29   using TMath::PoissonI;
 
   60       this->push_back(mean);
 
   61       this->push_back(sigma);
 
   62       this->push_back(signal);
 
   63       this->push_back(background);
 
   78   inline result_type 
g1(
const JGauss& 
gauss, 
const JElement_t& point) 
 
   82     const double u  = (point.getX() - gauss[0]) / gauss[1];
 
   83     const double fs =  gauss[2] * 
exp(-0.5*u*u) / (sqrt(2.0*
PI) * gauss[1]);
 
   84     const double fb =  gauss[3];
 
   85     const double f1 =  fs + fb;
 
   87     const double p  =  PoissonI(point.getY(), f1);
 
   89     result.chi2 = -log(p);
 
   91     result.gradient[0] = fs * u   / gauss[1];                                      
 
   92     result.gradient[1] = fs * u*u / gauss[1]  -  fs / gauss[1];                    
 
   93     result.gradient[2] = fs       / gauss[2];                                      
 
   94     result.gradient[3] = 1.0;                                                      
 
   96     result.gradient *= 1.0 - point.getY()/f1;                                      
 
  109 int main(
int argc, 
char **argv)
 
  122     JParser<> zap(
"Program to test JGandalf algorithm.");
 
  132   catch(
const exception& error) {
 
  133     FATAL(error.what() << endl);
 
  136   using namespace JFIT;
 
  138   ASSERT(numberOfEvents > 0);
 
  142   TF1 fs(
"fs", 
"exp(-0.5 * (x-[0])*(x-[0]) / ([1]*[1]))");
 
  145   fs.FixParameter(0, gauss[0]);
 
  146   fs.FixParameter(1, gauss[1]);
 
  149   const Double_t xmin = -5.0;
 
  150   const Double_t xmax = +5.0;
 
  157   TH1D      
H[] = { TH1D(
"ha", 
"", 101,   -0.1,    +0.1),
 
  158                     TH1D(
"hb", 
"", 101,   -0.1,    +0.1),
 
  159                     TH1D(
"hc", 
"", 101, -100.0,  +100.0),
 
  160                     TH1D(
"hd", 
"", 101, -100.0,  +100.0) };
 
  164   for (
int i = 0; i != numberOfEvents; ++i) {
 
  166     STATUS(
"event: " << setw(10) << i << 
'\r'); 
DEBUG(endl);
 
  168     TH1D h0(
"h0", NULL, nx, xmin, xmax);
 
  172     h0.FillRandom(
"fs", (Int_t) gauss[2]);
 
  173     h0.FillRandom(
"fb", (Int_t) gauss[3]);
 
  177     for (Int_t i = 1; i <= h0.GetNbinsX(); ++i) {
 
  178       data.push_back(JElement_t(h0.GetBinCenter (i),
 
  179                                 h0.GetBinContent(i)));
 
  186     for (
size_t i = 0; i != 4; ++i) {
 
  192                        h0.GetEntries() * (xmax - xmin) / nx - h0.GetMinimum(),
 
  199     const double chi2 = fit(
g1, data.begin(), data.end());
 
  204     DEBUG(
"Chi2 " << chi2 << endl);
 
  206     const double Y[] = { fit.
value[0]                       - gauss[0],
 
  207                          fit.
value[1]                       - gauss[1],
 
  208                          fit.
value[2] * nx / (xmax - xmin)  - gauss[2],
 
  209                          fit.
value[3] * nx                  - gauss[3] };
 
  211     for (
int i = 0; i != 
sizeof(Q)/
sizeof(Q[0]); ++i) {
 
  217   for (
int i = 0; i != 
sizeof(Q)/
sizeof(Q[0]); ++i) {
 
  222     timer.print(cout, 
true, 
micro_t);
 
  229     for (
int i = 0; i != 
sizeof(
H)/
sizeof(
H[0]); ++i) {
 
  237   for (
int i = 0; i != 
sizeof(Q)/
sizeof(Q[0]); ++i) {
 
  241   ASSERT(Q[0].getSTDev() < precision[0]);
 
  242   ASSERT(Q[1].getSTDev() < precision[1]);
 
  243   ASSERT(Q[2].getSTDev() < precision[2]);
 
  244   ASSERT(Q[3].getSTDev() < precision[3]);
 
Utility class to parse command line options. 
 
The elements in a collection are sorted according to their abscissa values and a given distance opera...
 
std::vector< parameter_type > parameters
 
double getMean(vector< double > &v)
get mean of vector content 
 
#define ASSERT(A,...)
Assert macro. 
 
I/O formatting auxiliaries. 
 
#define make_field(A,...)
macro to convert parameter to JParserTemplateElement object 
 
JGauss()
Default constructor. 
 
Auxiliary class for CPU timing and usage. 
 
General purpose messaging. 
 
Fit method based on the Levenberg-Marquardt method. 
 
std::ostream & longprint(std::ostream &out)
Set long printing. 
 
Utility class to parse command line options. 
 
Data structure for return value of fit function. 
 
double gauss(const double x, const double sigma)
Gauss function (normalised to 1 at x = 0). 
 
std::ostream & shortprint(std::ostream &out)
Set short printing. 
 
#define DEBUG(A)
Message macros. 
 
then set_variable FORMULA *[0] exp(-0.5 *(x-[1])*(x-[1])/([2]*[2]))" set_variable OUTPUT_FILE histogram.root JHistogram1D -o $WORKDIR/$OUTPUT_FILE -F "$FORMULA" -
 
Double_t g1(const Double_t x)
Function. 
 
int main(int argc, char *argv[])