Skype discussion with Judita

[ytatus94]

• 20170518

[18.05.17, 9:56:36 AM] Judita Mamuzic: Hi Yu-Ting, thanks for asking the Higgsino group to add the NUHM2 points! I wrote there you will provide the sensitivity plots. Can you informally find out when the processing will be ready? if it is close, then we can wait, if not, perhaps we can run the truth samples I guess while we wait for the EWK 2/3 L, with Higgsino group all is faster and more clear... ok, I have a meeting now we can chat later Hi Yu-Ting, we got approved for compressed+compressed production :-)

[18.05.17, 1:41:30 PM] Yu-Ting: Hi Judita, I was eating my lunch in R1

[18.05.17, 1:41:45 PM] Judita Mamuzic: Hi, are we meeting today? I am super busy, catching up with many topics after the Bucharest workshop do you think Christopher would make the production request? I think you could provide the sensitivity plots but perhaps we can discuss all at the meeting

[18.05.17, 1:44:32 PM] Yu-Ting: No, I believe Chris cannot do that. No one know how to do that except Abe. Pat said he will find time to learn how to request the compressed+compressed after the grading is done. (The semester just finished and he need to grad the final and turn in the scores of students.)

[18.05.17, 1:45:32 PM] Judita Mamuzic: Ah, so you discussed already, it is not too difficult, we have the jobOptions from previous round, we just need to change the processes

[18.05.17, 1:46:00 PM] Yu-Ting: If I understood correctly, we still can use accessible+compressed to do our study because the compressed+compressed is similar to Jesse's study. Isn't it?

[18.05.17, 1:46:10 PM] Judita Mamuzic: and perhaps add the MadSpin decays yes, I think it would be good if you could to this sensitivity study it is very close, by Monday and we are (again) waiting for samples, although, this time they are on a very good way

[18.05.17, 1:47:22 PM] Yu-Ting: Abe kept most of things before he quit his job but no one spends time to know and to learn how to do that yet.

[18.05.17, 1:47:47 PM] Judita Mamuzic: it is all at the JIRA request so this should be all OK, but one needs to test the jobs, submit the request, so it requires some time

[18.05.17, 1:49:53 PM] Yu-Ting: By the way, if you have time, could you please send a mail to Sarah to ask her about my previous mail and also ask Nicky about the status of the common ntuples? They rarely reply my mail but they reply yours

[18.05.17, 1:50:01 PM] Judita Mamuzic: I could give it a try, perhaps start the request, but perhaps somebody can take over if I’m out of time I think they got the email, I think they will reply, I see n point in bugging them more its not that they don’t reply to you, I think they are busy I worked with them before, it was similar :-) We got all results at the end nicely but it did take a while ok, I could got to R1 need to leave early today, so lets have ameeting nowish?

[18.05.17, 5:13:16 PM] Yu-Ting: Ok, see you in R1

[18.05.17, 5:13:31 PM] Judita Mamuzic: ok, I leave now, will be there in 7min or so

[18.05.17, 5:13:38 PM] Yu-Ting: okay

[18.05.17, 5:27:25 PM] Judita Mamuzic:

mc15_13TeV:mc15_13TeV.370620.MGPy8EG_A14N23LO_NUHM2_m12_500_weak.evgen.EVNT.e4209
mc15_13TeV:mc15_13TeV.370619.MGPy8EG_A14N23LO_NUHM2_m12_400_weak.evgen.EVNT.e4209
mc15_13TeV:mc15_13TeV.370621.MGPy8EG_A14N23LO_NUHM2_m12_600_weak.evgen.EVNT.e4209
mc15_13TeV:mc15_13TeV.370618.MGPy8EG_A14N23LO_NUHM2_m12_350_weak.evgen.EVNT.e4209
mc15_13TeV:mc15_13TeV.370622.MGPy8EG_A14N23LO_NUHM2_m12_700_weak.evgen.EVNT.e4209
mc15_13TeV:mc15_13TeV.370623.MGPy8EG_A14N23LO_NUHM2_m12_800_weak.evgen.EVNT.e4209
mc15_13TeV:mc15_13TeV.370617.MGPy8EG_A14N23LO_NUHM2_m12_300_weak.evgen.EVNT.e4209

https://twiki.cern.ch/twiki/bin/viewauth/AtlasProtected/TruthDAOD https://gitlab.cern.ch/atlas-phys-susy-wg/SimpleAnalysis

[ytatus94]

• 20170519

[19.05.17, 2:15:10 PM] Judita Mamuzic: Hi Yu-Ting, did the truth3 generation go well?

[19.05.17, 2:17:42 PM] Yu-Ting: I sent a derivation request yesterday. And I am still reading the tutorial. Didn't start the production yet.

