Logo ROOT  
Reference Guide
Go to the documentation of this file.
1/// \file
2/// \ingroup tutorial_roostats
3/// \notebook
4/// StandardTestStatDistributionDemo.C
6/// This simple script plots the sampling distribution of the profile likelihood
7/// ratio test statistic based on the input Model File. To do this one needs to
8/// specify the value of the parameter of interest that will be used for evaluating
9/// the test statistic and the value of the parameters used for generating the toy data.
10/// In this case, it uses the upper-limit estimated from the ProfileLikleihoodCalculator,
11/// which assumes the asymptotic chi-square distribution for -2 log profile likelihood ratio.
12/// Thus, the script is handy for checking to see if the asymptotic approximations are valid.
13/// To aid, that comparison, the script overlays a chi-square distribution as well.
14/// The most common parameter of interest is a parameter proportional to the signal rate,
15/// and often that has a lower-limit of 0, which breaks the standard chi-square distribution.
16/// Thus the script allows the parameter to be negative so that the overlay chi-square is
17/// the correct asymptotic distribution.
19/// \macro_image
20/// \macro_output
21/// \macro_code
23/// \author Kyle Cranmer
25#include "TFile.h"
26#include "TROOT.h"
27#include "TH1F.h"
28#include "TCanvas.h"
29#include "TSystem.h"
30#include "TF1.h"
31#include "TSystem.h"
33#include "RooWorkspace.h"
34#include "RooAbsData.h"
48using namespace RooFit;
49using namespace RooStats;
51bool useProof = false; // flag to control whether to use Proof
52int nworkers = 0; // number of workers (default use all available cores)
54// -------------------------------------------------------
55// The actual macro
57void StandardTestStatDistributionDemo(const char *infile = "", const char *workspaceName = "combined",
58 const char *modelConfigName = "ModelConfig", const char *dataName = "obsData")
61 // the number of toy MC used to generate the distribution
62 int nToyMC = 1000;
63 // The parameter below is needed for asymptotic distribution to be chi-square,
64 // but set to false if your model is not numerically stable if mu<0
65 bool allowNegativeMu = true;
67 // -------------------------------------------------------
68 // First part is just to access a user-defined file
69 // or create the standard example file if it doesn't exist
70 const char *filename = "";
71 if (!strcmp(infile, "")) {
72 filename = "results/example_combined_GaussExample_model.root";
73 bool fileExist = !gSystem->AccessPathName(filename); // note opposite return code
74 // if file does not exists generate with histfactory
75 if (!fileExist) {
76#ifdef _WIN32
77 cout << "HistFactory file cannot be generated on Windows - exit" << endl;
78 return;
80 // Normally this would be run on the command line
81 cout << "will run standard hist2workspace example" << endl;
82 gROOT->ProcessLine(".! prepareHistFactory .");
83 gROOT->ProcessLine(".! hist2workspace config/example.xml");
84 cout << "\n\n---------------------" << endl;
85 cout << "Done creating example input" << endl;
86 cout << "---------------------\n\n" << endl;
87 }
89 } else
90 filename = infile;
92 // Try to open the file
93 TFile *file = TFile::Open(filename);
95 // if input file was specified byt not found, quit
96 if (!file) {
97 cout << "StandardRooStatsDemoMacro: Input file " << filename << " is not found" << endl;
98 return;
99 }
101 // -------------------------------------------------------
102 // Now get the data and workspace
104 // get the workspace out of the file
105 RooWorkspace *w = (RooWorkspace *)file->Get(workspaceName);
106 if (!w) {
107 cout << "workspace not found" << endl;
108 return;
109 }
111 // get the modelConfig out of the file
112 ModelConfig *mc = (ModelConfig *)w->obj(modelConfigName);
114 // get the modelConfig out of the file
115 RooAbsData *data = w->data(dataName);
117 // make sure ingredients are found
