This package implements an analysis of ATLAS Experiment data designed to look for Higgs bosons produced in association with dark matter particles. The Higgs decay is identified by a diphoton resonance, while the dark matter particle would manifest as missing transverse energy in the detector.
The code has been structured so that all of the general analysis settings are stored in the configuration file in the data/ directory. *A typical user should only need to adjust the config file. Changes to categorizations are temporary exceptions, and should be added to the DMEvtSelect class. The code is designed to be as automatic as possible. Each class looks to see if all of the necessary inputs have been produced previously before generating them from scratch.
The DMMaster.cxx program is the user interface for the analysis. All other classes and macros can be run through this program by specifying the proper job option and config file settings. A full list of the job options is given below in the "Running the code" section.
New users will need to modify data/settingsHDM_sys.cfg and lxbatch scripts in order to run the code with full functionality. All input and output file locations for the programs are specified in the settings (masterInput, masterOutput, packageLocation, clusterFileLocation, and fileName*). Sub- directories will be created as necessary by the program.
The only input files currently required are centrally-produced HGamma group MxAODs (h008 tag or later). From these, the entire analysis can be generated.
First compile the master program, which executes the analysis code:
> make bin/DMMaster
Then to run,
> ./bin/DMMaster <Program> <SettingsFile>
The program can be any of the following options:
- Cleanup (clean old files from previous analysis runs)
- MassPoints (make mass files as inputs for and model)
- GetSystematics (make cutflows and categorizations for all syst. variations)
- RankSystematics (make a ranking of systematic uncertainties for each sample)
- PlotVariables (plot interesting kinematic variables)
- SigParam (build the signal PDF from MC)
- BkgModel (build the background model)
- Workspace (build the statistical model)
- ResubmitWorkspace (submit failed Workspace jobs again)
- TossPseudoExp (toss pseudo experiment ensemble)
- PlotPseudoExp (plot the results of pseudo experiments)
- TestStat (calculate p0 and CLs)
- ResubmitTestStat (submit failed TestStat jobs again)
- MuLimit (get the 95% CL limit on the parameter of interest)
- Optimizer (optimize the analysis selection with meta job)
- OptAnalysis (analyze the results of Optimizer)
The code will automatically run any required upstream programs in order to ensure that it has all required inputs. For instance, if you want to create a workspace from scratch, just run,
./bin/DMMaster Workspace data/settingsHDM.cfg
This is true for every program EXCEPT PlotVariables and MuLimit, which are separate executables.
Make sure that you are running in a directory from which EOS is accesssible.
Luminosity, higgs mass, m_yy range, file names, script locations, production mode information should all go here. Default job options also are included. The sys tag refers to the fact that MxAODs with systematic variations can now be specified. Be careful: there are differences between the nominal and systematic MxAOD path specifications. Specifically, one must specify the entire file location for systematics MxAODs (root://eosatlas//eos...), since the files are spread across many locations.
This namespace should store all general analysis methods. The idea is to avoid duplication of methods to minimize the pain of changing things.
This is the master 'wrapper' class for the analysis. Using this class, all the analysis tools can be run. The file organization is automated, using a directory structure based on the 'masterInput' and 'masterOutput' strings in settingsHDM, as well as the masterJobName.
This program uses a TTree of data events to produce a series of mass points that can be used as inputs for the signal parameterization or workspace creation. The cutflow is implemented using the DMEvtSelect class.
This program uses signal MC to fit the resonance shape for the SM Higgs and the DM signal and saves the parameters for use in workspace generation. The fit is performed on masspoints generated with DMMassPoints. Signal cross- sections are provided by the BRXSReader tool.
This program implements all of the possible background models, and can return either a RooAbsPdf object, a CombinedPdf, or add a PDF directly to the analysis workspace. It is designed to be generic enough for use in all H->yy analyses.
This program produces the statistical model for the DM analysis. It includes SM signal, a single DM signal, and background PDFs, as well as associated systematic uncertainties. The parameter of interest is "mu_DM", the signal strength for the dark matter production process. The signal strength of the Standard Model Higgs Boson "mu_SM" is set to 1. The background normalization comes from data. There is the option of fitting the SM signal strengths individually.
This program calculates the 95% CL, CLs, and p0 values for a given DM signal. It is the default fitting program.
This program implements the pseudo-dataset generation and fitting. It is designed to either run locally or on a cluster, and outputs a TTree file.
This program has tools for analyzing toy MC data.
Reads tables of SM Higgs cross sections and branching ratios and provides an easy-to-use interface.
Checks to see whether the output files for a given program are available. It also assembles a list of failed jobs that can be prepared for resubmission.
This class implements the cutflow and counters for the analysis. It can be initialized using a pointer to the DMTree.
This class is automatically generated based on the MxAOD structure. It provides a useful interface for code to the TTrees. It has been modified from the class which is automatically spit out from the TTree::MakeClass method in order to simplify the systematic uncertainties implementation.
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent