Dynamic Bonds #31
Description
Implement dynamic bonds based on a distance criterion between two groups (which may either have no overlap or may be 100% identical).
The basic idea is to split this into roughly 3 steps:
- Find all possible bonds by using a neighbor list. This list should not contain duplicates. We should use a separate neighbor list for the plugin because otherwise there might be funny unwanted interactions if particle interactions or neighbor list parameters change elsewhere in the code. We also might be able to make this very efficient because we are only interested in the intersection of the two groups, so we do only need the tree and not a verlet list and all the other overhead associated with a full neighbor list. T
- Select bonds from all possible bonds based on some criterion (only N bonds per particle, only M bonds per step, ... ) Here we also need to keep track of all existing bonds in the system so that we can compare against it, should be done in a way which is transferable between CPU/CPU, so no vectors/maps.
- Form the bonds by using m_bond_data->addBondedGroup, update the exclusions for the neighbor list handling the particle interactions.
Right now, there is no easy way to add or remove bonds to the BondedGroupData , except for adding them on the CPU with m_bond_data->addBondedGroup. Therefore, we'll first try step 1 and 2 on the GPU and see how the performance is. A bond reservoir will cause issues when used with MPI. One potential useful extension would be to write a function which can add multiple bonded groups at once to avoid paying overhead multiple times.
While groups are very flexible, in MPI we will have to make sure that the buffer layer is large enough to be able to communicate all potential bond partners between ranks ( using particle types would be easier for this).
It might be interesting to keep track of the distance between possible bonding pairs as well to be able to have selection criteria based on this as well, e.g. closer bonds are formed first.