118 if (!data || !mc) {
119 w->Print();
120 cout << "data or ModelConfig was not found" << endl;
121 return;
122 }
124 mc->Print();
125 // -------------------------------------------------------
126 // Now find the upper limit based on the asymptotic results
127 RooRealVar *firstPOI = (RooRealVar *)mc->GetParametersOfInterest()->first();
128 ProfileLikelihoodCalculator plc(*data, *mc);
129 LikelihoodInterval *interval = plc.GetInterval();
130 double plcUpperLimit = interval->UpperLimit(*firstPOI);
131 delete interval;
132 cout << "\n\n--------------------------------------" << endl;
133 cout << "Will generate sampling distribution at " << firstPOI->GetName() << " = " << plcUpperLimit << endl;
134 int nPOI = mc->GetParametersOfInterest()->getSize();
135 if (nPOI > 1) {
136 cout << "not sure what to do with other parameters of interest, but here are their values" << endl;
137 mc->GetParametersOfInterest()->Print("v");
138 }
140 // -------------------------------------------------------
141 // create the test stat sampler
144 // to avoid effects from boundary and simplify asymptotic comparison, set min=-max
145 if (allowNegativeMu)
146 firstPOI->setMin(-1 * firstPOI->getMax());
148 // temporary RooArgSet
149 RooArgSet poi;
150 poi.add(*mc->GetParametersOfInterest());
152 // create and configure the ToyMCSampler
153 ToyMCSampler sampler(ts, nToyMC);
154 sampler.SetPdf(*mc->GetPdf());
155 sampler.SetObservables(*mc->GetObservables());
156 sampler.SetGlobalObservables(*mc->GetGlobalObservables());
157 if (!mc->GetPdf()->canBeExtended() && (data->numEntries() == 1)) {
158 cout << "tell it to use 1 event" << endl;
159 sampler.SetNEventsPerToy(1);
160 }
161 firstPOI->setVal(plcUpperLimit); // set POI value for generation
162 sampler.SetParametersForTestStat(*mc->GetParametersOfInterest()); // set POI value for evaluation
164 if (useProof) {
165 ProofConfig pc(*w, nworkers, "", false);
166 sampler.SetProofConfig(&pc); // enable proof
167 }
169 firstPOI->setVal(plcUpperLimit);
170 RooArgSet allParameters;
171 allParameters.add(*mc->GetParametersOfInterest());
172 allParameters.add(*mc->GetNuisanceParameters());
173 allParameters.Print("v");
175 SamplingDistribution *sampDist = sampler.GetSamplingDistribution(allParameters);
176 SamplingDistPlot plot;
177 plot.AddSamplingDistribution(sampDist);
178 plot.GetTH1F(sampDist)->GetYaxis()->SetTitle(
179 Form("f(-log #lambda(#mu=%.2f) | #mu=%.2f)", plcUpperLimit, plcUpperLimit));
180 plot.SetAxisTitle(Form("-log #lambda(#mu=%.2f)", plcUpperLimit));
182 TCanvas *c1 = new TCanvas("c1");
183 c1->SetLogy();
184 plot.Draw();
185 double min = plot.GetTH1F(sampDist)->GetXaxis()->GetXmin();
186 double max = plot.GetTH1F(sampDist)->GetXaxis()->GetXmax();
188 TF1 *f = new TF1("f", Form("2*ROOT::Math::chisquared_pdf(2*x,%d,0)", nPOI), min, max);
189 f->Draw("same");
190 c1->SaveAs("standard_test_stat_distribution.pdf");
#define f(i)
Definition: RSha256.hxx:104
#define gROOT
Definition: TROOT.h:406
char * Form(const char *fmt,...)
R__EXTERN TSystem * gSystem
Definition: TSystem.h:559
Int_t getSize() const
RooAbsArg * first() const
virtual void Print(Option_t *options=0) const
This method must be overridden when a class wants to print itself.
RooAbsData is the common abstract base class for binned and unbinned datasets.
Definition: RooAbsData.h:49
virtual Int_t numEntries() const
Return number of entries in dataset, i.e., count unweighted entries.
Definition: RooAbsData.cxx:307
Bool_t canBeExtended() const
If true, PDF can provide extended likelihood term.
Definition: RooAbsPdf.h:238
virtual Double_t getMax(const char *name=0) const
Get maximum of currently defined range.
RooArgSet is a container object that can hold multiple RooAbsArg objects.
Definition: RooArgSet.h:29
Bool_t add(const RooAbsArg &var, Bool_t silent=kFALSE) override
Add element to non-owning set.
Definition: RooArgSet.cxx:261
RooRealVar represents a variable that can be changed from the outside.
