[RF] Extend testSumW2Error to use an actual weighted dataset

The test has also been updated to use a simpler model without the
RooWorkspace, and the initial fit parameters are kicked away from the
generator parameters to get realistic fit behavior.

Author: guitargeek

roofit/roofit/test/testSumW2Error.cxx

@@ -3,7 +3,9 @@
 
 #include "RooFitResult.h"
 #include "RooAbsPdf.h"
-#include "RooWorkspace.h"
+#include "RooGaussian.h"
+#include "RooExponential.h"
+#include "RooAddPdf.h"
 #include "RooRandom.h"
 #include "RooDataHist.h"
 #include "RooDataSet.h"
@@ -17,51 +19,55 @@ TEST(SumW2Error, BatchMode)
    auto &msg = RooMsgService::instance();
    msg.setGlobalKillBelow(RooFit::WARNING);
 
-   RooWorkspace ws{"workspace"};
-
-   auto *x = ws.factory("x[-10, 10]");
-   ws.factory("Gaussian::sig(x, mu[-1, 1], s[0.1, 5])");
-   ws.factory("Chebychev::bkg(x, {c1[0.1, -1, 1]})");
-   auto *shp = static_cast<RooAbsPdf *>(ws.factory("SUM::shp(Nsig[0, 20000] * sig, Nbkg[0, 20000] * bkg)"));
-   auto &model = *shp;
-
-   // parameters
-   auto *mu = ws.var("mu");
-   auto *s = ws.var("s");
-   auto *c1 = ws.var("c1");
-   auto *Nsig = ws.var("Nsig");
-   auto *Nbkg = ws.var("Nbkg");
-
-   auto resetParameters = [&]() {
-      mu->setVal(0.0);
-      mu->setError(0.0);
-      s->setVal(2.0);
-      s->setError(0.0);
-      c1->setVal(0.1);
-      c1->setError(0.0);
-      Nsig->setVal(10000.0);
-      Nsig->setError(0.0);
-      Nbkg->setVal(10000.0);
-      Nbkg->setError(0.0);
+   RooRealVar x{"x", "x", 0, 0, 10};
+   RooRealVar mu{"mu", "mu", 3, 0, 10};
+   RooRealVar s{"s", "s", 1.0, 0.1, 5};
+   RooRealVar c1{"c1", "c1", -0.5, -3, -0.1};
+   RooRealVar f{"f", "f", 0.2, 0.0, 1.0};
+
+   RooGaussian sig{"sig", "sig", x, mu, s};
+   RooExponential bkg{"bkg", "bkg", x, c1};
+   RooAddPdf model{"model", "model", {sig, bkg}, {f}};
+
+   auto resetParametersToInitialFitValues = [&]() {
+      mu.setVal(4.0);
+      mu.setError(0.0);
+      s.setVal(2.0);
+      s.setError(0.0);
+      c1.setVal(-0.4);
+      c1.setError(0.0);
+      f.setVal(0.3);
+      f.setError(0.0);
    };
 
    std::size_t nEvents = 1000;
 
    RooRandom::randomGenerator()->SetSeed(4357);
-   std::unique_ptr<RooAbsData> dataHist{shp->generateBinned(*x, nEvents)};
-   std::unique_ptr<RooAbsData> dataSet{shp->generate(*x, nEvents)};
+   std::unique_ptr<RooAbsData> dataHist{model.generateBinned(x, nEvents)};
+   std::unique_ptr<RooAbsData> dataSet{model.generate(x, nEvents)};
+
+   // these datasets will be filled with a weight that is not unity
+   RooRealVar weight("weight", "weight", 0.5, 0.0, 1.0);
+   RooDataHist dataHistWeighted("dataHistWeighted", "dataHistWeighted", x);
+   RooDataSet dataSetWeighted("dataSetWeighted", "dataSetWeighted", {x, weight}, "weight");
+
+   for (std::size_t i = 0; i < nEvents; ++i) {
+      dataHistWeighted.add(*dataSet->get(), 0.5); // filling the histogram from a dataset is easier
+      dataSetWeighted.add(*dataSet->get(), 0.5);
+   }
 
    auto fit = [&](RooAbsData &data, bool sumw2 = false, bool batchmode = false, std::string const &minimizer = "Minuit",
                   int printLevel = -1) {
       using namespace RooFit;
 
