Skip to content

Latest commit

 

History

History
104 lines (70 loc) · 2.81 KB

README.md

File metadata and controls

104 lines (70 loc) · 2.81 KB
Raw

Onetime setup

./install.sh

Loading the "launch" environment

This is the environment you use for preparing inputs and launching workers.

source setup.sh

Entering the GENIE container

This is the environment you use for running gspladd.

shifter --image=mjkramer/sim2x2:genie_edep.LFG_testing.20230228 /bin/bash --init-file /environment

Note that the run_gmkspl.sh script is meant to be called from outside the container (e.g. by spline_worker.py). Internally, the script enters the container.

Making splines

These instructions assume you're working with the LFG_testing.20230228 tag of GENIE and the D22_22a_02_11b tune. Feel free to change those values in run_gmkspl.sh and in the commands below.

First, free nucleons

Prepare input for free nucleons

mkdir -p input/nucleons
./prep_input.py --mode nucleons > input/nucleons/input.txt

Generate splines for free nucleons

seq 12 | parallel -n0 -u ./spline_worker.py \
    -i input/nucleons/input.txt -o splines/nucleons

It should take about two hours in the worst case, so do this in a screen session.

Merge splines for free nucleons

From inside the container:

gspladd -d splines/nucleons -o splines/D22_22a_02_11b.nucleons.LFG_testing.20230228.spline.xml

This merged file is used as input when generating the splines for nuclei.

Then, various nuclei

Prepare input for nuclei

First, open up prep_input.py and make sure that the NUCLEI variable contains all of the targets listed in the GENIE max path file for the geometry you plan to simulate.

mkdir -p input/nuclei
./prep_input.py --mode nuclei > input/nuclei/input.txt

Generate splines for nuclei

seq 48 | parallel -n0 -u ./spline_worker.py \
    -i input/nuclei/input.txt -o splines/nuclei \
    --xsec splines/D22_22a_02_11b.nucleons.LFG_testing.20230228.spline.xml

This took a day or two for the 21 nuclei in the MINERvA+2x2 geometry. Note that there are six spline files (one per neutrino signed-flavor) for each nucleus, so in that case, there were 126 total files to produce.

For faster turnaround, you can increase the nunber of workers beyond 24 (ideally on a batch node). You're free to add more workers at any point.

Merge all of the splines

From inside the container:

gspladd -d splines/nuclei -o splines/D22_22a_02_11b_nuclei_LFG_testing.20230228_spline.xml

gspladd -f splines/D22_22a_02_11b_nucleons_LFG_testing.20230228_spline.xml,splines/D22_22a_02_11b_nuclei_LFG_testing.20230228_spline.xml \
    -o splines/D22_22a_02_11b_all_LFG_testing.20230228_spline.xml

It looks like the second step is redundant; the merged all and nuclei files are identical, so I guess the free nucleon cross sections are already embedded in the spline files for each nucleus.