Daniel Seemaier

https://github.com/KaHIP/KaHIP/blob/master/parallel/parallel_src/lib/io/parallel_graph_io.cpp#L585 Let `n=1382867`, `size=2048`, `peID=2046`. Then, `from=1383096`, `to=min(1383772, 1382867-1)=1382866`, thus `local_no_nodes=-229` (actually it overflows since it's a `ULONG`) and ParHIP crashes in line 585. This PR should work for any...

Build: `cmake .. -DSTRONG_LV_SOCIAL=On` Run: `env LD_LIBRARY_PATH=./ ./kaffpa --preconfiguration=slvsocial --k=4 ~/graphs/cnr-2000.graph` With `STRONG_LV_SOCIAL` enabled, CMake clones the VieClus repository and builds a shared library from it (see PR in the...

Use enum names in `config_to_str()` output?

1

This is essentially the `enum_ostream` example with a `config_to_str()` call added: ``` #include #include #include #include "CLI11.hpp" enum class Level : int { High, Medium, Low }; std::ostream &operator

Clear hash map / erase all elements

2

Is there a fast way to delete all elements of a growable or non-growable hash map without freeing its memory?

Implement edge weights

1

It would be nice to have an option for edge weights based on the distance between adjacent vertices.

Currently, graphs are always generated with 64 bit edge and vertex IDs. Would be nice if we had a CMake flag to use 32 bit data types instead.

Currently, KaGen can only use one thread per MPI process. Since the generators are (mostly) communication free, we could implement additional OpenMPI parallelization without too much hustle.

Currently, the edge partitioning code ignores vertex and edge weights. This PR adds rudimentary support, **but**: for weighted input graphs, it might be harder to find "good" values for the...

See https://github.com/KIT-ITI10/kamping/pull/208#discussion_r838639081

Use up-to-date clang-format-version and reformat.

2

Bad formatting: ``` C++ template struct is_builtin_mpi_op< kamping::ops::min, T, typename std::enable_if::type> : std::true_type { static MPI_Op op() { return MPI_MIN; } }; ``` Bad formatting: ``` C++ template struct UserOperationPtr...