...
Say you've finished all chains for particular set of parameters (all cells from 1 to the number of angle slices are green). We now want to consolidate those into a single larger file. As I write this, I notice that all 60 180 angle slice chains for R_{grid} = 60 28 and 70 for 26, R_{bin} = 250 150 are done (green), but have not been consolidated. So we'll use those as an example. Consolidating .
Consolidating these chains again involves running a slurm script. Specifically, the
ConsolidateSNeOnSkyFits.slurm script. The user will need to edit the following variables in this file:
rand
comoving_bin
fitter_ids
grid_densities
hemispheres_sets
angle_divs_set
The 'rand' variable should be either 0, if the chains being consolidated are from a real data set, or 1 if the chains being consolidated are from a bootstrapped data set
The 'comoving_bin' variable should be a single integer, whatever the size of the comoving bin is in the chains to be consolidated
The other 4 variables should actually be lists, formatted as [XX,YY,ZZ] (NOTE: the lack of spaces is essential!!!!)
The fitter_ids list should be whatever ids belong to the chains to be processed. For real data, this should always be [1]. For randomized data, this can be whatever the randomized fitter ids are. For example, [1,2,3,4,5]
The other three lists should be the same length, equal to the number of different R_{grid} variables you want to consolidate with this processing run.
The grid_densities variable should be the grid densities of each chain.
The hemispheres_sets variable should pretty much always be [both,both,both] , one both for each grid element
The angle_divs_set should be the number of slices into which the hemisphere is divided.
For example, to process the two data sets from we're interested in, I will set:
rand=0
comoving_bin=150
fitter_ids=[1]
grid_densities=[26,28]
hemispheres_sets=[both,both]
angle_divs_set=[180,180]
You can do so using vi:
vi ConsolidateSNeOnSkyFits.slurm A useful manual for basic vi editing tips can be found here: https://www.ele.uri.edu/faculty/vetter/Other-stuff/vi/text_edit.html
The commands you'll mostly need are:
-Entering edit mode: type i, then you can type characters and delete them at your leisure.
-Saving your edits: Type : and then w
-Exiting vi once you've saved your edits: Type : and then q
-Exiting vi WITHOUT SAVING (so without typing : and w before hand): Type :, then q, then !
With those edits made, let's execute the command:
$ sbatch ConsolidateSNeOnSkyFits.slurm
submitted batch job 19202558
Note the output file. When the batch job is done, open op the output file. The figures of merit (chi square improvement) are printed near the end of the output file.
vi slurm-19202558.out
Down at the bottom, we see the results:
For grid density: 26
Ratio of real to null r Chi Sqr = 0.16698040463400232
For grid density: 28
Ratio of real to null r Chi Sqr = 0.13617911931972834
Those two numbers are the values we copy and paste into the spreadsheet, under the Chi Sqr Ratio column. .