1 // This file is part of Eigen, a lightweight C++ template library 2 // for linear algebra. 3 // 4 // Copyright (C) 2008-2015 Gael Guennebaud <gael.guennebaud (at) inria.fr> 5 // 6 // This Source Code Form is subject to the terms of the Mozilla 7 // Public License v. 2.0. If a copy of the MPL was not distributed 8 // with this file, You can obtain one at http://mozilla.org/MPL/2.0/. 9 10 #include "sparse.h" 11 12 template<typename T> 13 typename Eigen::internal::enable_if<(T::Flags&RowMajorBit)==RowMajorBit, typename T::RowXpr>::type 14 innervec(T& A, Index i) 15 { 16 return A.row(i); 17 } 18 19 template<typename T> 20 typename Eigen::internal::enable_if<(T::Flags&RowMajorBit)==0, typename T::ColXpr>::type 21 innervec(T& A, Index i) 22 { 23 return A.col(i); 24 } 25 26 template<typename SparseMatrixType> void sparse_block(const SparseMatrixType& ref) 27 { 28 const Index rows = ref.rows(); 29 const Index cols = ref.cols(); 30 const Index inner = ref.innerSize(); 31 const Index outer = ref.outerSize(); 32 33 typedef typename SparseMatrixType::Scalar Scalar; 34 typedef typename SparseMatrixType::StorageIndex StorageIndex; 35 36 double density = (std::max)(8./(rows*cols), 0.01); 37 typedef Matrix<Scalar,Dynamic,Dynamic,SparseMatrixType::IsRowMajor?RowMajor:ColMajor> DenseMatrix; 38 typedef Matrix<Scalar,Dynamic,1> DenseVector; 39 typedef Matrix<Scalar,1,Dynamic> RowDenseVector; 40 typedef SparseVector<Scalar> SparseVectorType; 41 42 Scalar s1 = internal::random<Scalar>(); 43 { 44 SparseMatrixType m(rows, cols); 45 DenseMatrix refMat = DenseMatrix::Zero(rows, cols); 46 initSparse<Scalar>(density, refMat, m); 47 48 VERIFY_IS_APPROX(m, refMat); 49 50 // test InnerIterators and Block expressions 51 for (int t=0; t<10; ++t) 52 { 53 Index j = internal::random<Index>(0,cols-2); 54 Index i = internal::random<Index>(0,rows-2); 55 Index w = internal::random<Index>(1,cols-j); 56 Index h = internal::random<Index>(1,rows-i); 57 58 VERIFY_IS_APPROX(m.block(i,j,h,w), refMat.block(i,j,h,w)); 59 for(Index c=0; c<w; c++) 60 { 61 VERIFY_IS_APPROX(m.block(i,j,h,w).col(c), refMat.block(i,j,h,w).col(c)); 62 for(Index r=0; r<h; r++) 63 { 64 VERIFY_IS_APPROX(m.block(i,j,h,w).col(c).coeff(r), refMat.block(i,j,h,w).col(c).coeff(r)); 65 VERIFY_IS_APPROX(m.block(i,j,h,w).coeff(r,c), refMat.block(i,j,h,w).coeff(r,c)); 66 } 67 } 68 for(Index r=0; r<h; r++) 69 { 70 VERIFY_IS_APPROX(m.block(i,j,h,w).row(r), refMat.block(i,j,h,w).row(r)); 71 for(Index c=0; c<w; c++) 72 { 73 VERIFY_IS_APPROX(m.block(i,j,h,w).row(r).coeff(c), refMat.block(i,j,h,w).row(r).coeff(c)); 74 VERIFY_IS_APPROX(m.block(i,j,h,w).coeff(r,c), refMat.block(i,j,h,w).coeff(r,c)); 75 } 76 } 77 78 VERIFY_IS_APPROX(m.middleCols(j,w), refMat.middleCols(j,w)); 79 VERIFY_IS_APPROX(m.middleRows(i,h), refMat.middleRows(i,h)); 80 for(Index r=0; r<h; r++) 81 { 82 VERIFY_IS_APPROX(m.middleCols(j,w).row(r), refMat.middleCols(j,w).row(r)); 