Update repo: committing code updates for Lc->pKpi analysis #255

strogolo · 2021-12-10T18:08:23Z

Dear @fcatalan92, @fgrosa,
I am doing this PR in order to push in the repository the updates that I have done for the non-prompt Lc analysis in pp collisions at 13 TeV.
In particular I have updated:

GetRawYieldsDplusDs.py
ProjectDplusDsSparse.py
ProjectDplusDsTree.py
RunFullAnalysis.sh
and I have committed as new macro/files:
ComputeEffAccWeightedAvg.py
ComputeEffAccWeightedAvg.C
config_LctopKpi_Fit_pp13TeV_basic.yml

Please let me know if something is not correct and has to be fixed.
Thanks and regards,
Stefano

fcatalan92

Hi @strogolo, thank you for your contribution. I added a few comments.
Moreover, I would prefer to have only one macro for the weighted average of the efficiency. I would keep ComputeEffAccWeightedAvg.py, we have still a few macros that are both in C++ and in Python but it is not a good idea.

fcatalan92 · 2021-12-13T10:22:38Z

ComputeEffAccWeightedAvg.py

+gROOT.SetBatch(args.batch)
+SetGlobalStyle(padleftmargin=0.14, padbottommargin=0.12, titlesize=0.045, labelsize=0.04)


Is this necessary? You are not drawing anything in the macro.

In the previous version of the script I was not drawing the histograms, while now I'm doing it in the latest update

fcatalan92 · 2021-12-13T10:23:19Z

ComputeEffAccWeightedAvg.py

+effFileAll = TFile.Open(args.effFileAll)
+hEffAllPrompt = effFileAll.Get('hAccEffPrompt')
+hEffAllFD = effFileAll.Get('hAccEffFD')
+SetObjectStyle(hEffAllPrompt, color=kRed+1, markerstyle=kFullCircle)
+SetObjectStyle(hEffAllFD, color=kAzure+4, markerstyle=kOpenSquare, markersize=1.5, linewidh=2, linestyle=7)
+
+effFileNonRes = TFile.Open(args.effFileNonRes)
+hEffNonResPrompt = effFileNonRes.Get('hAccEffPrompt')
+hEffNonResFD = effFileNonRes.Get('hAccEffFD')
+SetObjectStyle(hEffNonResPrompt, color=kRed+1, markerstyle=kFullCircle)
+SetObjectStyle(hEffNonResFD, color=kAzure+4, markerstyle=kOpenSquare, markersize=1.5, linewidh=2, linestyle=7)
+
+effFileKStar = TFile.Open(args.effFileKStar)
+hEffKStarPrompt = effFileKStar.Get('hAccEffPrompt')
+hEffKStarFD = effFileKStar.Get('hAccEffFD')
+SetObjectStyle(hEffKStarPrompt, color=kRed+1, markerstyle=kFullCircle)
+SetObjectStyle(hEffKStarFD, color=kAzure+4, markerstyle=kOpenSquare, markersize=1.5, linewidh=2, linestyle=7)
+
+effFileDelta = TFile.Open(args.effFileDelta)
+hEffDeltaPrompt = effFileDelta.Get('hAccEffPrompt')
+hEffDeltaFD = effFileDelta.Get('hAccEffFD')
+SetObjectStyle(hEffDeltaPrompt, color=kRed+1, markerstyle=kFullCircle)
+SetObjectStyle(hEffDeltaFD, color=kAzure+4, markerstyle=kOpenSquare, markersize=1.5, linewidh=2, linestyle=7)
+
+effFileLambda1520 = TFile.Open(args.effFileLambda1520)
+hEffLambda1520Prompt = effFileLambda1520.Get('hAccEffPrompt')
+hEffLambda1520FD = effFileLambda1520.Get('hAccEffFD')
+SetObjectStyle(hEffLambda1520Prompt, color=kRed+1, markerstyle=kFullCircle)
+SetObjectStyle(hEffLambda1520FD, color=kAzure+4, markerstyle=kOpenSquare, markersize=1.5, linewidh=2, linestyle=7)


I would do a for loop since you are copy-pasting the same operations.

