Introduction

This resource contains the full statistical model for CMS Combine ref v10.0.2 related to the paper "Simultaneous probe of the charm and bottom quark Yukawa couplings using ttH events", by the CMS collaboration.

The instructions below include the commands needed to reproduce the workspaces used for the results presented in the publication. For additional technical documentation please consult the CMS Combine documentation.

Requirements

Since the datacards (besides txt files) often include binary files such as ROOT files, the repository is set up to use Git Large File Storage to track them. This makes cloning the repository much more efficient. In case that you do not have git lfs installed on your system, please follow the git lfs installation instructions.

For the commands below, you may require running ulimit -s unlimited; ulimit -u unlimited to avoid memory issues.

Analysis specific instructions

The instructions below include the Combine version, commands used to create RooFit workspaces and the combine commands to produce the analysis measurements. All of the relevant options for combine commands are documented as well. Combine version: v10.0.2

Physics models

This section contains a brief description of the physics models used in the analysis and the references to the implementation within combine.

Results for ttH and ttZ signal strengths

We use the standard multiSignalModel as physics model and set the POIs in the text2workspace.py command via -P HiggsAnalysis.CombinedLimit.PhysicsModel:multiSignalModel -v 0 --channel-masks --PO 'map=.*/ttH_hbb:rate_ttHbb[1.,-10.,10.]' --PO 'map=.*/ttH_hcc:rate_ttHcc[1.,-10.,10.]' --PO 'map=.*/ttZ_zbb:rate_ttZbb[1.,-10.,10.]' --PO 'map=.*/ttZ_zcc:rate_ttZcc[1.,-10.,10.]'. No additional model file is needed. The POIs of the analysis are rate_ttHcc, rate_ttHbb, rate_ttZcc, and rate_ttZbb.

The complete command to create the workspace is:

text2workspace.py datacard.txt \
-o workspace.root \
-m 125.38 \
-P HiggsAnalysis.CombinedLimit.PhysicsModel:multiSignalModel \
-v 0 \
--channel-masks \
--PO map=.*/ttH_hbb:rate_ttHbb[1.,-10.,10.] \
--PO map=.*/ttH_hcc:rate_ttHcc[1.,-10.,10.] \
--PO map=.*/ttZ_zbb:rate_ttZbb[1.,-10.,10.] \
--PO map=.*/ttZ_zcc:rate_ttZcc[1.,-10.,10.]' \
--for-fits \
--no-wrappers \
--use-histsum \
--X-pack-asympows \
--optimize-simpdf-constraints=cms

Results in the kappa-framework

For the interpretation in the kappa-framework, we use the standard K1C.py physics model provided in the models folder. The POIs of the 1D (2D) interpretation are kappa_c (and kappa_b).

The command to create the respective workspace is:

text2workspace.py datacard.txt \
-o workspace.root \
-m 125.38 \
-P HiggsAnalysis.CombinedLimit.LHCHCGModels:K1C \
--for-fits \
--no-wrappers \
--use-histsum \
--X-pack-asympows \
--optimize-simpdf-constraints=cms \
--channel-masks

For performing the fits in the kappa framework, the datacards have to be slightly modified. The signal processes ttZ_zcc and ttZ_zbb have to be downgraded to background processes, hence the processIDs have to be changed from -3 and -2 to 17 and 16. In addition, we have to insert the following lines to the combined card to leave the background rates freely floating:

SF_norm_ttZbb                              rateParam  * ttZ_zbb      1
SF_norm_ttZcc                              rateParam  * ttZ_zcc      1

Datacard description

We add the full combined datacard, which includes the validation region (VR), control regions (CRs), and signal regions (SRs) of all channels (fully hadronic (0L/FH), semileptonic (1L/SL), and dileptonic (2L/DL)), as described in the publication. The sum of the CRs and SRs used in the signal extraction is denoted as analysis region (AR).