[19.05.17, 2:18:26 PM] Judita Mamuzic: OK, I guess we would know if the samples entered the train, so we need to make it locally ok, it should be straight forward, but I also know it typically goes with many details to consider, let me know when they are finished, and I will run the next step with SimpleAnalysis

[19.05.17, 2:19:38 PM] Yu-Ting: I don't know when will the train start so I will produce the derivation locally.

[19.05.17, 2:19:53 PM] Judita Mamuzic: yes, that is very good ok, I just wanted to check wish you a nice afternoon

[19.05.17, 2:20:56 PM] Yu-Ting: I will let you know after the samples are ready. Thank you. Hi Judita, what you need is a truth3 root file, right? Because the output contains several files

[19.05.17, 3:23:17 PM] Judita Mamuzic: yes yes, only the root

[19.05.17, 3:24:42 PM] Yu-Ting: okay, I read another twiki tutorial yesterday not the one you gave me. After comparing the one you gave me, I realized the tutorial I read contains the analysis. To produce the truth3 is only one step, and I already submit it. Will let you know when the root files are ready.

[19.05.17, 3:25:50 PM] Judita Mamuzic: great thanks :-)

[19.05.17, 5:54:23 PM] Yu-Ting: Hi Judita, the TRUTH3 DAOD are ready You can find them at /afs/cern.ch/user/y/yushen/public/ForJudita

[ytatus94]

• 20170520

[20.05.17, 3:55:04 PM] Judita Mamuzic: Hi Yu-Ting, I have started, but I have problems finding afs space, will need more time

[20.05.17, 3:59:05 PM] Yu-Ting: That is okay. I didn't start yet...

[20.05.17, 3:59:13 PM] Judita Mamuzic: its ok

[20.05.17, 3:59:17 PM] Yu-Ting: I was working on housekeeping things

[20.05.17, 3:59:20 PM] Judita Mamuzic: :-D today is wine festival, you missed it?

[20.05.17, 3:59:47 PM] Yu-Ting: And I will go to downtown to renew my monthly pass and go to Asian supermarket

[20.05.17, 4:00:01 PM] Judita Mamuzic: sounds like a nice plan

[20.05.17, 4:00:14 PM] Yu-Ting: I heard that but I don't know where is it.

[20.05.17, 4:00:20 PM] Judita Mamuzic: I need a few hours, will try to start them before 17h, but they will also need some time to finish I don’t have the link, but your Oklahoma group must know...

[20.05.17, 4:01:23 PM] Yu-Ting: Usually I cannot work in Saturday because lots of things need to do except works.

[20.05.17, 4:01:44 PM] Judita Mamuzic: yes, I allways try, but it only rare that I manage ok, I let you with you Satuday plan, and I will try to submit them soon the further we go, we will need the filter efficiencies but you cannot use them without the small ntuples...

[20.05.17, 4:03:27 PM] Yu-Ting: That is why I said I might not get significance plots in time.

[20.05.17, 4:03:38 PM] Judita Mamuzic: that should be fast following that script you showed at the meeting we are on a good way today we can have the ingredients, tomorrow setting up it can work we also don’t need all points for now so if it is difficult, just take 1, e.g. 700

[20.05.17, 4:05:08 PM] Yu-Ting: Anyway I will try to make the plots tonight or tomorrow

[20.05.17, 4:05:15 PM] Judita Mamuzic: yes, it is possible

[20.05.17, 11:32:10 PM] Yu-Ting: Hi Judita, I notice the official TRUTH3 ntuples are ready. You can find them using: rucio download mc15_13TeV.*.MGPy8EG_A14N23LO_NUHM2_m12_*_weak.merge.DAOD_TRUTH3.*

[20.05.17, 11:32:40 PM] Judita Mamuzic: great, thanks I have put the simpleanalysis ntuples to my public you can take a look I continue tomorrow good night

[20.05.17, 11:33:15 PM] Yu-Ting: good night

[ytatus94]

• 20170521

[21.05.17, 6:16:43 PM] Yu-Ting: Hi Judita, just let you know my status. The framework finally runs but none of event pass the SR requirements. And I notice it is because of MET/HT cut. I will start to investigate this

[21.05.17, 6:42:46 PM] Judita Mamuzic: Hi @Yu-Ting, which cuts did you use?

[21.05.17, 6:43:15 PM] Yu-Ting: METOverHT

[21.05.17, 6:43:37 PM] Judita Mamuzic: I mean which Higgsino SR cuts?

[21.05.17, 6:44:08 PM] Yu-Ting: I use the cuts list in page 3 https://indico.cern.ch/event/637730/contributions/2593606/attachments/1459588/2254873/Higgsino_20170515_intro.pdf But only m1/2=700 and 800 have METOverHT > 5.

