Re: vectors in selection cuts in TTrees

From: Philippe Canal <pcanal_at_fnal.gov>
Date: Thu, 30 Sep 2010 15:10:27 -0500

  Hi,

> I am not clear about something though, the above formula
> would mean the effective loop looks something like this:

> for(int i=0; i<mc_n;i++) {
> ....
> for (int j=0; j<mc_parent_index->size(); j++) {

This is not correct. The implied look in your case really has only _one_ running index which is used for both array dereferencing. (There would be a 2nd implied look if you replace the

    mc_pdgId[ mc_parent_index[][0] ]
by

    mc_pdgId[ mc_parent_index[][] ]

> I don't want to lose any of the branches. I want to copy the entire
> event (meaning all branches for the matching entry).

You don't lose any branches. My point is that you copy the _entire_ event whenever at least one of the muon matches the criteria. (and hence the resulting file with have both muons that do match and muon that do not matches the criteria).

> It might not work if there were more muons from other sources.

I think it should still work.

Cheers,
Philippe.

On 9/30/10 2:31 PM, suvayu ali wrote:

Hi Philippe,
> On 30 September 2010 07:06, Philippe Canal<pcanal_at_fnal.gov> wrote:

>>   Hi,
>>
>> CopyTree("abs(mc_pdgId[]) == 13&&    abs(mc_pdgId[
>> mc_parent_index[Iteration$][0] ])== 24")
>>
>> is actually equivalent to:
>>
>> bool keepentry = false;
>> for(int i=0; i<mc_n;i++) {
>>    if ( abs(mc_pdgId[i]) = 13
>> &&  mc_parent_index->size()>0
>> &&  mc_parent_index->at(i).size()>0
>> &&  abs(mc_pdgId[ mc_parent_index[i][0] ])== 24 )
>>    {
>>       keepentry = true;
>>    }
>> }
>> if (keepentry) newtree->Fill();
>>

> Okay.

>
>> Note that CopyTree can only copy entire Events, so at best you would get a
>> TTree
>> with all the event when at least one 'element' match the criterium.
>>

> I don't want to lose any of the branches. I want to copy the entire
> event (meaning all branches for the matching entry). Given my
> intention, is my use of CopyTree correct? Or is there a more efficient
> way to achieve the same?

>
>> A priori, the formula you wrote seems correct, you could try this slight
>> variation:
>>
>>     CopyTree("abs(mc_pdgId[]) == 13&&    abs(mc_pdgId[ mc_parent_index[][0]
>> ])== 24")
>>

> Thank you for this suggestion. It seems to gives me the desired
> entries. I am not clear about something though, the above formula
> would mean the effective loop looks something like this:

>

> bool keepentry = false;
> for(int i=0; i<mc_n;i++) {
> if ( abs(mc_pdgId[i]) = 13&& mc_parent_index->size()>0) {
> for (int j=0; j<mc_parent_index->size(); j++) {
> if (abs(mc_pdgId[ mc_parent_index[j][0] ])== 24 ) {
> keepentry = true;
> }
> }
> }
> }
> if (keepentry) newtree->Fill();

>

> Although I am getting the desired result, I have a feeling it only
> works in my case because there is only one muon with a W boson as a
> parent. It might not work if there were more muons from other sources.
> Is my understanding correct here?

>
>> Also you can investigate whether the formula is return the value you expect
>> by using
>>
>>    physics->Scan("abs(mc_pdgId[]) == 13:abs(mc_pdgId[ mc_parent_index[][0]
>> ])== 24:abs(mc_pdgId[]) == 13&&  abs(mc_pdgId[mc_parent_index[][0] ])==
>> 24");
>>

> This tip was incredibly useful. Thanks a lot for all the help.

>
>> Cheers,
>> Philippe.
>>
>>   On 9/30/10 3:11 AM, suvayu ali wrote:
>>> Hi ROOT users,
>>>
>>> I am trying to skim an ntuple based on MC data. The MC branches I am
>>> using for my selection look like this,
>>>
>>> Int_t                           mc_n;
>>> vector<int>                  *mc_pdgId;
>>> vector<vector<int>    >    *mc_parent_index;
>>>
>>> I want to make a selection equivalent to this,
>>>
>>>    for(int i=0; i<mc_n;i++){
>>>      if (abs(mc_pdgId[i])==13) {
>>>        firstparentindex.push_back(mc_parent_index[i][0]);
>>>      }
>>>    }
>>>    if (firstparentindex.size()>0){
>>>      for(int i=0; i<mc_n;i++){
>>>        if (i==firstparentindex[0]&&    abs(mc_pdgId[i])==24) {
>>>          ismuondecay =  true;
>>>        }
>>>      }
>>>    }
>>>
>>> I wanted to avoid fiddling with MakeClass() so I tried something like
>>> this,
>>>
>>> TTree *newTree = physics->CopyTree("abs(mc_pdgId[]) == 13&&
>>> abs(mc_pdgId[ mc_parent_index[Iteration$][0] ])== 24");
>>>
>>> But this gives me an empty TTree! If I remove the later cut asking for
>>> the parent, I do get a TTree which has W's in them. I test this by
>>> looking at the histogram drawn by double clicking on the mc_pdgId
>>> branch from a TBrowser.
>>>
>>> root [1] TBrowser b
>>> root [2] htemp->Print("all")
>>> TH1.Print Name  = htemp, Entries= 86496, Total sum= 86496
>>>   .....
>>>   fSumw[3]=0, x=-24.5
>>>   fSumw[4]=349, x=-23.5
>>>   .....
>>>   fSumw[52]=556, x=24.5
>>>   fSumw[53]=0, x=25.5
>>>
>>> What am I doing wrong? Thanks for any insights.
>>>
>
>

Received on Thu Sep 30 2010 - 22:10:32 CEST