I re-wrote everything using loops in python. It took me sometime to get everything working since I am not super-expert of python. Nevertheless, I checked the output and the result is identical to the old version of the script.

fcatalan92 · 2021-12-13T10:24:36Z

ComputeEffAccWeightedAvg.py

+SetObjectStyle(hEffAllPrompt, color=kRed+1, markerstyle=kFullCircle)
+SetObjectStyle(hEffAllFD, color=kAzure+4, markerstyle=kOpenSquare, markersize=1.5, linewidh=2, linestyle=7)


Setting the style for these histograms looks not needed since you are not drawing/storing them

It is needed in the last version of the script in order to draw the histograms.

GetRawYieldsDplusDs.C

fcatalan92 · 2021-12-13T11:12:02Z

ProjectDplusDsTree.py

+        if args.LctopKpireso == 1:
+            dataFramePrompt = FilterBitDf(dataFramePrompt, 'cand_type', [bitLcNonRes], 'and')
+        if args.LctopKpireso == 2:
+            dataFramePrompt = FilterBitDf(dataFramePrompt, 'cand_type', [bitLcKStar], 'and')
+        if args.LctopKpireso == 3:
+            dataFramePrompt = FilterBitDf(dataFramePrompt, 'cand_type', [bitLcDelta], 'and')
+        if args.LctopKpireso == 4:
+            dataFramePrompt = FilterBitDf(dataFramePrompt, 'cand_type', [bitLcLambda1520], 'and')


You can rewrite this by making a list of the different bits and accessing the list with args.LctopKpireso to avoid the copy-paste

Suggested change

if args.LctopKpireso == 1:

dataFramePrompt = FilterBitDf(dataFramePrompt, 'cand_type', [bitLcNonRes], 'and')

if args.LctopKpireso == 2:

dataFramePrompt = FilterBitDf(dataFramePrompt, 'cand_type', [bitLcKStar], 'and')

if args.LctopKpireso == 3:

dataFramePrompt = FilterBitDf(dataFramePrompt, 'cand_type', [bitLcDelta], 'and')

if args.LctopKpireso == 4:

dataFramePrompt = FilterBitDf(dataFramePrompt, 'cand_type', [bitLcLambda1520], 'and')

if args.LctopKpireso in range(1, 5):

dataFramePrompt = FilterBitDf(dataFramePrompt, 'cand_type', [bitsLcResonance[args.LctopKpireso]], 'and')

fcatalan92 · 2021-12-13T11:12:51Z

ProjectDplusDsTree.py

+        if args.LctopKpireso == 1:
+            dataFrameFD = FilterBitDf(dataFrameFD, 'cand_type', [bitLcNonRes], 'and')
+        if args.LctopKpireso == 2:
+            dataFrameFD = FilterBitDf(dataFrameFD, 'cand_type', [bitLcKStar], 'and')
+        if args.LctopKpireso == 3:
+            dataFrameFD = FilterBitDf(dataFrameFD, 'cand_type', [bitLcDelta], 'and')
+        if args.LctopKpireso == 4:
+            dataFrameFD = FilterBitDf(dataFrameFD, 'cand_type', [bitLcLambda1520], 'and')


Suggested change

if args.LctopKpireso == 1:

dataFrameFD = FilterBitDf(dataFrameFD, 'cand_type', [bitLcNonRes], 'and')

if args.LctopKpireso == 2:

dataFrameFD = FilterBitDf(dataFrameFD, 'cand_type', [bitLcKStar], 'and')

if args.LctopKpireso == 3:

dataFrameFD = FilterBitDf(dataFrameFD, 'cand_type', [bitLcDelta], 'and')

if args.LctopKpireso == 4:

dataFrameFD = FilterBitDf(dataFrameFD, 'cand_type', [bitLcLambda1520], 'and')

if args.LctopKpireso in range(1, 5):