[21.05.17, 6:45:26 PM] Judita Mamuzic: hmm 5 is really high cut for METIverHt lets try without it could you make distribution plots? and it would be good to have normalisation

[21.05.17, 6:47:13 PM] Yu-Ting: And after checking the METOverHT value, there are inf

[21.05.17, 6:47:23 PM] Judita Mamuzic: for that we need cross-section from the SUSYtools, and to get the filter efficiencies

[21.05.17, 6:47:55 PM] Yu-Ting: a few events with METOverHT > 5 but those event does't pass leading jet pT cut

[21.05.17, 6:48:04 PM] Judita Mamuzic: could you make plots of just after the baseline selection

[21.05.17, 6:48:10 PM] Yu-Ting: Distribution plot for METOverHT? There is no baseline in the ntuples only nLep_base and nLep_signal

[21.05.17, 6:49:25 PM] Judita Mamuzic: it would be good to have all: jet pt (Also jet[0], jet[2], jet[2]), lepton pt, met, HT, met/HT, Njets, Nleptons, b-jets pt, Nbjet, mT2, mll

[21.05.17, 6:49:44 PM] Yu-Ting: so I assume all the trees in the ntuple are baseline ok, I can make those plots. for all m1/2?

[21.05.17, 6:50:22 PM] Judita Mamuzic: ok, then it would be good to have these plots, but normalised lets say 500 and 700

[21.05.17, 6:50:43 PM] Yu-Ting: normalised? i don't understand

[21.05.17, 6:51:03 PM] Judita Mamuzic: yes, normalised to 36.1 fb, using xesc and filt eff you can do only m1/2 = 600

[21.05.17, 6:52:04 PM] Yu-Ting: the mc samples are weighted for 1pb-1 so I think I don't need to do normalised

[21.05.17, 6:52:36 PM] Judita Mamuzic: ah, so the xsec is in? not sure, is there event weight in the tree?

[21.05.17, 6:52:57 PM] Yu-Ting: and actually I don't know how to normalize those distribute Yes, there are event weight and xsec in the trees

[21.05.17, 6:53:26 PM] Judita Mamuzic: probably you need to use that, and since we don’t have in the SUSYtools the right filt eff, we need to add that, i.e. just multiply with the value

[21.05.17, 6:53:48 PM] Yu-Ting: but Joey said the mc sample are normalised in his slide so what you mean "normalized", you mean applying all the weight and cross section

[21.05.17, 6:55:15 PM] Judita Mamuzic: yes

[21.05.17, 6:55:16 PM] Yu-Ting: but if the cross sections in the SUSYTools are wrong, then this is meaningless

[21.05.17, 6:55:29 PM] Judita Mamuzic: no, they are not wrong, they are misisng filt eff which you can add per sample yourself if you run 1 point get filt eff for that one

[21.05.17, 6:56:12 PM] Yu-Ting: but the correct weight should be Lumi x Xsec x Eff x all the weights

[21.05.17, 6:56:15 PM] Judita Mamuzic: I think the weighting should be like: eventWeight*filteff*lumi yes, the xsec weight might be in the event weight so we need to make some checks**

[21.05.17, 6:57:21 PM] Yu-Ting: I dont' think so. because there is xsec and eventWeight in the tree. they are separated variables

*[21.05.17, 6:57:36 PM] Judita Mamuzic: I see, I get it now then it might be like this: `xseckfacfiltefflumi/sum_weights but1/sum_weights might be in theeventWeight`**

[21.05.17, 6:59:09 PM] Yu-Ting: we don't have filter eff

[21.05.17, 6:59:19 PM] Judita Mamuzic: ah, also, be careful, you need to read the process for the event

[21.05.17, 6:59:25 PM] Yu-Ting: and the sum of weight is another variable genWeight, I believe

[21.05.17, 6:59:36 PM] Judita Mamuzic: is there somehting for that? we have xsec per process

[21.05.17, 7:01:51 PM] Yu-Ting: there is xsec in ntuple

[21.05.17, 7:02:01 PM] Judita Mamuzic: yes, does it go per process? does it differ per event or is it allways the same?

[21.05.17, 7:02:14 PM] Yu-Ting: this variable should keep correct xsec

[21.05.17, 7:03:06 PM] Judita Mamuzic: if you draw just xsec, is it always the same for one NUHM2 point

[21.05.17, 7:03:07 PM] Yu-Ting: they varied

[21.05.17, 7:03:17 PM] Judita Mamuzic: or it has couple values ok and is there some tree for the final state?

