Oops, it seems that if I go to the later version of root this works! Thanks! -y ---------- Dr. Yuri Gershtein, Physics, Brown University. Based at Fermilab. Tel: (630)-840-3784. Fax: (630)-840-8886 E-Mail: Gerstein@fnal.gov On Sat, 8 Jun 2002, Rene Brun wrote: > Hi Yuri, > > An example exclude.C: > //file exclude.C > Double_t fline(Double_t *x, Double_t *par) > { > if (x[0] > 2.5 && x[0] < 3.5) { > TF1::RejectPoint(); > return 0; > } > return par[0] + par[1]*x[0]; > } > > void exclude() { > TF1 *f1 = new TF1("f1","[0] +[1]*x +gaus(2)",0,5); > f1->SetParameters(6,-1,5,3,0.2); > TH1F *h = new TH1F("h","background + signal",100,0,5); > h->FillRandom("f1",2000); > TF1 *fline = new TF1("fline",fline,0,5,2); > fline->SetParameters(2,-1); > h->Fit("fline","l"); > } > > > Rene Brun > > On Sat, 8 Jun 2002, Yuri Gershtein wrote: > > > Hi, > > > > so I have a peak which sits on a low statistics background, so > > I should use log likelihood. > > I want to fit just background without the peak, say with [1]*x+[0]. > > > > Can I define a non-contiguos range for the fit? > > Setting errors to zero doesn't seem to work for the log likelihood (I > > think in PAW it used to work, i.e. non-zero bin with zero error was > > excluded from the fit). > > > > Also, in PAW I knew how to control the FCN calculation, and though this > > was not graceful, it worked... > > > > Thanks for your help, > > Yuri > > > > > > ---------- > > Dr. Yuri Gershtein, Physics, Brown University. > > Based at Fermilab. Tel: (630)-840-3784. Fax: (630)-840-8886 > > E-Mail: Gerstein@fnal.gov > > > >
This archive was generated by hypermail 2b29 : Sat Jan 04 2003 - 23:50:56 MET