dataFramePrompt = FilterBitDf(dataFramePrompt, 'cand_type', [bitsLcResonance[args.LctopKpireso]], and')

fcatalan92 · 2021-12-13T11:17:45Z

RunFullAnalysis.sh

+    if [ ${Particle} == "LctopKpi" ]; then
+      python3 ${ProjectScript} ${cfgFileMC} ${CutSetsDir}/cutset${CutSets[$iCutSet]}.yml ${OutDirEfficiency}/Distr_${Particle}_MC${CutSets[$iCutSet]}.root
+      python3 ${ProjectScript} ${cfgFileMCNonRes} ${CutSetsDir}/cutset${CutSets[$iCutSet]}.yml  ${OutDirEfficiency}/Distr_${Particle}_NonRes_MC${CutSets[$iCutSet]}.root --LctopKpireso 1
+      python3 ${ProjectScript} ${cfgFileMCKStar} ${CutSetsDir}/cutset${CutSets[$iCutSet]}.yml  ${OutDirEfficiency}/Distr_${Particle}_KStar_MC${CutSets[$iCutSet]}.root --LctopKpireso 2
+      python3 ${ProjectScript} ${cfgFileMCDelta} ${CutSetsDir}/cutset${CutSets[$iCutSet]}.yml  ${OutDirEfficiency}/Distr_${Particle}_Delta_MC${CutSets[$iCutSet]}.root --LctopKpireso 3
+      python3 ${ProjectScript} ${cfgFileMCLambda1520} ${CutSetsDir}/cutset${CutSets[$iCutSet]}.yml  ${OutDirEfficiency}/Distr_${Particle}_Lambda1520_MC${CutSets[$iCutSet]}.root --LctopKpireso 4


Here you are removing the possibility to apply pT weights and other corrections for the Lc->pkpi. Maybe we need to think about how to improve the approach and build the command incrementally

I will implement the possiblity to apply pT weights when I'll have to do it.

fcatalan92 · 2021-12-13T11:23:33Z

RunFullAnalysis.sh

+    if [ ${Particle} == "LctopKpi" ]; then
+      echo $(tput setaf 4) Compute efficiency from ${OutDirEfficiency}/Distr_${Particle}_MC${CutSets[$iCutSet]}.root $(tput sgr0)
+      python3 ComputeEfficiencyDplusDs.py ${cfgFileFit} ${Cent} ${OutDirEfficiency}/Distr_${Particle}_MC${CutSets[$iCutSet]}.root ${OutDirEfficiency}/Efficiency_${Particle}${CutSets[$iCutSet]}.root --batch
+      echo $(tput setaf 4) Compute efficiency from ${OutDirEfficiency}/Distr_${Particle}_NonRes_MC${CutSets[$iCutSet]}.root $(tput sgr0)
+      python3 ComputeEfficiencyDplusDs.py ${cfgFileFit} ${Cent} ${OutDirEfficiency}/Distr_${Particle}_NonRes_MC${CutSets[$iCutSet]}.root ${OutDirEfficiency}/Efficiency_${Particle}_NonRes${CutSets[$iCutSet]}.root --batch
+      echo $(tput setaf 4) Compute efficiency from ${OutDirEfficiency}/Distr_${Particle}_KStar_MC${CutSets[$iCutSet]}.root $(tput sgr0)
+      python3 ComputeEfficiencyDplusDs.py ${cfgFileFit} ${Cent} ${OutDirEfficiency}/Distr_${Particle}_KStar_MC${CutSets[$iCutSet]}.root ${OutDirEfficiency}/Efficiency_${Particle}_KStar${CutSets[$iCutSet]}.root --batch
+      echo $(tput setaf 4) Compute efficiency from ${OutDirEfficiency}/Distr_${Particle}_Delta_MC${CutSets[$iCutSet]}.root $(tput sgr0)
+      python3 ComputeEfficiencyDplusDs.py ${cfgFileFit} ${Cent} ${OutDirEfficiency}/Distr_${Particle}_Delta_MC${CutSets[$iCutSet]}.root ${OutDirEfficiency}/Efficiency_${Particle}_Delta${CutSets[$iCutSet]}.root --batch
+      echo $(tput setaf 4) Compute efficiency from ${OutDirEfficiency}/Distr_${Particle}_Lambda1520_MC${CutSets[$iCutSet]}.root $(tput sgr0)
+      python3 ComputeEfficiencyDplusDs.py ${cfgFileFit} ${Cent} ${OutDirEfficiency}/Distr_${Particle}_Lambda1520_MC${CutSets[$iCutSet]}.root ${OutDirEfficiency}/Efficiency_${Particle}_Lambda1520${CutSets[$iCutSet]}.root --batch