[21.05.17, 7:03:44 PM] Yu-Ting: it is strange. I think they should be the same. yes there is FS

[21.05.17, 7:04:25 PM] Judita Mamuzic: different FS have different xsec

[21.05.17, 7:04:42 PM] Yu-Ting: yes, i noticed that by the way, about the mail you sent yesterday, I cannot visit /afs/cern.ch/work/m/mamuzic/public/ForYuTing/NUHM2 Permission denied

[21.05.17, 7:06:14 PM] Judita Mamuzic: hmm, strange, it is public… excellent it looks like FS and xsec are there so use: FS x xsec x genWeight x eventWeight x pileupWeight x filtEff this is without trigger weight for now Perhaps produce this for 1 point then we could try to have plots like this https://its.cern.ch/jira/secure/attachment/119863/119863_Higgsino_0505.pdf on the starting few pages I have changed the rights at my public, it should be OK now /afs/cern.ch/work/m/mamuzic/public/ForYuTing/NUHM2 we will be comparing it to this point in Higgsino analysis MGPy8EG_A14N23LO_SM_Higgsino_160_150_2LMET50_MadSpin_SusySkimHiggsino_v1.4_SUSY16_tree_NoSys.root the two parameters are N2 and N1, and I read out for NUHM2 what are the N2 and N1 for m1/2 = 600 point so this is how I chose the Higgsiono point for comparison sorry, no, it shoud not be multitplied by FS just xsec x genWeight x eventWeight x pileupWeight x filtEff it just says what is the FS, and the xsec is corectly added

[21.05.17, 10:36:18 PM] Yu-Ting: Hi Judita, I am fighting with the online reimbursement system of my school since 7pm. (It is very very hard to use. A very suck system.) and still working on it. I have several problems need to be fix for the reimbursement report, so I think it is impossible for me to show results tomorrow. :( Sorry for telling you that.

[21.05.17, 10:43:01 PM] Judita Mamuzic: OK, don’t worry, I will put up some slides for tomorrow, I might try to make some relative comparison plots, as with the script I have sent you, and perhaps I might ask for a bit of help tomorrow, if that’s ok. The plots with cuts we will provide during the week

[21.05.17, 10:51:55 PM] Yu-Ting: and i checked your public folder, it is still permission deny Screen Shot 2017-05-21 at 10.51.50 PM.png

[21.05.17, 10:52:16 PM] Judita Mamuzic: oh, that is strange

[21.05.17, 10:52:51 PM] Yu-Ting: do yo put a link of those files not a real copy?

[21.05.17, 10:53:09 PM] Judita Mamuzic: its a copy

[21.05.17, 10:54:10 PM] Yu-Ting: then I don't know what is the possible problem causing this.

[21.05.17, 10:54:44 PM] Judita Mamuzic: can you try now?

[21.05.17, 10:55:57 PM] Yu-Ting: still the same

[21.05.17, 10:56:48 PM] Judita Mamuzic: nd now? there is no more rights I can add

[21.05.17, 10:57:10 PM] Yu-Ting: nothing changed

[21.05.17, 10:57:15 PM] Judita Mamuzic: can you copy to your folder

[21.05.17, 10:57:23 PM] Yu-Ting: no I cannot

**[21.05.17, 10:57:29 PM] Judita Mamuzic: this is funny for me it says its

-rwxrwxrwx. 1 mamuzic zp  110 May 20 
16:51 user.NUHM2.370623.EwkHiggsino2016_000029.root.txt
-rwxrwxrwx. 1 mamuzic zp 1.4M May 20 16:51 user.NUHM2.370623.EwkHiggsino2016_000030.root.root
-rwxrwxrwx. 1 mamuzic zp  111 May 20 16:51 user.NUHM2.370623.EwkHiggsino2016_000030.root.txt**

[21.05.17, 10:57:50 PM] Yu-Ting:

[yushen@lxplus021 NUHM2]$ cp user.NUHM2.370623.EwkHiggsino2016_000030.root.txt ~/
cp: cannot stat `user.NUHM2.370623.EwkHiggsino2016_000030.root.txt': Permission denied

[21.05.17, 10:59:38 PM] Judita Mamuzic: can you try this one? /afs/cern.ch/user/m/mamuzic/public/ForYuTing not the workspace public, perhaps its some afs problem

[21.05.17, 11:00:27 PM] Yu-Ting: this one works

[21.05.17, 11:00:40 PM] Judita Mamuzic: I guess its some workspace problem I just copied the same folder, with same rights

[21.05.17, 11:00:55 PM] Yu-Ting: let me copy the files to my desktop

[21.05.17, 11:01:03 PM] Judita Mamuzic: ok, in principle just copy the brpv.py the others are truth, we dont need them anymore, you have been using the Higgsino samples already ok, I stop here good night

[21.05.17, 11:02:54 PM] Yu-Ting: good night

[ytatus94]

