Failed Eigen assertion when testing with MKL enabled

[peddie]

[ RUN      ] test_core_dataset/DatasetOperatorTester/0.test_multiply_dataset
albatross_unit_tests: albatross/third_party/eigen/Eigen/src/Core/Assign_MKL.h:132: static void Eigen::internal::Assignment<Eigen::Array<double, -1, 1, 0>, Eigen::CwiseUnaryOp<Eigen::internal::scalar_sqrt_op<double>, const Eigen::ArrayWrapper<const Eigen::Matrix<double, -1, 1, 0>>>, Eigen::internal::assign_op<double, double>, Eigen::internal::Dense2Dense>::run(DstXprType &, const Eigen::internal::Assignment<type-parameter-0-0, CwiseUnaryOp<Eigen::internal::scalar_sqrt_op<double>, type-parameter-0-1>, Eigen::internal::assign_op<double, double>, Eigen::internal::Dense2Dense, typename enable_if<vml_assign_traits<DstXprType, SrcXprNested>::EnableVml, void>::type>::SrcXprType &, const assign_op<double, double> &) [DstXprType = Eigen::Array<double, -1, 1, 0>, SrcXprType = Eigen::CwiseUnaryOp<Eigen::internal::scalar_sqrt_op<double>, const Eigen::ArrayWrapper<const Eigen::Matrix<double, -1, 1, 0>>>, Functor = Eigen::internal::assign_op<double, double>, Kind = Eigen::internal::Dense2Dense, EnableIf = void]: Assertion `dst.rows() == src.rows() && dst.cols() == src.cols()' failed.
Aborted

Looks like a mistaken assumption about sizes somewhere. This failure also occurs in:

• test_multiply_with_matrix_joint
• test_multiply_with_matrix_marginal
• test_multiply_with_sparse_matrix_joint
• test_multiply_with_sparse_matrix_marginal
• test_multiply_with_vector

[peddie]

This is triggered by the matrix with 0 rows at https://github.com/swift-nav/albatross/blob/master/tests/test_core_dataset.cc#L193:

struct EmptyTransform {
  auto get_matrix(Eigen::Index n) {
    Eigen::SparseMatrix<double> matrix(0, n);
    return matrix;
  }
};

Not sure yet whether we are doing something wrong or Eigen is, but http://gitlab.com/libeigen/eigen/-/issues/1135 suggests that this should probably work.

[peddie]

With the fix above, I don't get any assert failures, but I still see the warning Intel MKL ERROR: Parameter 6 was incorrect on entry to DGEMV . I'm not clear whether this is a problem for us in practice if the results for the test are still computed correctly. Here is the MKL description:

MKL_VERBOSE DGEMV(N,0,3,0x7fff24289c28,(nil),0,0x1c4f740,1,0x7fff24289c30,(nil),1) 881ns CNR:OFF Dyn:1 FastMM:1 TID:0  NThr:8

We know that we ask for a matrix with 0 rows in Eigen, but argument 6 of DGEMV is the leading dimension (which should be the number of columns). I can imagine something like: Eigen has computed the leading dimension to be 0 based on the fact that there are 0 rows and thus 0 elements, but DGEMV still expects to see 3 in there.