Hello Hans,
I totally agree with your request.
This functionality is necessary for the developments planned in
PROOF (the Parallel Root facility) that we intend to extend this year.
There are several possibilities to implement it. I am expecting some
user
feedback on this point and some prototyping will also be necessary.
This selection function could be interpreted, compiled via the script
compiler or it could refer to an already precompiled function.
PROOF has an additional requirement concerning parallelism.
The scheme must be able to work in a realistic situation where
a user wants to analyze a TChain of TTrees (multi files on multi
systems)
and N slaves processing the query in parallel.
There are several possible user interfaces
- call to a simple C style function returning the selection (weight)
for a given entry, eg
T.Draw("var","select.C",...
T.Draw("var","select.C(4.5)",...
T.Draw("var","select(...)",..
T.Draw("var","var1>3 && select(6)"
T.Draw("select1(2.5):select2(8):var3","select4.C++(5)",...
- a member function (say Select) of a class generated by MakeClass
- ???
Currently, you can already implement your function by using the class
generated by TTree::MakeClass (Loop function).
TTree::Draw is OK for simple things, but you can generate only one
histogram
at the time.
The Loop function is more general. You can for example loop on a first
TAG Tree and for the selected entries read entries in other Trees.
All these considerations should be taken into account before
implementing
your request.
Rene Brun
Hans Wenzel (lokal) wrote:
>
> hi
>
> would be nice if one could pass a logical function as an argument in
> CopyTree. E.g. in the example below i don't like the fact that i have to
> hardwire the cut values where i would like to use variables. In addition
> the selection might become more complicated.
>
> newtree =t->fTree->CopyTree("Chiprob>0.01&&Mass>5.129&&Mass<5.429");
>
> cheers
>
> hans
> --
> ________________________________________________________________________________
> Hans Wenzel | Phone:(721) 608 3418
> Inst. für Experimentelle Kernphysik | (721) 96453 45
> (home)
> Universität Karlsruhe | Fax: (721) 607 262
> .~. L
> Engesserstr. 7 | email: wenzel@fnal.gov
> /V\ I
> D-76128 Karlsruhe | http://kcdf1.fnal.gov/~wenzel/
> // \\ N
> Germany |
> /( )\ U
> ______________________________________________________________________
> ^`~'^__X_
This archive was generated by hypermail 2b29 : Tue Jan 02 2001 - 11:50:20 MET