here a for loop could be used

fcatalan92 · 2021-12-13T11:24:08Z

RunFullAnalysis.sh

+      python3 CombineAccTimesEff.py ${OutDirEfficiency}/Efficiency_${Particle}${CutSets[$iCutSet]}.root ${accFileName} ${OutDirEfficiency}/Eff_times_Acc_${Particle}${CutSets[$iCutSet]}.root --batch
+      python3 CombineAccTimesEff.py ${OutDirEfficiency}/Efficiency_${Particle}_NonRes${CutSets[$iCutSet]}.root ${accFileName} ${OutDirEfficiency}/Eff_times_Acc_${Particle}_NonRes${CutSets[$iCutSet]}.root --batch
+      python3 CombineAccTimesEff.py ${OutDirEfficiency}/Efficiency_${Particle}_KStar${CutSets[$iCutSet]}.root ${accFileName} ${OutDirEfficiency}/Eff_times_Acc_${Particle}_KStar${CutSets[$iCutSet]}.root --batch
+      python3 CombineAccTimesEff.py ${OutDirEfficiency}/Efficiency_${Particle}_Delta${CutSets[$iCutSet]}.root ${accFileName} ${OutDirEfficiency}/Eff_times_Acc_${Particle}_Delta${CutSets[$iCutSet]}.root --batch
+      python3 CombineAccTimesEff.py ${OutDirEfficiency}/Efficiency_${Particle}_Lambda1520${CutSets[$iCutSet]}.root ${accFileName} ${OutDirEfficiency}/Eff_times_Acc_${Particle}_Lambda1520${CutSets[$iCutSet]}.root --batch


maybe for loop also here

RunFullAnalysis.sh

strogolo · 2022-01-05T16:42:53Z

Hi @strogolo, thank you for your contribution. I added a few comments. Moreover, I would prefer to have only one macro for the weighted average of the efficiency. I would keep ComputeEffAccWeightedAvg.py, we have still a few macros that are both in C++ and in Python but it is not a good idea.

Hi @fcatalan92, I have implemented almost all of your suggestions. Regarding the double version of the macro ComputeEffAccWeightedAvg (in python and in C++) I understand your point. However, I just want to inform you that on my laptop I'm using the C++ version, while the python version was written for completeness for your repository. Thus, I was just wondering if it is good to have at least a version of the macro that I'm currently using as documentation.

Thanks and regards,
Stefano

ComputeEffAccWeightedAvg.py

fcatalan92 · 2022-01-08T17:09:03Z

Hi @strogolo, thank you for implementing the changes. I have just a few suggestions more.

On the ComputeEffAccWeightedAvg, have you checked that the python version is equivalent to the C++ one (same outputs)? If yes, I think we can merge only the former one. @fgrosa What do you think?

strogolo · 2022-01-10T11:31:10Z

Hi @strogolo, thank you for implementing the changes. I have just a few suggestions more.

On the ComputeEffAccWeightedAvg, have you checked that the python version is equivalent to the C++ one (same outputs)? If yes, I think we can merge only the former one. @fgrosa What do you think?

