relion
relion copied to clipboard
MPI Parallelize training of Bayesian Polishing
This can be trivially parallelized by MPI.
We can calculate target function values per movie over MPI and gather it to the rank 0 node to run optimization.
Jasenko is working on this.
I agree. This would be very helpful to speed up the training step. Any estimate when this feature will be implemented?
I have no idea. Jasenko left the group, although he will keep contributing to RELION development.