83 VERIFY_IS_APPROX(m.middleRows(i,h).row(r), refMat.middleRows(i,h).row(r)); 84 for(Index c=0; c<w; c++) 85 { 86 VERIFY_IS_APPROX(m.col(c).coeff(r), refMat.col(c).coeff(r)); 87 VERIFY_IS_APPROX(m.row(r).coeff(c), refMat.row(r).coeff(c)); 88 89 VERIFY_IS_APPROX(m.middleCols(j,w).coeff(r,c), refMat.middleCols(j,w).coeff(r,c)); 90 VERIFY_IS_APPROX(m.middleRows(i,h).coeff(r,c), refMat.middleRows(i,h).coeff(r,c)); 91 if(m.middleCols(j,w).coeff(r,c) != Scalar(0)) 92 { 93 VERIFY_IS_APPROX(m.middleCols(j,w).coeffRef(r,c), refMat.middleCols(j,w).coeff(r,c)); 94 } 95 if(m.middleRows(i,h).coeff(r,c) != Scalar(0)) 96 { 97 VERIFY_IS_APPROX(m.middleRows(i,h).coeff(r,c), refMat.middleRows(i,h).coeff(r,c)); 98 } 99 } 100 } 101 for(Index c=0; c<w; c++) 102 { 103 VERIFY_IS_APPROX(m.middleCols(j,w).col(c), refMat.middleCols(j,w).col(c)); 104 VERIFY_IS_APPROX(m.middleRows(i,h).col(c), refMat.middleRows(i,h).col(c)); 105 } 106 } 107 108 for(Index c=0; c<cols; c++) 109 { 110 VERIFY_IS_APPROX(m.col(c) + m.col(c), (m + m).col(c)); 111 VERIFY_IS_APPROX(m.col(c) + m.col(c), refMat.col(c) + refMat.col(c)); 112 } 113 114 for(Index r=0; r<rows; r++) 115 { 116 VERIFY_IS_APPROX(m.row(r) + m.row(r), (m + m).row(r)); 117 VERIFY_IS_APPROX(m.row(r) + m.row(r), refMat.row(r) + refMat.row(r)); 118 } 119 } 120 121 // test innerVector() 122 { 123 DenseMatrix refMat2 = DenseMatrix::Zero(rows, cols); 124 SparseMatrixType m2(rows, cols); 125 initSparse<Scalar>(density, refMat2, m2); 126 Index j0 = internal::random<Index>(0,outer-1); 127 Index j1 = internal::random<Index>(0,outer-1); 128 Index r0 = internal::random<Index>(0,rows-1); 129 Index c0 = internal::random<Index>(0,cols-1); 130 131 VERIFY_IS_APPROX(m2.innerVector(j0), innervec(refMat2,j0)); 132 VERIFY_IS_APPROX(m2.innerVector(j0)+m2.innerVector(j1), innervec(refMat2,j0)+innervec(refMat2,j1)); 133 134 m2.innerVector(j0) *= Scalar(2); 135 innervec(refMat2,j0) *= Scalar(2); 136 VERIFY_IS_APPROX(m2, refMat2); 137 138 m2.row(r0) *= Scalar(3); 139 refMat2.row(r0) *= Scalar(3); 140 VERIFY_IS_APPROX(m2, refMat2); 141 142 m2.col(c0) *= Scalar(4); 143 refMat2.col(c0) *= Scalar(4); 144 VERIFY_IS_APPROX(m2, refMat2); 145 146 m2.row(r0) /= Scalar(3); 147 refMat2.row(r0) /= Scalar(3); 148 VERIFY_IS_APPROX(m2, refMat2); 149 150 m2.col(c0) /= Scalar(4); 151 refMat2.col(c0) /= Scalar(4); 152 VERIFY_IS_APPROX(m2, refMat2); 153 154 SparseVectorType v1; 155 VERIFY_IS_APPROX(v1 = m2.col(c0) * 4, refMat2.col(c0)*4); 156 VERIFY_IS_APPROX(v1 = m2.row(r0) * 4, refMat2.row(r0).transpose()*4); 157 158 SparseMatrixType m3(rows,cols); 159 m3.reserve(VectorXi::Constant(outer,int(inner/2))); 160 for(Index j=0; j<outer; ++j) 161 for(Index k=0; k<(std::min)(j,inner); ++k) 162 m3.insertByOuterInner(j,k) = internal::convert_index<StorageIndex>(k+1); 163 for(Index j=0; j<(std::min)(outer, inner); ++j) 164 { 165 