Ratio32Theory

This will collect instructions for running theoretical predictions for the R32 analysis.

NNLO predictions

The NNLO predictions are divided into four channels (numbered 1 to 4) for eight different contributions (DU0, DU1, FR0, FR1, SU0, SU1, RVF, RRF). These have to be summed up on the top of:

• LO predictions: (1,2,3,4 - B)

• NLO predictions: (1,2,3,4 - UC0, UC1, VF, RF).

The basic syntax for running these predictions on the grid is the following

prun \
--noEmail \
--architecture ’@centos7’ \
--containerImage llorente/threejet-nnlo:0.7 \
--exec "runR32.sh <njet> <channel> <contribution> <pdf-order> <time> %RNDM:<base-seed>" \
--outDS group.phys-sm.R32_NNLO_nj<njet>_ch<channel>_<contribution>_v0001 \
--outputs "*.xml" \
--noBuild \
--official \
--nJobs=30000

Where:

• njet is 2 or 3, depending if you're running numerators or denominators for R32.

• channel is 1,2,3,4. All of them should be ran separately.

• contribution is the contribution, described above.

• pdf-order is 3 for LO and NLO, and 2 at NNLO (it will calculate all eigenvectors for LO and NLO only, but not for NNLO as this would blow up computing time).

• time is the time you want your job to run, in seconds. Normally we use 50000 for ~14 hour runs.

• base-seed is the seed of the first job in the jobset. When using %RNDM, it gets incremented for each of the 30000 jobs. One should change the base seed for each job that one sends with the same channel and contribution in order to avoid repeated events. I suggest to start with 100000000 (10^8) as base seed, then increment it by 50000 (50K) for each jobset that is sent.