumMat(mat.Mat());
+    tmp.AddColSumMat(1.0, mat.Mat());
     if(beta != 1.0) vec_.Scale(beta);
     vec_.AddVec(alpha,tmp);
   }

src/featbin/compute-fbank-feats.cc

@@ -153,7 +153,7 @@ int main(int argc, char *argv[]) {
       }
       if (subtract_mean) {
         Vector<BaseFloat> mean(features.NumCols());
-        mean.AddRowSumMat(features);
+        mean.AddRowSumMat(1.0, features);
         mean.Scale(1.0 / features.NumRows());
         for (int32 i = 0; i < features.NumRows(); i++)
           features.Row(i).AddVec(-1.0, mean);

src/featbin/compute-mfcc-feats.cc

@@ -153,7 +153,7 @@ int main(int argc, char *argv[]) {
       }
       if (subtract_mean) {
         Vector<BaseFloat> mean(features.NumCols());
-        mean.AddRowSumMat(features);
+        mean.AddRowSumMat(1.0, features);
         mean.Scale(1.0 / features.NumRows());
         for (int32 i = 0; i < features.NumRows(); i++)
           features.Row(i).AddVec(-1.0, mean);

src/featbin/compute-plp-feats.cc

@@ -153,7 +153,7 @@ int main(int argc, char *argv[]) {
       }
       if (subtract_mean) {
         Vector<BaseFloat> mean(features.NumCols());
-        mean.AddRowSumMat(features);
+        mean.AddRowSumMat(1.0, features);
         mean.Scale(1.0 / features.NumRows());
         for (size_t i = 0; i < features.NumRows(); i++)
           features.Row(i).AddVec(-1.0, mean);

src/featbin/compute-spectrogram-feats.cc

@@ -116,7 +116,7 @@ int main(int argc, char *argv[]) {
       }
       if (subtract_mean) {
         Vector<BaseFloat> mean(features.NumCols());
-        mean.AddRowSumMat(features);
+        mean.AddRowSumMat(1.0, features);
         mean.Scale(1.0 / features.NumRows());
         for (int32 i = 0; i < features.NumRows(); i++)
           features.Row(i).AddVec(-1.0, mean);

src/featbin/remove-mean.cc

@@ -52,7 +52,7 @@ int main(int argc, char *argv[]) {
         continue;
       }
       Vector<BaseFloat> mean(feats.NumCols());
-      mean.AddRowSumMat(feats);
+      mean.AddRowSumMat(1.0, feats);
       mean.Scale(1.0 / feats.NumRows());
       for (int32 i = 0; i < feats.NumRows(); i++)
         feats.Row(i).AddVec(-1.0, mean);

src/gmm/mle-full-gmm-test.cc

@@ -309,7 +309,7 @@ UnitTestEstimateFullGmm() {
     Vector<BaseFloat> mean(dim);
     cov.AddMatMat(1.0, feats, kTrans, feats, kNoTrans, 0.0);
     cov.Scale(1.0 / feats.NumRows());
-    mean.AddRowSumMat(feats);
+    mean.AddRowSumMat(1.0, feats);
     mean.Scale(1.0 / feats.NumRows());
     cov.AddVecVec(-1.0, mean, mean);
     BaseFloat logdet = cov.LogDet();

src/matrix/kaldi-matrix.cc

@@ -763,8 +763,7 @@ void MatrixBase<double>::CopyFromTp(const TpMatrix<double> & M,
 
 template<typename Real>
 void MatrixBase<Real>::CopyRowsFromVec(const VectorBase<Real> &rv) {
-  KALDI_ASSERT(rv.Dim() == num_rows_*num_cols_);
+  if (rv.Dim() == num_rows_*num_cols_) {
-
     if (stride_ == num_cols_) {
       // one big copy operation.
       const Real *rv_data = rv.Data();
@@ -779,12 +778,19 @@ void MatrixBase<Real>::CopyRowsFromVec(const VectorBase<Real> &rv) {
         rv_data += num_cols_;
       }
     }
+  } else if (rv.Dim() == num_cols_) {
+    const Real *rv_data = rv.Data();
+    for (MatrixIndexT r = 0; r < num_rows_; r++)
+      std::memcpy(RowData(r), rv_data, sizeof(Real)*num_cols_);
+  } else {
+    KALDI_ERR << "Wrong sized arguments";
+  }
 }
 