VERIFY(j==numext::real(m3.innerVector(j).nonZeros())); 166 if(j>0) 167 VERIFY(j==numext::real(m3.innerVector(j).lastCoeff())); 168 } 169 m3.makeCompressed(); 170 for(Index j=0; j<(std::min)(outer, inner); ++j) 171 { 172 VERIFY(j==numext::real(m3.innerVector(j).nonZeros())); 173 if(j>0) 174 VERIFY(j==numext::real(m3.innerVector(j).lastCoeff())); 175 } 176 177 VERIFY(m3.innerVector(j0).nonZeros() == m3.transpose().innerVector(j0).nonZeros()); 178 179 // m2.innerVector(j0) = 2*m2.innerVector(j1); 180 // refMat2.col(j0) = 2*refMat2.col(j1); 181 // VERIFY_IS_APPROX(m2, refMat2); 182 } 183 184 // test innerVectors() 185 { 186 DenseMatrix refMat2 = DenseMatrix::Zero(rows, cols); 187 SparseMatrixType m2(rows, cols); 188 initSparse<Scalar>(density, refMat2, m2); 189 if(internal::random<float>(0,1)>0.5f) m2.makeCompressed(); 190 Index j0 = internal::random<Index>(0,outer-2); 191 Index j1 = internal::random<Index>(0,outer-2); 192 Index n0 = internal::random<Index>(1,outer-(std::max)(j0,j1)); 193 if(SparseMatrixType::IsRowMajor) 194 VERIFY_IS_APPROX(m2.innerVectors(j0,n0), refMat2.block(j0,0,n0,cols)); 195 else 196 VERIFY_IS_APPROX(m2.innerVectors(j0,n0), refMat2.block(0,j0,rows,n0)); 197 if(SparseMatrixType::IsRowMajor) 198 VERIFY_IS_APPROX(m2.innerVectors(j0,n0)+m2.innerVectors(j1,n0), 199 refMat2.middleRows(j0,n0)+refMat2.middleRows(j1,n0)); 200 else 201 VERIFY_IS_APPROX(m2.innerVectors(j0,n0)+m2.innerVectors(j1,n0), 202 refMat2.block(0,j0,rows,n0)+refMat2.block(0,j1,rows,n0)); 203 204 VERIFY_IS_APPROX(m2, refMat2); 205 206 VERIFY(m2.innerVectors(j0,n0).nonZeros() == m2.transpose().innerVectors(j0,n0).nonZeros()); 207 208 m2.innerVectors(j0,n0) = m2.innerVectors(j0,n0) + m2.innerVectors(j1,n0); 209 if(SparseMatrixType::IsRowMajor) 210 refMat2.middleRows(j0,n0) = (refMat2.middleRows(j0,n0) + refMat2.middleRows(j1,n0)).eval(); 211 else 212 refMat2.middleCols(j0,n0) = (refMat2.middleCols(j0,n0) + refMat2.middleCols(j1,n0)).eval(); 213 214 VERIFY_IS_APPROX(m2, refMat2); 215 } 216 217 // test generic blocks 218 { 219 DenseMatrix refMat2 = DenseMatrix::Zero(rows, cols); 220 SparseMatrixType m2(rows, cols); 221 initSparse<Scalar>(density, refMat2, m2); 222 Index j0 = internal::random<Index>(0,outer-2); 223 Index j1 = internal::random<Index>(0,outer-2); 224 Index n0 = internal::random<Index>(1,outer-(std::max)(j0,j1)); 225 if(SparseMatrixType::IsRowMajor) 226 VERIFY_IS_APPROX(m2.block(j0,0,n0,cols), refMat2.block(j0,0,n0,cols)); 227 else 228 VERIFY_IS_APPROX(m2.block(0,j0,rows,n0), refMat2.block(0,j0,rows,n0)); 229 230 if(SparseMatrixType::IsRowMajor) 231 VERIFY_IS_APPROX(m2.block(j0,0,n0,cols)+m2.block(j1,0,n0,cols), 232 refMat2.block(j0,0,n0,cols)+refMat2.block(j1,0,n0,cols)); 233 else 234 VERIFY_IS_APPROX(m2.block(0,j0,rows,n0)+m2.block(0,j1,rows,n0), 235 refMat2.block(0,j0,rows,n0)+refMat2.block(0,j1,rows,n0)); 236 237 Index i = internal::random<Index>(0,m2.outerSize()-1); 238 