cellular level mesh partitioning
From the microcard task force implementation meeting rose the question: how to partition the mesh for the cellular level simulations? I am trying to summarize the main requirements from the meeting, so we can start the implementation, however I feel that my understanding is still limited.
The subcellular level simulations require a mesh partitioning for distribution of the computation between different processor ranks. For that we need to implement an algorithm, which will take into account the geometry of the intracellular and extracellular space and create the partitions. The division should follow the requirements:
- a cell should not be split between the partitions, a separate partition is created for each cell
- partition boundaries should be aligned with the myocyte membrane
- each element has a tag number associated with a partition (the tag clearly distinguishes between extracellular and intracellular, e.g. 1 for intracellular space of myocyte and 101 for extracellular space associated with the myocyte with tag 1)
- a single rank can compute several partitions with different tags
- WP7 provides input data for the subdivision of the extracellular space (attribution of the myocyte to the data)
- the interfaces between subdomains should be specified separately (e.g. if there are 100 sets for the cells, there will be 100 interfaces for each of them)
Please correct me, if some of the requirements are wrong and add further requirements that may be missing in the list (@fatemeh, @axel.loewe ).