 template<typename Real>
 template<typename OtherReal>
 void MatrixBase<Real>::CopyRowsFromVec(const VectorBase<OtherReal> &rv) {
-  KALDI_ASSERT(rv.Dim() == num_rows_*num_cols_);
+  if (rv.Dim() == num_rows_*num_cols_) {
     const OtherReal *rv_data = rv.Data();
     for (MatrixIndexT r = 0; r < num_rows_; r++) {
       Real *row_data = RowData(r);
@@ -793,7 +799,18 @@ void MatrixBase<Real>::CopyRowsFromVec(const VectorBase<OtherReal> &rv) {
       }
       rv_data += num_cols_;
     }
+  } else if (rv.Dim() == num_cols_) {
+    const OtherReal *rv_data = rv.Data();
+    Real *first_row_data = RowData(0);
+    for (MatrixIndexT c = 0; c < num_cols_; c++)
+      first_row_data[c] = rv_data[c];
+    for (MatrixIndexT r = 1; r < num_rows_; r++)
+      std::memcpy(RowData(r), first_row_data, sizeof(Real)*num_cols_);
+  } else {
+    KALDI_ERR << "Wrong sized arguments.";
   }
+}
+  
 
 template
 void MatrixBase<float>::CopyRowsFromVec(const VectorBase<double> &rv);
@@ -802,8 +819,7 @@ void MatrixBase<double>::CopyRowsFromVec(const VectorBase<float> &rv);
 
 template<typename Real>
 void MatrixBase<Real>::CopyColsFromVec(const VectorBase<Real> &rv) {
-  KALDI_ASSERT(rv.Dim() == num_rows_*num_cols_);
+  if (rv.Dim() == num_rows_*num_cols_) {
-
     const Real *v_inc_data = rv.Data();
     Real *m_inc_data = data_;
 
@@ -814,6 +830,18 @@ void MatrixBase<Real>::CopyColsFromVec(const VectorBase<Real> &rv) {
       v_inc_data += num_rows_;
       m_inc_data ++;
     }
+  } else if (rv.Dim() == num_rows_) {
+    const Real *v_inc_data = rv.Data();
+    Real *m_inc_data = data_;
+    for (MatrixIndexT r = 0; r < num_rows_; r++) {
+      BaseFloat value = *(v_inc_data++);
+      for (MatrixIndexT c = 0; c < num_cols_; c++)
+        m_inc_data[c] = value;
+      m_inc_data += stride_;
+    }
+  } else {
+    KALDI_ERR << "Wrong size of arguments.";
+  }
 }
 
 

src/matrix/kaldi-matrix.h

@@ -130,14 +130,17 @@ class MatrixBase {
                   MatrixTransposeType Trans = kNoTrans);
   
   /// Inverse of vec() operator. Copies vector into matrix, row-by-row.
-  /// Note that rv.Dim() must equal NumRows()*NumCols().
+  /// Note that rv.Dim() must either equal NumRows()*NumCols() or
+  /// NumCols()-- this has two modes of operation.
   void CopyRowsFromVec(const VectorBase<Real> &v);
   template<typename OtherReal>
   void CopyRowsFromVec(const VectorBase<OtherReal> &v);
 
   /// Copies vector into matrix, column-by-column.
-  /// Note that rv.Dim() must equal NumRows()*NumCols().
+  /// Note that rv.Dim() must either equal NumRows()*NumCols() or NumRows();
+  /// this has two modes of operation.
   void CopyColsFromVec(const VectorBase<Real> &v);
+  
   /// Copy vector into specific column of matrix.
   void CopyColFromVec(const VectorBase<Real> &v, const MatrixIndexT col);
   /// Copy vector into specific row of matrix.
@@ -745,6 +748,12 @@ std::istream & operator >> (std::istream & In, MatrixBase<Real> & M);
 template<typename Real>
 std::istream & operator >> (std::istream & In, Matrix<Real> & M);
 
+
+template<class Real>
+bool SameDim(const MatrixBase<Real> &M, const MatrixBase<Real> &N) {
+  return (M.NumRows() == N.NumRows() && M.NumCols() == N.NumCols());
+}
+
 /// @} end of \addtogroup matrix_funcs_io
 
