"Fit data is empty" 当 `TH1::SetDefaultSumw2` 开启时

"Fit data is empty" when `TH1::SetDefaultSumw2` is on

我有来自 NaI(Tl) 检测器的数据,该检测器具有大量的 bin 和每个 bin 的低计数(大约 5-30)。

我写了这个函数来提取数据并绘制它;然而,当我试图减去背景时,我得到 "Warning in : Fit data is empty"。这仅在我 TH1::SetDefaultSumw2(kTRUE) 未注释时发生。

为什么会发生这种情况(是我的信号太弱了吗?)我能否解决此问题同时仍确保正确的错误传播?

数据太长,无法内嵌;但是,我已经托管了文件:Co60 and Background.

代码(文件名Graph1.C):

#include <vector>
#include <cmath>
#include "TCanvas.h"
#include "TROOT.h"
#include "TGraphErrors.h"
#include "TH1.h"
#include "TF1.h"
#include "TLegend.h"
#include "TArrow.h"
#include "TLatex.h"

#define NBINS 16384
#define LOWBIN 0
#define HIGHBIN NBINS

using std::vector;
using std::ifstream;

void Graph1(char const* filename, char const* bgfilename, char const* title,
            int start_bin = LOWBIN, int end_bin = HIGHBIN) {
  TH1::SetDefaultSumw2(kTRUE);
  gStyle->SetOptFit();
  std::ifstream in;
  in.open(filename);

  std::ifstream bg;
  bg.open(bgfilename);

  std::string titleString = title + std::string(";Bin;Count");

  TH1D *counts = new TH1D("Counts", titleString.c_str(), NBINS, LOWBIN, HIGHBIN);
  TH1D bgcounts("Bg", "", NBINS, LOWBIN, HIGHBIN);

  int bin = 0;

  while (in.good() && bg.good()) {
    double temp;
    in >> temp;
    counts->AddBinContent(bin, temp);
    bg >> temp;
    bgcounts.AddBinContent(bin, temp);
    ++bin;
  }

  counts->Add(&bgcounts, -1.0);

  TF1 *fit = new TF1("fit", "gaus", start_bin, end_bin);
  counts->Fit(fit, "R");
  counts->GetXaxis()->SetRange(start_bin, end_bin);

  auto mycanvas = new TCanvas();
  mycanvas->SetGrid();

  counts->SetStats(false);
  counts->Draw("C E");

  mycanvas->Update();
  mycanvas->Modified();
}

Root 被调用:

.x Graph1.C("Co60.TKA", "Background.TKA", "Co60", 12200, 13500)

使用 "LL" 拟合参数解决了问题:

An improved Log Likelihood fit in case of very low statistics and when bincontentsare not integers. Do not use this option if bin contents are large (greater than 100).

https://root.cern.ch/root/html534/guides/users-guide/FittingHistograms.html#the-fit-method