Attached are instructions and scripts for carrying out a parameter space scan with ShelfMC, the simulation package for the ARIANNA detector.
Because some of the plotted outputs looked like colored stripes and did not offer any insight into how effective volume changed with some variables, I made some changes to the simulation and plotting scripts so that different maximum, minimum, and increment values can be chosen for each variable. Now rather than having fixed, hard-coded values for all variables, the parameter space scan and plotting is more flexible for use with variable inputs.
| Quote: |
|
I am trying to write a script that will plot a 2d histogram of effective volume versus two of ShelfMC's parameters.
The script prompts the user for which two parameters (out of five that we vary in our parameter space scan) to plot along the x- and y-axes, as well as what values to hold the other 3 parameters constant at. It then collects the necessary root files from simulation results, generates a plotting script, and runs the plotting script to produce a plot in pdf form.
After many struggles I have the script written to the point where it functions, but the plots don't look right. Some plots look like they could be actual data (like Veff_A_I), and others just look flat-out wrong (like Veff_R_S).
I have yet to pin down the cause of this, but hopefully will be able to sometime in the near future.
|
|
This document will explain how to dowload, configure, and run a parameter space search for ShelfMC on a computing cluster.
These scripts explore the ShelfMC parameter space by varying ATTEN_UP, REFLECT_RATE, ICETHICK, FIRNDEPTH, and STATION_DEPTH for certain rages.
The ranges and increments can be found in setup.sh.
In order to vary STATION_DEPTH, some changes were made to the ShelfMC code. Follow these steps to allow STATION_DEPTH to be an input parameter.
1.cd to ShelfMC directory
2.Do $sed -i -e 's/ATDepth/STATION_DEPTH/g' *.cc
3.Open declaration.hh. Replace line 87 "const double ATDepth = 0.;" with "double STATION_DEPTH;"
4.In functions.cc go to line 1829. This is the ReadInput() method. Add the lines below to the end of this method.
GetNextNumber(inputfile, number); // new line for station Depth
STATION_DEPTH = (double) atof(number.c_str()); //new line
5.Do $make clean all
#######Script Descriptions########
setup.sh -> This script sets up the necessary directories and setup files for all the runs
scheduler.sh -> This script submits and monitors all jobs.
#######DOWNLOAD########
1.Download setup.sh and scheduler.sh
2.Move both files into your ShelfMC directory
3.Do $chmod u+x setup.sh and $chmod u+x scheduler.sh
######CONFIGURE#######
1.Open setup.sh
2.On line 4, modify the job name
3.On line 6, modify group name
4.On line 11, specify your ShelfMC directory
5.On lines 15-33, specify the minimum and maximum values you wish to simulate for each variable, as well as what value to increment by
6.On line 36, modify your run name
7.On line 37, specify the NNU per run
8.On line 38, specify the starting seed
9.On line 40, specify the number of processors per node on your cluster
10.On lines 42-79, edit the input.txt parameters that you want to keep constant for every run
11.On line 80, specify the location of the LP_gain_manual.txt
12.On line 143, change walltime depending on total NNU. Remember this wall time will be 20x shorter than a single processor run.
13.On line 144, change job prefix
14.On line 146, change the group name if needed
15.Save file
16.Open scheduler.sh
17.On line 4, specify your ShelfMC directory
18.On line 5, modify run name. Make sure it is the same runName as you have in setup.sh
19.Ensure that lines 15-33 match with setup.sh
20.On lines 62 and 66, replace cond0092 with your username for the cluster
21.On line 69, you can pick how many nodes you want to use at any given time. It is set to 6 intially.
22.Save file
#######RUN#######
1.Do $qsub setup.sh
2.Wait for setup.sh to finish. This script is creating the setup files for all runs. This may take about an hour.
3.When setup.sh is done, there should be a new directory in your home directory. Move this directory to your ShelfMC directory.
4.Do $screen to start a new screen that the scheduler can run on. This is incase you lose connection to the cluster mid run.
5.Make sure that the minimum, maximum, and increment values for each variable match those in setup.sh
6.Do $./scheduler.sh to start script. This script automatically submits jobs and lets you see the status of the runs. This will run for several hours.
7.The scheduler makes a text file of all jobs called jobList.txt in the ShelfMC dir. Make sure to delete jobList.txt before starting a whole new run.
######RESULT#######
1.When Completed, there will be a great amount of data in the run files, about 460GB (This amount varies depending on how many values looped over for all variables.
2.The run directory is organized in tree, results for particular runs can be found by cd'ing deeper into the tree.
3.In each run directory, there will be a resulting root file, all the setup files, and a log file for the run.
|