• 20170522

[22.05.17, 12:19:07 PM] Yu-Ting: Hi Judita, I have m1/2=600 distributions /afs/cern.ch/user/y/yushen/public/ForJudita/optimization_MC_MGPy8EG_A14N23LO_NUHM2_m12_600_weak_SusySkimHiggsino_v1.4_SUSY16_tree_NoSys.root/hist-SusySkimHiggsino_v1.4_SUSY16_Signal_tree.root But they are not scaled to the correct luminosity because after scaling the plots look strange. Maybe I did something wrong. So I decide to give you before scaling plots. Here is how I scale the plots

float weight = xsec x eventWeight x leptonWeight x bTagWeight x jvtWeight x pileupWeight x genWeight;
weight *= luminosity;

where luminosity is set to 36095 at very beginning. I didn't find anything wrong for the weight calculation. I will continue to check my framework after lunch

[22.05.17, 12:44:35 PM] Judita Mamuzic: Hi Yu-Ting, this is great can you produce the same also for the MGPy8EG_A14N23LO_SM_Higgsino_160_150_2LMET50_MadSpin_SusySkimHiggsino_v1.4_SUSY16_tree_NoSys.root and draw all these distributions on the same plot instead of omitting to scale with lumi, you could scale each of them with the Integral then they are just relative plots integral before cuts one question MET/HT, how you define HT?

[22.05.17, 1:06:20 PM] Yu-Ting: What do you mean scale each of them with the integral? Do you mean normalize to one and compare plots?

[22.05.17, 1:06:33 PM] Judita Mamuzic: yes for each of the histograms, draw Higgsino and NUHM2 on the same plot in tow different colors is that possible?

[22.05.17, 1:07:57 PM] Yu-Ting: The MET/HT is a variable in the Higgsion framework, and I checked the source code, it only shows float METOverHT = met / Ht30; and

  // Ht: Scalar sum of jets
  float Ht30 = getHt( configMgr->obj, configMgr->obj->cJets.size(), 30.0 );

[22.05.17, 1:08:09 PM] Judita Mamuzic: if not, then send me these in pdf

[22.05.17, 1:08:36 PM] Yu-Ting: But I couldn't find the definition of getHt() because it is in another package and doesn't include in the Higgsino framework

[22.05.17, 1:09:01 PM] Judita Mamuzic: ok, we will make these checks later so we have 1 hour, can you produce these plots in pdf? and if there is enough time, run the same for the Higgsino point, and make plots with both and normalised to 1

[22.05.17, 1:11:43 PM] Yu-Ting: yes, i am working on it now.

[22.05.17, 1:11:51 PM] Judita Mamuzic: perfect! Hi Yu-Ting, what is the status? should I make plots of the root file you have sent me and use just that? I will make those, and then we add the overlayed ones if you manage

[22.05.17, 1:51:07 PM] Yu-Ting: Hi Judita, the plots are ready one second ago. I move them to /afs/cern.ch/user/y/yushen/public/ForJudita

[22.05.17, 1:51:21 PM] Judita Mamuzic: wohoo!

[22.05.17, 1:54:50 PM] Yu-Ting: I don't have b-jet pT, mT2 plots because there is no information about these two in the production ntuple.

[22.05.17, 1:55:05 PM] Judita Mamuzic: plots look very nice the jets pt is strange leptons pt I mean**

[22.05.17, 1:55:52 PM] Yu-Ting: baseline or signal

[22.05.17, 1:56:12 PM] Judita Mamuzic: could you redo with log scale?

[22.05.17, 1:56:19 PM] Yu-Ting: do you mean higgsino one? I notice there is only one bin in both baseline and signal leptons pT The plots are correct

[22.05.17, 2:00:58 PM] Judita Mamuzic: theu look amazing we just need to make some snese

[22.05.17, 2:01:09 PM] Yu-Ting: I checked the production ntuple. My plot only has one bin because the lep pT are all less then 50 And my first bin is 0-50 so there is only one bin in my plot. I can make thinner bin but it should be after the meeting

[22.05.17, 2:02:41 PM] Judita Mamuzic: yes, they might ask so this is all signal selection, no cuts yes, I see, we need finer binning, this is low pt analysis we need the bins below 50 GeV

[22.05.17, 2:03:28 PM] Yu-Ting: no cut, I only selected signal and baseline leptons. There is a signal flag for the leptons

[22.05.17, 2:03:57 PM] Judita Mamuzic: we need to connect

[22.05.17, 2:04:10 PM] Yu-Ting: I am on vidyo now

[22.05.17, 2:04:27 PM] Judita Mamuzic: me too

[22.05.17, 2:09:58 PM] Yu-Ting: I will ask MET/HT to Mike

[22.05.17, 2:18:37 PM] Judita Mamuzic: I have uploaded take a look

[22.05.17, 2:18:45 PM] Yu-Ting: thank you.

[22.05.17, 2:18:49 PM] Judita Mamuzic: who will give the talk?

[22.05.17, 2:19:36 PM] Yu-Ting: I don't familiar the first two pages Could you please give the talk?