 

src/matrix/kaldi-vector.cc

@@ -551,7 +551,7 @@ Real VectorBase<Real>::SumLog() const {
 }
 
 template<typename Real>
-void VectorBase<Real>::AddRowSumMat(const MatrixBase<Real> &rM) {
+void VectorBase<Real>::AddRowSumMat(Real alpha, const MatrixBase<Real> &rM) {
   // note the double accumulator
   double sum;
   KALDI_ASSERT(dim_ == rM.NumCols());
@@ -560,12 +560,12 @@ void VectorBase<Real>::AddRowSumMat(const MatrixBase<Real> &rM) {
     for (MatrixIndexT j = 0; j < rM.NumRows(); j++) {
       sum += rM(j, i);
     }
-    data_[i] += sum;
+    data_[i] += alpha * sum;
   }
 }
 
 template<typename Real>
-void VectorBase<Real>::AddColSumMat(const MatrixBase<Real> &rM) {
+void VectorBase<Real>::AddColSumMat(Real alpha, const MatrixBase<Real> &rM) {
   // note the double accumulator
   double sum;
   KALDI_ASSERT(dim_ == rM.NumRows());
@@ -574,7 +574,7 @@ void VectorBase<Real>::AddColSumMat(const MatrixBase<Real> &rM) {
     for (MatrixIndexT j = 0; j < rM.NumCols(); j++) {
       sum += rM(i, j);
     }
-    data_[i] += sum;
+    data_[i] += alpha * sum;
   }
 }
 

src/matrix/kaldi-vector.h

@@ -242,11 +242,11 @@ class VectorBase {
   /// negative.
   Real SumLog() const;
 
-  /// Adds sum of the rows of M to existing contents.
+  /// Adds sum of the rows of M to existing contents, times alpha.
-  void AddRowSumMat(const MatrixBase<Real>& M);
+  void AddRowSumMat(Real alpha, const MatrixBase<Real>& M);
 
   /// Adds sum of the columns of M to existing contents.
-  void AddColSumMat(const MatrixBase<Real>& M);
+  void AddColSumMat(Real alpha, const MatrixBase<Real>& M);
 
   /// Returns log(sum(exp())) without exp overflow
   /// If prune > 0.0, ignores terms less than the max - prune.

src/matrix/matrix-lib-test.cc

@@ -209,6 +209,27 @@ static void UnitTestSpAddVec() {
   }
 }
 
+template<class Real> static void UnitTestCopyRowsAndCols() {
+  // Test other mode of CopyRowsFromVec, and CopyColsFromVec,
+  // where vector is duplicated.
+  for (int32 i = 0; i < 30; i++) {
+    int32 dimM = 1 + rand() % 5, dimN = 1 + rand() % 5;
+    Vector<float> w(dimN); // test cross-type version of
+    // CopyRowsFromVec.
+    Vector<Real> v(dimM);
+    Matrix<Real> M(dimM, dimN), N(dimM, dimN);
+    InitRand(&v);
+    InitRand(&w);
+    M.CopyColsFromVec(v);
+    N.CopyRowsFromVec(w);
+    for (int32 r = 0; r < dimM; r++) {
+      for (int32 c = 0; c < dimN; c++) {
+        KALDI_ASSERT(M(r, c) == v(r));
+        KALDI_ASSERT(N(r, c) == w(c));
+      }
+    }    
+  }
+}
 
 template<class Real> static void UnitTestSpliceRows() {
 
@@ -354,8 +375,10 @@ static void UnitTestSimpleForVec() {  // testing some simple operaters on vector
     Matrix<Real> M(dimM, dimN);
     InitRand(&M);
     Vector<Real> Vr(dimN), Vc(dimM);
-    Vr.AddRowSumMat(M);
+    Vr.AddRowSumMat(0.5, M);
-    Vc.AddColSumMat(M);
+    Vc.AddColSumMat(0.5, M);
+    Vr.Scale(2.0);
+    Vc.Scale(2.0);
 