if(SparseMatrixType::IsRowMajor) { 239 m2.innerVector(i) = m2.innerVector(i) * s1; 240 refMat2.row(i) = refMat2.row(i) * s1; 241 VERIFY_IS_APPROX(m2,refMat2); 242 } else { 243 m2.innerVector(i) = m2.innerVector(i) * s1; 244 refMat2.col(i) = refMat2.col(i) * s1; 245 VERIFY_IS_APPROX(m2,refMat2); 246 } 247 248 Index r0 = internal::random<Index>(0,rows-2); 249 Index c0 = internal::random<Index>(0,cols-2); 250 Index r1 = internal::random<Index>(1,rows-r0); 251 Index c1 = internal::random<Index>(1,cols-c0); 252 253 VERIFY_IS_APPROX(DenseVector(m2.col(c0)), refMat2.col(c0)); 254 VERIFY_IS_APPROX(m2.col(c0), refMat2.col(c0)); 255 256 VERIFY_IS_APPROX(RowDenseVector(m2.row(r0)), refMat2.row(r0)); 257 VERIFY_IS_APPROX(m2.row(r0), refMat2.row(r0)); 258 259 VERIFY_IS_APPROX(m2.block(r0,c0,r1,c1), refMat2.block(r0,c0,r1,c1)); 260 VERIFY_IS_APPROX((2*m2).block(r0,c0,r1,c1), (2*refMat2).block(r0,c0,r1,c1)); 261 262 if(m2.nonZeros()>0) 263 { 264 VERIFY_IS_APPROX(m2, refMat2); 265 SparseMatrixType m3(rows, cols); 266 DenseMatrix refMat3(rows, cols); refMat3.setZero(); 267 Index n = internal::random<Index>(1,10); 268 for(Index k=0; k<n; ++k) 269 { 270 Index o1 = internal::random<Index>(0,outer-1); 271 Index o2 = internal::random<Index>(0,outer-1); 272 if(SparseMatrixType::IsRowMajor) 273 { 274 m3.innerVector(o1) = m2.row(o2); 275 refMat3.row(o1) = refMat2.row(o2); 276 } 277 else 278 { 279 m3.innerVector(o1) = m2.col(o2); 280 refMat3.col(o1) = refMat2.col(o2); 281 } 282 if(internal::random<bool>()) 283 m3.makeCompressed(); 284 } 285 if(m3.nonZeros()>0) 286 VERIFY_IS_APPROX(m3, refMat3); 287 } 288 } 289 } 290 291 void test_sparse_block() 292 { 293 for(int i = 0; i < g_repeat; i++) { 294 int r = Eigen::internal::random<int>(1,200), c = Eigen::internal::random<int>(1,200); 295 if(Eigen::internal::random<int>(0,4) == 0) { 296 r = c; // check square matrices in 25% of tries 297 } 298 EIGEN_UNUSED_VARIABLE(r+c); 299 CALL_SUBTEST_1(( sparse_block(SparseMatrix<double>(1, 1)) )); 300 CALL_SUBTEST_1(( sparse_block(SparseMatrix<double>(8, 8)) )); 301 CALL_SUBTEST_1(( sparse_block(SparseMatrix<double>(r, c)) )); 302 CALL_SUBTEST_2(( sparse_block(SparseMatrix<std::complex<double>, ColMajor>(r, c)) )); 303 CALL_SUBTEST_2(( sparse_block(SparseMatrix<std::complex<double>, RowMajor>(r, c)) )); 304 305 CALL_SUBTEST_3(( sparse_block(SparseMatrix<double,ColMajor,long int>(r, c)) )); 306 CALL_SUBTEST_3(( sparse_block(SparseMatrix<double,RowMajor,long int>(r, c)) )); 307 308 r = Eigen::internal::random<int>(1,100); 309 c = Eigen::internal::random<int>(1,100); 310 if(Eigen::internal::random<int>(0,4) == 0) { 311 r = c; // check square matrices in 25% of tries 312 } 313 314 CALL_SUBTEST_4(( sparse_block(SparseMatrix<double,ColMajor,short int>(short(r), short(c))) )); 315 CALL_SUBTEST_4(( sparse_block(SparseMatrix<double,RowMajor,short int>(short(r), short(c))) )); 316 } 317 } 318