const unsigned int n_dofs = local_source.size();
-//TODO[WB]: This function could me made more efficient. First, it allocates memory, which could be avoided by passing in another argument as a scratch array. second, the elementwise access is really slow if we use PETSc vectors/matrices. This should be fixed eventually
+//TODO[WB]: This function could me made more efficient. First, it allocates memory, which could be avoided by passing in another argument as a scratch array (or by using the dof_indices cache for non-hp DoFHandlers, see the functions in the Implementation class above). second, the elementwise access is really slow if we use PETSc vectors/matrices. This should be fixed eventually
// get indices of dofs
std::vector<unsigned int> dofs (n_dofs);
const unsigned int n_dofs = local_source.m();
-//TODO[WB]: This function could me made more efficient. First, it allocates memory, which could be avoided by passing in another argument as a scratch array. second, the elementwise access is really slow if we use PETSc vectors/matrices. This should be fixed eventually
+//TODO[WB]: This function could me made more efficient. First, it allocates memory, which could be avoided by passing in another argument as a scratch array (or by using the dof_indices cache for non-hp DoFHandlers, see the functions in the Implementation class above). second, the elementwise access is really slow if we use PETSc vectors/matrices. This should be fixed eventually
// get indices of dofs
std::vector<unsigned int> dofs (n_dofs);