Hi @fcatalan92, I have removed the debug print as requested.
Concerning your question: I did not check in details that the output of the python version is equivalent to the C++ one. From a quick look it seems everything equivalent, but I did not perform an apple-to-apple comparison because the main macro I'm using for the analysis is the C++ one, while I wrote the python for completeness and because someone could prefer to use the python version. Look, if it is such a huge problem I can remove ComputeEffAccWeightedAvg (both python and C++) since it is a macro specific for the Lc->pKpi analysis and just keep it locally updated.

fgrosa

Ciao @strogolo thanks a lot! I have a couple of comments that just go in the direction to simplify a bit the code (if they are fine with you you can just click on "commit suggestion"). For the pT weights, indeed it will have to be done, but it's ok for me to do it in a second step (also because in the meantime we might want to merge this PR #249). For the c++/python script I have no strong opinion, it is fine for me to keep only one of the two or both.

ComputeEffAccWeightedAvg.C

ComputeEffAccWeightedAvg.py

RunFullAnalysis.sh

fcatalan92 · 2022-01-11T09:02:21Z

@strogolo Thank you for the updates. I made just a minor comment.
IMHO, the most important thing is to have only one macro to maintain. So, if it is not possible to validate the python one (my preferred option), I would keep only the C++ macro.

Co-authored-by: Fabrizio <fabrizio.grosa@cern.ch>

strogolo · 2022-01-11T09:33:48Z

@strogolo Thank you for the updates. I made just a minor comment. IMHO, the most important thing is to have only one macro to maintain. So, if it is not possible to validate the python one (my preferred option), I would keep only the C++ macro.

Hi @fcatalan92 thanks for your reply. What I can propose is that we merge both macros and in the next weeks, when I'll do the systematic studies, I'll try to use the python script to check that works properly. Once this is verified, we can remove the C++ macro from the repository. Could this be a viable option? In this way we can go on with this PR (without blocking others) and in parallel have the time to do some checks. Thanks, Stefano

ComputeEffAccWeightedAvg.py

fcatalan92

@strogolo fine for me to merge now and remove the C++ macro after the checks.

fcatalan92 requested changes Dec 13, 2021

View reviewed changes

fcatalan92 reviewed Dec 13, 2021

View reviewed changes

RunFullAnalysis.sh Outdated Show resolved Hide resolved

strogolo force-pushed the master branch from 2e97f61 to 698bdde Compare January 5, 2022 11:54

strogolo requested a review from fcatalan92 January 7, 2022 08:57

fcatalan92 requested changes Jan 8, 2022

View reviewed changes

strogolo added 10 commits January 10, 2022 12:23

Update GetRawYieldsDplusDs: split in LctopKpi and LctopK0s

c067000

Add projection for each resonant channel (LctopKpi)

4cca6e0

Add projection (TTree) for each resonant channel (LctopKpi)

48cd6c9

Add macro/script to compute LctopKpi efficiency BR-weighted

1538368

Update RunFullAnalysis.sh for LctopKpi analysis

8343546

Re-write ComputeEffAccWeightedAvg.py with loops

d2b7a7c

Fix kLctopKpi and kLctopK0s

36e3312

Introduce loop to check LcResonance BIT

1878f88

Implement for loop on Lc ResoChannel

7025378

Remove debug print output

d21be60

strogolo force-pushed the master branch from 4e90aac to d21be60 Compare January 10, 2022 11:25

