Closed joesilber closed 4 years ago
Working for me. Except the script plot_posparams is now deprecated, I think.
Structurally this looks good. Thanks for these updates. It would be nice if @julienguy could confirm that it still works for him prior to merging to catch numpy version etc. stuff, but if merging this is a blocking factor for @joesilber it's fine with me to merge and patch up any compatibility issues post-facto.
Except the script plot_posparams is now deprecated, I think.
That script should work
Structurally this looks good. Thanks for these updates. It would be nice if @julienguy could confirm that it still works for him prior to merging to catch numpy version etc. stuff, but if merging this is a blocking factor for @joesilber it's fine with me to merge and patch up any compatibility issues post-facto.
It's not blocking me, I can start another branch as needed.
This branch addresses issue #64
The positioner parameter fitting is handled for each positioner in a new module
fithandler.py
.The multiprocessing batches are much simpler to manage: for each positioner, shoot a data table over to the
fithandler
module and let it run.Long jobs are more robust as well. For each positioner, a separate data table is written as soon as it has been processed. This is vastly more memory efficient than before (running all 5020 tables in memory at once) and a crash doesn't lose all the work.