The naming scheme of the regions is the following: (DL|SL|FH)_ttHcc_FR2_(DL|SL|FH)_cat(BB|BJ|CC|CJ|LF|Hcc|Hbb|Zcc|Zbb)_(SR|MidScoreVR), where (BB|BJ|CC|CJ|LF|Hcc|Hbb|Zcc|Zbb) stands for a category in the VR/AR enriched in tt+bb/cc/bj/cj/lf production or one of the signal processes (ttH(cc), ttH(bb), ttZ(cc), ttZ(bb)).

Running the fit

We use the functionality of channel masks in Combine for the individual fits. In the following we explain how to run the fit for the VR and the AR (final signal extraction fit).

To run the VR fit with the POIs being the background normalization scale factors:

combine \
 -M MultiDimFit \
 --algo singles \
 --saveFitResult \
 --saveWorkspace \
 --saveNLL \
 --X-rtd REMOVE_CONSTANT_ZERO_POINT=1 \
 -m 125.38 \
 --robustFit 1 \
 --cminDefaultMinimizerStrategy 0 \
 --X-rtd MINIMIZER_MaxCalls=999999999 \
 --cminDefaultMinimizerTolerance 0.1 \
 --X-rtd FAST_VERTICAL_MORPH \
 --cminPreScan \
 --cminPreFit 1 \
 -n _nominal_exp_fit \
 -d workspace_root. \
 -t -1 \
 --setParameters rgx{CMS_HIG24018_SFnorm_.*}=1.,rgx{rate_tt.*}=1.,rgx{mask_.*_SR.*}=1 \
 --setParameterRanges rgx{CMS_HIG24018_SFnorm_.*}=-3.,3.:rgx{rate_ttZbb.*}=-5.,5.:rgx{rate_ttZcc.*}=-5.,5.:rgx{rate_ttHbb.*}=-5.,5.:rgx{rate_ttHcc.*}=-50.,50. \
--redefineSignalPOIs CMS_HIG24018_SFnorm_ttcc,CMS_HIG24018_SFnorm_ttcj,CMS_HIG24018_SFnorm_ttbb,CMS_HIG24018_SFnorm_ttbj,CMS_HIG24018_SFnorm_ttlf \
 --freezeParameters rate_ttHcc,rate_ttHbb,rate_ttZcc,rate_ttZbb

And for the AR (i.e. CRs+SRs) with 4 POIs for ttH(cc), ttH(bb), ttZ(cc), and ttZ(bb):

combine \
 -M MultiDimFit \
 --algo singles \
 --saveFitResult \
 --saveWorkspace \
 --saveNLL \
 --X-rtd REMOVE_CONSTANT_ZERO_POINT=1 \
 -m 125.38 \
 --robustFit 1 \
 --cminDefaultMinimizerStrategy 0 \
 --X-rtd MINIMIZER_MaxCalls=999999999 \
 --cminDefaultMinimizerTolerance 0.1 \
 --X-rtd FAST_VERTICAL_MORPH \
 --cminPreScan \
 --cminPreFit 1 \
 -n _nominal_exp_fit \
 -d workspace.root \
 -t -1 \
 --setParameters rgx{CMS_HIG24018_SFnorm_.*}=1.,rgx{rate_tt.*}=1.,rgx{mask_.*_MidScoreVR.*}=1 \
 --setParameterRanges rgx{CMS_HIG24018_SFnorm_.*}=-3.,3.:rgx{rate_ttZbb.*}=-5.,5.:rgx{rate_ttZcc.*}=-5.,5.:rgx{rate_ttHbb.*}=-5.,5.:rgx{rate_ttHcc.*}=-50.,50. \
 --redefineSignalPOIs rate_ttHcc,rate_ttHbb,rate_ttZcc,rate_ttZbb

Combination with HIG-21-008

We further combine the results of this analysis with CMS-HIG-21-008, the Search for Higgs boson decay to a charm quark-antiquark pair in proton-proton collisions at sqrt{s} = 13 TeV. Those datacards are found in input_combination/input_HIG-21-008 and modified cards for CMS-HIG-24-018 can be found in input_combination/input_HIG-24-018, as the combination only targets the interpretation using the kappa framework. The combined card can be found in input_combination/combination.txt.

Fits can be run analougous to the standalone HIG-24-018 results discussed above.