[22.05.17, 2:20:14 PM] Judita Mamuzic: sure, next time is you plots are very nice

[22.05.17, 2:20:37 PM] Yu-Ting: and I can answer the distributions if there is any question

[22.05.17, 2:20:46 PM] Judita Mamuzic: we need to clarrify a few things until next week, but they look ver good yes, that would be great

[22.05.17, 2:21:43 PM] Yu-Ting: And after the meeting, i have some minor questions want to ask you.

[22.05.17, 2:21:51 PM] Judita Mamuzic: yes, sure :-D More work :-D but ok

[22.05.17, 3:16:11 PM] Yu-Ting: Thank you for the presentation :)

[22.05.17, 3:16:31 PM] Judita Mamuzic: np :-D thanks for the very nice plots so we should meet after the meeting

[22.05.17, 3:16:53 PM] Yu-Ting: yes

[22.05.17, 3:40:45 PM] Judita Mamuzic: meeting time?

[22.05.17, 3:42:51 PM] Yu-Ting: now?

[22.05.17, 3:43:00 PM] Judita Mamuzic: yes, lets meet in R1

[22.05.17, 3:43:08 PM] Yu-Ting: okay see you later

[22.05.17, 3:43:17 PM] Judita Mamuzic: I am leaving the office, will be there in a few min

[ytatus94]

• 20170525

[25.05.17, 3:09:42 PM] Judita Mamuzic: Hi Yu-Ting, I will not connect today to the meeting, I am using the holiday :-) Cheers

[25.05.17, 3:11:19 PM] Yu-Ting: Hi Judita, thank you for letting me know. I am also out of town today.

[25.05.17, 3:11:41 PM] Judita Mamuzic: Cool! :-)

[25.05.17, 3:12:09 PM] Yu-Ting: I found the mean value of the filter efficiency on Tursday but I don't know how to get the filter efficiency for each final states. I sent mail to ask Serhan because Da Xu told me he might know. But Serhan didn't reply me.

[25.05.17, 3:13:11 PM] Judita Mamuzic: it would be better to get them per process, but if its not provided, then we use what we have thanks for finding it and making all these checks!

[25.05.17, 3:14:42 PM] Yu-Ting: I also tried to use the v1.4 ntuple to calculate the significance yesterday but I got very huge number of background events in SR. For #signal < 50 but for #bkg > 100000, I don't think there is sensitivity using this kind of Nsig and Nbkg.

[25.05.17, 3:15:20 PM] Judita Mamuzic: but did you do the correct weighting

[25.05.17, 3:15:27 PM] Yu-Ting: For the SimpleAnalysis, I only produced the TRUTH3 for Higgsino 160 150. And I am reading the code of the framework

[25.05.17, 3:15:29 PM] Judita Mamuzic: for each process without weighting, you dont really know if its large or not

[25.05.17, 3:15:41 PM] Yu-Ting: Yes, I did

[25.05.17, 3:16:02 PM] Judita Mamuzic: Ah, great!

[25.05.17, 3:16:07 PM] Yu-Ting: I think the Nsig is reasonable but Nbkg is too huge.

[25.05.17, 3:16:31 PM] Judita Mamuzic: perhaps there is some cut that is missing, or some weighting is off?

[25.05.17, 3:16:32 PM] Yu-Ting: I want to ask Joey to provide his values to make a cross check using bkg samples.

[25.05.17, 3:16:41 PM] Judita Mamuzic: yes, that is a very good way

[25.05.17, 3:16:43 PM] Yu-Ting: But I didn't do it yet.

[25.05.17, 3:16:59 PM] Judita Mamuzic: yes, great, that is a good way, so glad you managed so many things ok, back to the holiday we continue later wish you a nice day!

[25.05.17, 3:17:51 PM] Yu-Ting: thank you. I will be back to Geneva on Sunday night But I bring my laptop so I will still do some work in the evening.

[25.05.17, 3:19:57 PM] Judita Mamuzic: did you see we need to document our analysis

[25.05.17, 3:20:13 PM] Yu-Ting: yes I saw that JIRA ticket

[25.05.17, 3:20:20 PM] Judita Mamuzic: I could do the intro, and the mass spectrum and xsec justification, and you could add the distribution plots ok, so we need to do that I think this week (?) at least something and once we have the significance plots, we will add them with the comp+comp production ok, ciao for now…

[25.05.17, 3:21:57 PM] Yu-Ting: sure, ciao

[ytatus94]

• 20170529

[29.05.17, 11:35:06 AM] Yu-Ting: Hi Judita, I am not going to present today because my results for the sensitivity are very strange. The number of signal is < 50 but number of background > 100000. And after mailing with Joey last weekend, there are a lot of things about the common ntuple I don't know and need to clarify with him again. He also suggested me to do the cutflow comparisons first. I am working on it now. And I don't have progress on the truth level study. I only spent time to read the source code of the common ntuple maker when I was on my trip.