-      resetParameters();
+      resetParametersToInitialFitValues();
 
-      return std::unique_ptr<RooFitResult>{model.fitTo(data, Extended(), Save(), SumW2Error(sumw2), Strategy(1),
+      return std::unique_ptr<RooFitResult>{model.fitTo(data, Save(), SumW2Error(sumw2), Strategy(1),
                                                        BatchMode(batchmode), Minimizer(minimizer.c_str()),
                                                        PrintLevel(printLevel))};
    };
 
+   // Compare batch mode vs. scalar mode for non-SumW2 fits on UNWEIGHTED datasets
    EXPECT_TRUE(fit(*dataSet, 0, 0, "Minuit")->isIdentical(*fit(*dataSet, 0, 1, "Minuit")))
       << " different results for Minuit fit to RooDataSet without SumW2Error correction.";
    EXPECT_TRUE(fit(*dataHist, 0, 0, "Minuit")->isIdentical(*fit(*dataHist, 0, 1, "Minuit")))
@@ -71,10 +77,12 @@ TEST(SumW2Error, BatchMode)
    EXPECT_TRUE(fit(*dataHist, 0, 0, "Minuit2")->isIdentical(*fit(*dataHist, 0, 1, "Minuit2")))
       << " different results for Minuit2 fit to RooDataHist without SumW2Error correction.";
 
-   // We can't compare the covariance matrix in these cases, because it is
+   // We can't compare the covariance matrix in these next cases, because it is
    // externally provided. Still, it's okay because the parameter values and
    // errors are compared, where the errors are inferred from the external
    // covariance matrix.
+
+   // Compare batch mode vs. scalar mode for SumW2 fits on UNWEIGHTED datasets
    EXPECT_TRUE(fit(*dataSet, 1, 0, "Minuit")->isIdenticalNoCov(*fit(*dataSet, 1, 1, "Minuit")))
       << " different results for Minuit fit to RooDataSet with SumW2Error correction.";
    EXPECT_TRUE(fit(*dataHist, 1, 0, "Minuit")->isIdenticalNoCov(*fit(*dataHist, 1, 1, "Minuit")))
@@ -83,4 +91,14 @@ TEST(SumW2Error, BatchMode)
       << " different results for Minuit2 fit to RooDataSet with SumW2Error correction.";
    EXPECT_TRUE(fit(*dataHist, 1, 0, "Minuit2")->isIdenticalNoCov(*fit(*dataHist, 1, 1, "Minuit2")))
       << " different results for Minuit2 fit to RooDataHist with SumW2Error correction.";
+
+   // Compare batch mode vs. scalar mode for SumW2 fits on WEIGHTED datasets
+   EXPECT_TRUE(fit(dataSetWeighted, 1, 0, "Minuit")->isIdenticalNoCov(*fit(dataSetWeighted, 1, 1, "Minuit")))
+      << " different results for Minuit fit to weighted RooDataSet with SumW2Error correction.";
+   EXPECT_TRUE(fit(dataHistWeighted, 1, 0, "Minuit")->isIdenticalNoCov(*fit(dataHistWeighted, 1, 1, "Minuit")))
+      << " different results for Minuit fit to weighted RooDataHist with SumW2Error correction.";
+   EXPECT_TRUE(fit(dataSetWeighted, 1, 0, "Minuit2")->isIdenticalNoCov(*fit(dataSetWeighted, 1, 1, "Minuit2")))
+      << " different results for Minuit2 fit to weighted RooDataSet with SumW2Error correction.";
+   EXPECT_TRUE(fit(dataHistWeighted, 1, 0, "Minuit2")->isIdenticalNoCov(*fit(dataHistWeighted, 1, 1, "Minuit2")))
+      << " different results for Minuit2 fit to weighted RooDataHist with SumW2Error correction.";
 }