     Vector<Real> V2r(dimN), V2c(dimM);
     for (MatrixIndexT k = 0; k < dimM; k++) {
@@ -2972,6 +2995,7 @@ template<class Real> static void MatrixUnitTest() {
   UnitTestMat2Vec<Real>();
   UnitTestSpLogExp<Real>();
   KALDI_LOG << " Point H";
+  UnitTestCopyRowsAndCols<Real>();
   UnitTestSpliceRows<Real>();
   UnitTestAddSp<Real>();
   UnitTestRemoveRow<Real>();

src/nnet_cpu/nnet-biasedlinearity.h

@@ -72,7 +72,7 @@ class BiasedLinearity : public UpdatableComponent {
     // compute gradient
     linearity_corr_.AddMatMat(1.0, err, kTrans, input, kNoTrans, momentum_);
     bias_corr_.Scale(momentum_);
-    bias_corr_.AddRowSumMat(err);
+    bias_corr_.AddRowSumMat(1.0, err);
     // l2 regularization
     if (l2_penalty_ != 0.0) {
       linearity_.AddMat(-learn_rate_*l2_penalty_*input.NumRows(), linearity_);

src/nnet_cpu/nnet-rnnlm.cc

@@ -128,7 +128,7 @@ void Rnnlm::Backpropagate(const MatrixBase<BaseFloat> &in_err) {
   // update layer2
   W2_.AddMatMat(-learn_rate_, h2_, kTrans, in_err, kNoTrans, 1.0);
   b2_corr_.SetZero();
-  b2_corr_.AddRowSumMat(in_err);
+  b2_corr_.AddRowSumMat(1.0, in_err);
   b2_.AddVec(-learn_rate_, b2_corr_);
 
   // LAYER1
@@ -136,7 +136,7 @@ void Rnnlm::Backpropagate(const MatrixBase<BaseFloat> &in_err) {
   U1_corr_.SetZero();
   b1_corr_.SetZero();
   // accumulate gradient for layer1
-  b1_corr_.AddRowSumMat(e2_);
+  b1_corr_.AddRowSumMat(1.0, e2_);
   for(int32 r=0; r<e2_.NumRows();r++) {
     V1_corr_.Row(in_seq_[r]-1).AddVec(1.0, e2_.Row(r));
   }
@@ -174,7 +174,7 @@ void Rnnlm::Backpropagate(const MatrixBase<BaseFloat> &in_err) {
     }
 
     // accumulate graidient
-    b1_corr_.AddRowSumMat(E);
+    b1_corr_.AddRowSumMat(1.0, E);
     for(int32 r=0; r<E.NumRows();r++) {
       // :TODO: IS IT CORRECT?
       // V1_corr_.Row(in_seq_[r+step]).AddVec(1.0,E.Row(r));

src/sgmm/estimate-am-sgmm-ebw.cc

@@ -273,8 +273,8 @@ double EbwAmSgmmUpdater::UpdateM(const MleAmSgmmAccs &num_accs,
   
   Vector<double> num_count_vec(I), den_count_vec(I), impr_vec(I);
   for (int32 j = 0; j < num_accs.num_states_; j++) {
-    num_count_vec.AddRowSumMat(num_accs.gamma_[j]);
+    num_count_vec.AddRowSumMat(1.0, num_accs.gamma_[j]);
-    den_count_vec.AddRowSumMat(den_accs.gamma_[j]);
+    den_count_vec.AddRowSumMat(1.0, den_accs.gamma_[j]);
   }
 
   for (int32 i = 0; i < I; i++) {
@@ -363,8 +363,8 @@ double EbwAmSgmmUpdater::UpdateWParallel(const MleAmSgmmAccs &num_accs,
   
   Vector<double> num_count_vec(I), den_count_vec(I), impr_vec(I);
   for (int32 j = 0; j < num_accs.num_states_; j++) {
-    num_count_vec.AddRowSumMat(num_accs.gamma_[j]);
+    num_count_vec.AddRowSumMat(1.0, num_accs.gamma_[j]);
-    den_count_vec.AddRowSumMat(den_accs.gamma_[j]);
+    den_count_vec.AddRowSumMat(1.0, den_accs.gamma_[j]);
   }
   
   // Get the F_i and g_i quantities-- this is done in parallel (multi-core),
@@ -448,8 +448,8 @@ double EbwAmSgmmUpdater::UpdateN(const MleAmSgmmAccs &num_accs,
   