[29.05.17, 11:39:28 AM] Judita Mamuzic: Hi Yu-Ting, I also didn’t manage to prepare the EVNT files for compressed+compressed. I have that talk on Wednesday, and I had serious computing farm problems, so it took all the time. I would like to focus on the talk until Wednesday, and try to improve some plots for that…this means a small delay for you. But until then you can clarify the cut-flow and event weighting on the background and compressed+accessible, and then after Wednesday add the compressed+compressed I will connect to the Higgsino meeting today, but I guess we don’t have the update for today, it will have to go via Jira ticket but we can start to add details to the note you could add the parts with plots for accessible+compressed we showed last time and after Wednesday, I will add the intro does this sound good for you?

[29.05.17, 11:41:51 AM] Yu-Ting: Yes

[29.05.17, 11:42:43 AM] Judita Mamuzic: ok :-)

[ytatus94]

• 20170531

[31.05.17, 3:58:23 PM] Yu-Ting: Do you finish your meeting/presentation?

[31.05.17, 4:03:56 PM] Judita Mamuzic: there is one more presentation in the meeting btw, are you connecting to susy meetings? and electroweak you should and background forum

[31.05.17, 4:04:58 PM] Yu-Ting: no I didn't connect to those meeting I find something strange in the m12 common ntuple so I ask Joey's help. For the cutflow comparison, I finished it yesterday but my values are quite different. I sent a mail to Joey too. And he found he did somewrong in the framework which produces the common ntuple.

[31.05.17, 4:08:45 PM] Judita Mamuzic: wow, great!

[31.05.17, 4:09:08 PM] Yu-Ting: I need to wait he produces new version of the ntuple and update his cutflow values. So I cannot prepare the new kinematic plots using m12=600. I am working on the significance plots now. After I finish this, I will start to work on the truth level study.

[31.05.17, 4:10:31 PM] Judita Mamuzic: sounds good I will start to work on the event generation tomorrow today I have a few more things to finish

[31.05.17, 4:11:27 PM] Yu-Ting: it's okay, I just want to let you know the status :)

[31.05.17, 4:11:55 PM] Judita Mamuzic: great, I’m very happy you helped them to find a bug this speeds up everybody :-) when I have the evgen files, I will let you know, and we could meet to see the status good work :-D

[31.05.17, 4:12:44 PM] Yu-Ting: thank you.

[ytatus94]

• 20170602

[02.06.17, 10:37:48 AM] Judita Mamuzic: Hi Yu-Ting Julia has sent around the slides for the EWK meeting today, and it is nice they already included the NUHM2 plots but I was wondering, could you perhaps redo them with a bit better range, and finer binning mll could be in the range 0-500 GeV, and ptll also 0-500 GeV, and perhaps double the number of bins we have now, then there will be 4 bins for the current one, that should be ok if you manage, could you please send them to met to see, and if we are happy, we could ask Julia to update the slides perhaps put 100 bins for both ranges then 1 bin will be 5 GeV

[02.06.17, 10:46:26 AM] Yu-Ting: Hi Judita, I am in the dental clinic now. I can try to do that after my dental appointment.

[02.06.17, 10:46:35 AM] Judita Mamuzic: oh, sorry, I forgot ok, good luck with the dentist never mind the plots if you manage later, would be good, but if not, then we will defend in words :-D if they ask something but also, the real analysis is on compressed+compressed, so we need to continue that... ok, good luck meet you later

[02.06.17, 3:33:52 PM] Yu-Ting: b.tgz Hi Judita, Here are the new plots. But they look strange to me

[02.06.17, 3:37:58 PM] Judita Mamuzic: could you also change the range? 0-300 is fine for both

[02.06.17, 3:39:16 PM] Yu-Ting: Sorry, I am in OU group meeting now. I think it is tool late to update the range of plots after OU meeting.

[02.06.17, 3:39:29 PM] Judita Mamuzic: ok, lets not send these we will connect, and give arguemtns

[02.06.17, 3:47:18 PM] Yu-Ting: 0-300 for mll and signal pT distributions?

[02.06.17, 3:47:38 PM] Judita Mamuzic: yes, but I am thainking not to send them there is a peak at Z for mll which is very different from the Higgsion I prefer we make the study with all these details this is compressed+accessible, so that one has this mass difference and there are decays with on shell Z but for compressed+compressed we will not have so it kind of sticks out so better make the study for compressed+compressed if Pat is there, you could discuss with him do you agree? its a guess lets leave it as it is we can say we already have more fresh plots, and are working on the comp+comp lets leave it as it is

[02.06.17, 3:54:59 PM] Yu-Ting: ok and the plots I sent to you are normalized to one

[02.06.17, 3:55:29 PM] Judita Mamuzic: thanks for providing the plots so fast, but I think perhaps not shake things yes, they seem correct, and the on-Z peak totally makes sense

[02.06.17, 3:55:45 PM] Yu-Ting: if I normalize to correct luminosity, then the "size" of the distributions are quite different.

