Chris Richardson

The problem with the `Eigen::` structure is that it is not that easy to dynamically add data. cf: ``` Eigen::Matrix a; ... a.conservativeResize(a.rows() + 1, NoChange); a.bottomRows()

What c++ compiler?

Can you post the errors?

Maybe it should have `#include ` since `std::sort` is used in `MPI.h` - strange that this did not come up before, though, and is not an issue on the CI...

OK, it's probably just due to machine precision and rounding. It is a bit difficult to control. Maybe we need to supply the precision as an argument.

This is a good idea, let's do it once the current work on MeshTag refinement is in main.

There are questions about scalability of this code. When every process communicates with every other process, it becomes a quadratic scaling, which is not good. Obviously, it can be difficult...

Would be useful and convenient for time-stepping problems.

What should go in the citation? The FEniCS Book (which is gradually becoming obsolete)? Or should we push to zenodo, get a doi, and use that as the citation? (which...

This is due to the serial implementation using the same code as in parallel. For example, it calculates the dual graph for partitioning the mesh (not needed in serial).