Hi Jeff,
You are using the wrong TTree::Branch function.
There are 3 Branch functions:
a, one for basic types
b, one for objects
c, one for TClonesArray
You create objects, you should use mode b. In this mode
you have to specify the class name and the address of a pointer
to an object. One could also provide a function accepting
an object by reference. I did not implement this function
because in the reality one create objects via new.
In attachment 1, you will find a modification of your program
to process your example. This will produce the file charm.root.
Note that I have specified split mode=0 because you want
to use your objects (data and functions).
You can run the following Root session:
Root > TFile f("charm.root")
Root > T.Draw("ppsi.Mag()")
In attachment 2, you will find a variant using the split mode.
Note the "." at the end of the branch names. If you use split
mode on branches with the same class, you must distinguish
the fX attribute of scat from the fX of ppsi.
You can run the following session:
Root > TFile f("charm.root")
Root > T.Draw("ppsi.fX")
In the two examples above, I use one of the Root C++ extensions,
a shortcut for
Root > TTree *T = (TTree*)f.Get("T");
Root > T->Draw("ppsi.fX")
The shortcut mode is OK when you type command by command.
If you write a macro, we recommend the C++ standard way
if you intend to compile your macro at a later stage via
the script compiler for instance.
It would be better to have a class (say Event) having
4 TLorentzVector objects as data members.
In this case, you will have only one call to TTree::Branch,
Root taking care of splitting the 4 objects into 4 separate
branches.
Your comments clearly indicates that many things are missing
in the documentation explaining the various ways to create a
TTree. I will try to use your nice example in a future upgrade
of the documentation.
Rene Brun
On Fri, 28 Jan 2000, Jeff Templon wrote:
> Hi,
>
> OK, so I am trying to use ROOT (I thought it would be easier than
> writing a Python wrapper for ROOT trees ... turns out I was right,
> keep reading ...).
>
> I wrote a macro which ran the very first time (!) which is supposed to
> generate a tree full of simulated J/psi electroproduction events.
> Each event will contain four TLorentzVector objects. This apparently
> happened as far as I can tell.
>
> Pain #1 - reattaching the Tree in the ROOT file seems to be much more
> complex than hi/fil 1 charm.root was
>
> Pain #2 - I don't see an easy way to go thru the entire Tree and plot
> say the magnitudes of the four-momenta (I thought the tree was
> supposed to keep all the object code intact which is why we were using
> C++ ... oh dear).
>
> TFile f("charm.root");
> TTree *mytree = (TTree*)f.Get("T"); // what a contorted command
> mytree.Draw("ppsi.fY"); // yields some error
> Error in <TCanvas::Range>: illegal world coordinates range: x1=13.684271, y1=-0.131250, x2=-1.520475, y2=1.181250
> Error in <TCanvas::RangeAxis>: illegal axis coordinates range: xmin=12.163796, ymin=0.000000, xmax=-0.000000, ymax=1.050000
> mytree.Draw("ppsi.fZ")
> Error in <TCanvas::Range>: illegal world coordinates range: x1=13.684271, y1=-0.393750, x2=-1.520475, y2=3.543750
> Error in <TCanvas::RangeAxis>: illegal axis coordinates range: xmin=12.163796, ymin=0.000000, xmax=-0.000000, ymax=3.150000
> mytree.Draw("ppsi.Mag()")
>
> *ERROR 26 :
> Unknown name : "ppsi.Mag()"
> Error in <TCanvas::Range>: illegal world coordinates range: x1=13.684271, y1=-0.393750, x2=-1.520475, y2=3.543750
> Error in <TCanvas::RangeAxis>: illegal axis coordinates range: xmin=12.163796, ymin=0.000000, xmax=-0.000000, ymax=3.150000
>
> ppsi.Mag() is the biggest disappointment. What is the use of having
> all these objects in the tree if I can't use their methods?
>
> Pain #3 - wading thru the documentation gives me no clue of how to do
> what I want to do, except for possibly the contorted, non-simple
> method shown under "The General Way" on URL
>
> http://root.cern.ch/root/HowtoReadTree.html
>
> Say it ain't so!!! Isn't OO supposed to make my life easier?
>
> JT
>
> ps In the very possible case that I have already blown it in my macro,
> here it is.
>
> // #-> charm.py -- simulate j/psi electroproduction at JLab
> // c++ version i hope
>
> // from Numeric import * [ left over from python version ]
> // from vec4 import *
> // import RNG
>
> // #-> simulation driver parameters
>
> main(int argc, char **argv)
> {
> float mj = 3096.88; // # J/psi mass in MeV
> float me = 0.510999; // # electron mass
> float mp = 938.272;
>
> float beamen = 11.0 * 1000.0; // # GeV * 1000 MeV/GeV
>
> float ll_the = (3.1415926 / 180) * 1.0; // last number is e- angle LL in deg
> float ul_the = (3.1415926 / 180) * 20.0; // last number is e- angle UL in deg
>
> // above is stored in radians since we will need to take cosines
> // below is not
>
> float ll_phe = -93.0; // LL on e- azimuthal angle (degrees now)
> float ul_phe = -87.0;
>
> float ll_pe = 2275; // LL on scat electron momentum (MeV/c)
> float ul_pe = 2780;
>
> int howmany = 10000; // # desired number of events
>
> // #-> set up random generators for electron. strategy: create sample
> // #-> arrays at start of program, just take values from them during
> // #-> execution
>
> TF1 *thegen = new TF1("thegen","sin(x)",ll_the,ul_the);
> TF1 *phegen = new TF1("phegen","1.0", ll_phe,ul_phe);
> TF1 *pegen = new TF1("pegen", "1.0", ll_pe, ul_pe );
>
> // #-> set up four-momenta for beam and target, which should not change
>
> TLorentzVector beam; beam.SetVectM(TVector3(0,0,beamen),me);
> TLorentzVector targ(0.0,0.0,0.0,mp);
>
> TFile hfile("charm.root","RECREATE","Charm");
>
> TTree *tree = new TTree("T","Charm Electroproduction");
>
> int nev = 0;
> TLorentzVector scat(1,1,1,1);
> TLorentzVector X;
>
> tree->Branch("beam",&beam,"fX/D:fY/D:fZ/D:fE/D");
> tree->Branch("targ",&targ,"fX/D:fY/D:fZ/D:fE/D");
> tree->Branch("scat",&scat,"fX/D:fY/D:fZ/D:fE/D");
> tree->Branch("ppsi",&X,"fX/D:fY/D:fZ/D:fE/D");
>
> for ( int nev = 1; nev <= howmany; nev++) {
> float thtmp = thegen.GetRandom();
> float phtmp = phegen.GetRandom();
> float ptmp = pegen.GetRandom();
>
> scat.SetVectM(TVector3(ptmp*sin(thtmp)*cos(phtmp),
> ptmp*sin(thtmp)*sin(phtmp),
> ptmp*cos(thtmp)),
> me);
>
> X = beam + targ - scat;
>
> tree->Fill();
> }
>
> hfile.Write();
> hfile.Close();
>
> return 0;
> }
>
This archive was generated by hypermail 2b29 : Tue Jan 02 2001 - 11:50:17 MET