   Vector<double> num_count_vec(I), den_count_vec(I), impr_vec(I);
   for (int32 j = 0; j < num_accs.num_states_; j++) {
-    num_count_vec.AddRowSumMat(num_accs.gamma_[j]);
+    num_count_vec.AddRowSumMat(1.0, num_accs.gamma_[j]);
-    den_count_vec.AddRowSumMat(den_accs.gamma_[j]);
+    den_count_vec.AddRowSumMat(1.0, den_accs.gamma_[j]);
   }
   
   for (int32 i = 0; i < I; i++) {
@@ -522,8 +522,8 @@ double EbwAmSgmmUpdater::UpdateVars(const MleAmSgmmAccs &num_accs,
   
   Vector<double> num_count_vec(I), den_count_vec(I), impr_vec(I);
   for (int32 j = 0; j < num_accs.num_states_; j++) {
-    num_count_vec.AddRowSumMat(num_accs.gamma_[j]);
+    num_count_vec.AddRowSumMat(1.0, num_accs.gamma_[j]);
-    den_count_vec.AddRowSumMat(den_accs.gamma_[j]);
+    den_count_vec.AddRowSumMat(1.0, den_accs.gamma_[j]);
   }
 
   for (int32 i = 0; i < I; i++) {

src/sgmm2/am-sgmm.cc

@@ -538,7 +538,7 @@ BaseFloat AmSgmm2::LogLikelihood(const Sgmm2PerFrameDerivedVars &per_frame_vars,
     substate_cache.remaining_log_like = max;
     int32 num_substates = loglikes.NumCols();
     substate_cache.likes.Resize(num_substates); // zeroes it.
-    substate_cache.likes.AddRowSumMat(loglikes); // add likelihoods [not in log!] for
+    substate_cache.likes.AddRowSumMat(1.0, loglikes); // add likelihoods [not in log!] for
     // each column [i.e. summing over the rows], so we get the sum for
     // each substate index.  You have to multiply by exp(remaining_log_like)
     // to get a real likelihood.
@@ -613,7 +613,7 @@ void AmSgmm2::SplitSubstatesInGroup(const Vector<BaseFloat> &pdf_occupancies,
     int32 split_m; // substate to split.
     {
       Vector<BaseFloat> substate_count(tgt_M);
-      substate_count.AddRowSumMat(c_j);
+      substate_count.AddRowSumMat(1.0, c_j);
       BaseFloat *data = substate_count.Data();
       split_m = std::max_element(data, data+cur_M) - data;
     }

src/sgmm2/estimate-am-sgmm-ebw.cc

@@ -50,10 +50,10 @@ void EbwAmSgmm2Updater::Update(const MleAmSgmm2Accs &num_accs,
   
   Vector<double> gamma_num(num_accs.num_gaussians_);
   for (int32 j1 = 0; j1 < num_accs.num_groups_; j1++)
-    gamma_num.AddRowSumMat(num_accs.gamma_[j1]); 
+    gamma_num.AddRowSumMat(1.0, num_accs.gamma_[j1]); 
   Vector<double> gamma_den(den_accs.num_gaussians_);
   for (int32 j1 = 0; j1 < den_accs.num_groups_; j1++)
-    gamma_den.AddRowSumMat(den_accs.gamma_[j1]); 
+    gamma_den.AddRowSumMat(1.0, den_accs.gamma_[j1]); 
 
   BaseFloat tot_impr = 0.0;
 

src/sgmm2/estimate-am-sgmm.cc

@@ -617,7 +617,7 @@ void MleAmSgmm2Updater::Update(const MleAmSgmm2Accs &accs,
   
   Vector<double> gamma_i(accs.num_gaussians_);
   for (int32 j1 = 0; j1 < accs.num_groups_; j1++)
-    gamma_i.AddRowSumMat(accs.gamma_[j1]); // add sum of rows of
+    gamma_i.AddRowSumMat(1.0, accs.gamma_[j1]); // add sum of rows of
   // accs.gamma_[j1], to gamma_i.
   
   if (flags & kSgmmPhoneProjections)

src/transform/lda-estimate.cc

@@ -63,7 +63,7 @@ void LdaEstimate::Estimate(int32 target_dim,
   // total covariance
   double sum = zero_acc_.Sum();
   Vector<double> total_mean(dim);
-  total_mean.AddRowSumMat(first_acc_);
+  total_mean.AddRowSumMat(1.0, first_acc_);
   total_mean.Scale(1/sum);
   SpMatrix<double> total_covar(total_second_acc_);
   total_covar.Scale(1/sum);