strogolo requested a review from fcatalan92 January 10, 2022 11:26

fgrosa reviewed Jan 10, 2022

View reviewed changes

fgrosa mentioned this pull request Jan 10, 2022

Various implementations #249

Merged

strogolo and others added 7 commits January 11, 2022 10:18

Update: re-arrange weighted average loop

aa451fd

Co-authored-by: Fabrizio <fabrizio.grosa@cern.ch>

Add TMath in ComputeEffAccWeightedAvg

6df53c6

Co-authored-by: Fabrizio <fabrizio.grosa@cern.ch>

Remove the file with all efficiencies

a1ac0d3

Co-authored-by: Fabrizio <fabrizio.grosa@cern.ch>

Change loop length

6906329

Co-authored-by: Fabrizio <fabrizio.grosa@cern.ch>

Change TFile size

796c35e

Co-authored-by: Fabrizio <fabrizio.grosa@cern.ch>

Change array size

8d939e8

Co-authored-by: Fabrizio <fabrizio.grosa@cern.ch>

Change loop length

cc8b5c2

Co-authored-by: Fabrizio <fabrizio.grosa@cern.ch>

strogolo and others added 7 commits January 11, 2022 10:24

Update uncertainty calculation for weighted average

8ef6855

Co-authored-by: Fabrizio <fabrizio.grosa@cern.ch>

Remove loading of global efficiency

3b45f2a

Co-authored-by: Fabrizio <fabrizio.grosa@cern.ch>

Remove opening of global efficiency

c770718

Co-authored-by: Fabrizio <fabrizio.grosa@cern.ch>

Indentation of SetObjectStyle

2ad5e07

Co-authored-by: Fabrizio <fabrizio.grosa@cern.ch>

Indentation of SetObjectStyle

e85d97d

Co-authored-by: Fabrizio <fabrizio.grosa@cern.ch>

Remove total BR Lc->pKpi

ade6921

Co-authored-by: Fabrizio <fabrizio.grosa@cern.ch>

Remove case of unweighted average

c2fb726

Co-authored-by: Fabrizio <fabrizio.grosa@cern.ch>

strogolo commented Jan 11, 2022

View reviewed changes

ComputeEffAccWeightedAvg.py Outdated Show resolved Hide resolved

Remove effC and wC

d871a5b

fcatalan92 approved these changes Jan 11, 2022

View reviewed changes

fcatalan92 mentioned this pull request Jan 11, 2022

Check equivalency of Python and C++ macro for weighted efficiency #259

Open

fcatalan92 merged commit a19663d into DmesonAnalysers:master Jan 11, 2022

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Update repo: committing code updates for Lc->pKpi analysis #255

Update repo: committing code updates for Lc->pKpi analysis #255

strogolo commented Dec 10, 2021

fcatalan92 left a comment

fcatalan92 Dec 13, 2021

strogolo Jan 5, 2022

fcatalan92 Dec 13, 2021

strogolo Jan 5, 2022

fcatalan92 Dec 13, 2021

strogolo Jan 5, 2022

fcatalan92 Dec 13, 2021

strogolo Jan 5, 2022

fcatalan92 Dec 13, 2021

strogolo Jan 5, 2022

fcatalan92 Dec 13, 2021

strogolo Jan 5, 2022

fcatalan92 Dec 13, 2021

strogolo Jan 5, 2022

fcatalan92 Dec 13, 2021

strogolo Jan 5, 2022

strogolo commented Jan 5, 2022

fcatalan92 commented Jan 8, 2022

strogolo commented Jan 10, 2022

fgrosa left a comment

fcatalan92 commented Jan 11, 2022

strogolo commented Jan 11, 2022

fcatalan92 left a comment

		gROOT.SetBatch(args.batch)
		SetGlobalStyle(padleftmargin=0.14, padbottommargin=0.12, titlesize=0.045, labelsize=0.04)

		SetObjectStyle(hEffAllPrompt, color=kRed+1, markerstyle=kFullCircle)
		SetObjectStyle(hEffAllFD, color=kAzure+4, markerstyle=kOpenSquare, markersize=1.5, linewidh=2, linestyle=7)

Update repo: committing code updates for Lc->pKpi analysis #255

Update repo: committing code updates for Lc->pKpi analysis #255

Conversation

strogolo commented Dec 10, 2021

fcatalan92 left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

strogolo commented Jan 5, 2022

fcatalan92 commented Jan 8, 2022

strogolo commented Jan 10, 2022

fgrosa left a comment

Choose a reason for hiding this comment

fcatalan92 commented Jan 11, 2022

strogolo commented Jan 11, 2022

fcatalan92 left a comment

Choose a reason for hiding this comment