Definition: RooRealVar.h:39
void setMin(const char *name, Double_t value)
Set minimum of name range to given value.
Definition: RooRealVar.cxx:486
virtual void setVal(Double_t value)
Set value of variable to 'value'.
Definition: RooRealVar.cxx:282
LikelihoodInterval is a concrete implementation of the RooStats::ConfInterval interface.
Double_t UpperLimit(const RooRealVar &param)
return the upper bound of the interval on a given parameter
ModelConfig is a simple class that holds configuration information specifying how a model should be u...
Definition: ModelConfig.h:30
const RooArgSet * GetGlobalObservables() const
get RooArgSet for global observables (return NULL if not existing)
Definition: ModelConfig.h:255
const RooArgSet * GetParametersOfInterest() const
get RooArgSet containing the parameter of interest (return NULL if not existing)
Definition: ModelConfig.h:237
const RooArgSet * GetNuisanceParameters() const
get RooArgSet containing the nuisance parameters (return NULL if not existing)
Definition: ModelConfig.h:240
virtual void Print(Option_t *option="") const override
overload the print method
const RooArgSet * GetObservables() const
get RooArgSet for observables (return NULL if not existing)
Definition: ModelConfig.h:249
RooAbsPdf * GetPdf() const
get model PDF (return NULL if pdf has not been specified or does not exist)
Definition: ModelConfig.h:234
The ProfileLikelihoodCalculator is a concrete implementation of CombinedCalculator (the interface cla...
ProfileLikelihoodTestStat is an implementation of the TestStatistic interface that calculates the pro...
Holds configuration options for proof and proof-lite.
Definition: ProofConfig.h:46
This class provides simple and straightforward utilities to plot SamplingDistribution objects.
Double_t AddSamplingDistribution(const SamplingDistribution *samplingDist, Option_t *drawOptions="NORMALIZE HIST")
adds the sampling distribution and returns the scale factor
void Draw(Option_t *options=0)
Draw this plot and all of the elements it contains.
void SetAxisTitle(char *varName)
TH1F * GetTH1F(const SamplingDistribution *samplDist=NULL)
Returns the TH1F associated with the give SamplingDistribution.
This class simply holds a sampling distribution of some test statistic.
const std::vector< Double_t > & GetSamplingDistribution() const
Get test statistics values.
ToyMCSampler is an implementation of the TestStatSampler interface.
Definition: ToyMCSampler.h:74
The RooWorkspace is a persistable container for RooFit projects.
Definition: RooWorkspace.h:43
RooAbsData * data(const char *name) const
Retrieve dataset (binned or unbinned) with given name. A null pointer is returned if not found.
void Print(Option_t *opts=0) const
Print contents of the workspace.
TObject * obj(const char *name) const
Return any type of object (RooAbsArg, RooAbsData or generic object) with given name)
Double_t GetXmax() const
Definition: TAxis.h:134
Double_t GetXmin() const
Definition: TAxis.h:133
The Canvas class.
Definition: TCanvas.h:23
1-Dim function class
Definition: TF1.h:213
A ROOT file is a suite of consecutive data records (TKey instances) with a well defined format.
Definition: TFile.h:54
static TFile * Open(const char *name, Option_t *option="", const char *ftitle="", Int_t compress=ROOT::RCompressionSetting::EDefaults::kUseCompiledDefault, Int_t netopt=0)
Create / open a file.
Definition: TFile.cxx:3997
TAxis * GetXaxis()
Get the behaviour adopted by the object about the statoverflows. See EStatOverflows for more informat...
Definition: TH1.h:320
TAxis * GetYaxis()
Definition: TH1.h:321
virtual void SetTitle(const char *title="")
Set the title of the TNamed.
Definition: TNamed.cxx:164
virtual const char * GetName() const
Returns name of object.
Definition: TNamed.h:47
virtual Bool_t AccessPathName(const char *path, EAccessMode mode=kFileExists)
Returns FALSE if one can access a file using the specified access mode.
Definition: TSystem.cxx:1294
return c1
Definition: legend1.C:41
The namespace RooFit contains mostly switches that change the behaviour of functions of PDFs (or othe...
Namespace for the RooStats classes.
Definition: Asimov.h:19
static constexpr double pc
Definition: file.py:1