[02.06.17, 3:56:17 PM] Judita Mamuzic: never mind the nomralisation, we dont really have the good filt.eff

[02.06.17, 3:56:18 PM] Yu-Ting: I don't understand your explanation about Z peak

[02.06.17, 3:56:30 PM] Judita Mamuzic: I will explain next time we meet

[02.06.17, 3:56:45 PM] Yu-Ting: let me try to implement the filter efficiency and to see the results again

[02.06.17, 3:56:46 PM] Judita Mamuzic: so the compressed+accessible has bigger mass differences accesible in particular and there mass difference is > m_Z and there are decays with on-shell Z that is why we see the peak at Z mass but compressed+comp has smaller mass diff

[02.06.17, 3:58:00 PM] Yu-Ting: ok. because the bigger mass differences, there are spaces to generate Z

[02.06.17, 3:58:00 PM] Judita Mamuzic: and Z is produced off shell got to go

[02.06.17, 3:58:29 PM] Yu-Ting: but the compressed+compressed has small mass difference, there is no space to produce off-shell Z

[02.06.17, 4:07:37 PM] Judita Mamuzic: ok, you could produce them in the range up to 70 GeV :-D ok, I prefer we don’t raise this question, the xsec is small for accessible+compressed and we will have the correct comp+comp prod for the next meeting this is work in progress or 0-50 GeV then it should be fine the Higgsino analysis focuses there in any case I lost sound, hmm

[02.06.17, 5:57:51 PM] Yu-Ting: me too try to reconnect but still the same

[02.06.17, 5:58:58 PM] Judita Mamuzic: hmm ok, wish you a nice weekend, I guess she stressed how much we need to make the trutl level analysis :-) but there were no extra questions, so I think this is all OK we continue on JIRA

[02.06.17, 6:13:18 PM] Yu-Ting: yes i am downloading new common ntuples now and will update the cutflow results using the new one.

[02.06.17, 6:14:08 PM] Judita Mamuzic: ah, great, how fast, that is amazing! :-D I will work on the jobOptions and event generation on the weekend will let you know when I have the samples ok, ciao for now :-)

[02.06.17, 6:14:41 PM] Yu-Ting: bye good weekend

[02.06.17, 6:15:05 PM] Judita Mamuzic: thanks! :-) same for you :-)

[ytatus94]

• 20170606

[06.06.17, 2:30:37 PM] Yu-Ting: Hi Judita, Pat wants to ask about the status of MC production.

[06.06.17, 2:31:20 PM] Judita Mamuzic: It is not yet finished, I need 1 more day, I think I have a meeting soon, so I will be away

[06.06.17, 2:31:40 PM] Yu-Ting: ok thank you When you finish your meeting, can we discuss what we want to present tomorrow?

[06.06.17, 5:20:04 PM] Judita Mamuzic: I will let you know when my meeting is finished

[06.06.17, 5:20:27 PM] Yu-Ting: thank you

[06.06.17, 6:13:06 PM] Judita Mamuzic: Hi Yu-Ting I will reply tomorrow 10 AM sounds good is this the same you wanted to discuss this evening?

[06.06.17, 6:13:57 PM] Yu-Ting: No I just want to discuss what we have to show tomorrow. Pat wants to discuss about the MC generation

[06.06.17, 6:14:39 PM] Judita Mamuzic: yes, I am working on it, but I wasnt working on it today up to now...

[06.06.17, 6:14:41 PM] Yu-Ting: Because we don't have compressed+compressed samples, I only can use compressed+accessible to show.

[06.06.17, 6:14:57 PM] Judita Mamuzic: ok, can I call?

[06.06.17, 6:15:04 PM] Yu-Ting: yes Call 11 minutes 10 seconds

[ytatus94]

• 20170607

[07.06.17, 3:57:31 PM] Yu-Ting: Hi Judita, I would like to update the mll plot in Joey's slide for tomorrow EB meeting h_mll.pdf Using this one. Pat said ok but I want to ask your opinion.

[07.06.17, 3:58:31 PM] Judita Mamuzic: it is ok, but can you put a better range, up to ~300 GeV and you could try the log scale on y axis this should be fast to change do you know how to?

[07.06.17, 3:58:52 PM] Yu-Ting: yes

[07.06.17, 3:59:08 PM] Judita Mamuzic: go ahead then :-D and its ATLAS Internal I think looks very nice

[07.06.17, 4:16:44 PM] Yu-Ting: done. I will work on the internal note now.

[07.06.17, 4:21:22 PM] Judita Mamuzic: great, