From 4e61a6400fe7eae171fec954f22f1c5127580074 Mon Sep 17 00:00:00 2001
From: Jonas Rembser <jonas.rembser@cern.ch>
Date: Sun, 25 Sep 2022 12:23:44 +0200
Subject: [PATCH] [RF] New unit test for `SplitRange()` feature in
 `RooAbsPdf::fitTo()`

The recent commit 922dec5157 fixed GitHub issue #11396, but there was
still no unit test implemented with the reproducer code from that issue.
This commit is implementing such a unit test.
---
 .../roofitcore/test/testRooSimultaneous.cxx   | 68 +++++++++++++++++++
 1 file changed, 68 insertions(+)

diff --git a/roofit/roofitcore/test/testRooSimultaneous.cxx b/roofit/roofitcore/test/testRooSimultaneous.cxx
index bfe6a37b67500..10816c982afdd 100644
--- a/roofit/roofitcore/test/testRooSimultaneous.cxx
+++ b/roofit/roofitcore/test/testRooSimultaneous.cxx
@@ -129,3 +129,71 @@ TEST(RooSimultaneous, CategoriesWithNoPdf)
    m1.setError(0.0);
    sim.fitTo(*ds, BatchMode(true), PrintLevel(-1));
 }
+
+/// GitHub issue #11396.
+/// Test whether the RooFit::SplitRange() command argument for simultaneous
+/// fits is correctly considered in multi-range fits.
+TEST(RooSimultaneous, MultiRangeFitWithSplitRange)
+{
+   using namespace RooFit;
+
+   int nEventsInCat = 11000;
+
+   RooWorkspace wsCat1{"wsCat1"};
+   wsCat1.factory("Gaussian::pdf_cat1(x_cat1[0,10],muCat1[4,0,10],sigma_cat1[1.0,0.1,10.0])");
+   RooAbsPdf &pdfCat1 = *wsCat1.pdf("pdf_cat1");
+   RooRealVar &xCat1 = *wsCat1.var("x_cat1");
+   xCat1.setRange("SideBandLo_cat1", 0, 2);
+   xCat1.setRange("SideBandHi_cat1", 6, 10);
+   std::unique_ptr<RooDataSet> dsCat1{pdfCat1.generate(xCat1, nEventsInCat)};
+
+   RooWorkspace wsCat2{"wsCat2"};
+   wsCat2.factory("Gaussian::pdf_cat2(x_cat2[0,10],muCat2[6,0,10],sigma_cat2[1.0,0.1,10.0])");
+   RooAbsPdf &pdfCat2 = *wsCat2.pdf("pdf_cat2");
+   RooRealVar &xCat2 = *wsCat2.var("x_cat2");
+   xCat2.setRange("SideBandLo_cat2", 0, 4);
+   xCat2.setRange("SideBandHi_cat2", 8, 10);
+   std::unique_ptr<RooDataSet> dsCat2{pdfCat1.generate(xCat2, nEventsInCat)};
+
+   RooCategory indexCat{"cat", "cat"};
+   indexCat.defineType("cat1");
+   indexCat.defineType("cat2");
+
+   RooSimultaneous simPdf{"simPdf", "", indexCat};
+   simPdf.addPdf(pdfCat1, "cat1");
+   simPdf.addPdf(pdfCat2, "cat2");
+
+   std::map<std::string, RooDataSet *> dsmap{{"cat1", dsCat1.get()}, {"cat2", dsCat2.get()}};
+
+   RooDataSet combData{"combData", "", {xCat1, xCat2}, Index(indexCat), Import(dsmap)};
+
+   std::unique_ptr<RooAbsReal> nll1{pdfCat1.createNLL(*dsCat1, Range("SideBandLo_cat1,SideBandHi_cat1"))};
+   std::unique_ptr<RooAbsReal> nll2{pdfCat2.createNLL(*dsCat2, Range("SideBandLo_cat2,SideBandHi_cat2"))};
+   std::unique_ptr<RooAbsReal> nllSim{simPdf.createNLL(combData, Range("SideBandLo,SideBandHi"), SplitRange())};
+
+   const double nll1Val = nll1->getVal();
+   const double nll2Val = nll2->getVal();
+   // In simultaneous PDFs, the probability is normalized over the categories,
+   // so we have to do that as well when computing the reference value. Since
+   // we do a ranged fit, we have to consider the ranges when calculating the
+   // number of events in data.
+   double nCat1InRange = std::unique_ptr<RooAbsData>
+   {
+      dsCat1->reduce(CutRange("SideBandLo_cat1,SideBandHi_cat1"))
+      } -> sumEntries();
+   double nCat2InRange = std::unique_ptr<RooAbsData>
+   {
+      dsCat2->reduce(CutRange("SideBandLo_cat2,SideBandHi_cat2"))
+      } -> sumEntries();
+   const double nllSimRefVal = (nCat1InRange + nCat2InRange) * std::log(2) + nll1Val + nll2Val;
+
+   const double nllSimVal = nllSim->getVal();
+
+   // For now, we can't cross check the likelihood value and the unit test
+   // only checks whether the simultaneous likelihood can be crated with the
+   // SplitRange() argument without the range overlap checks failing. Once the
+   // issue with multi-range simultaneous likelihood values is fixed, this can
+   // be commented out:
+   //
+   //  EXPECT_FLOAT_EQ(nllSimVal, nllSimRefVal);
+}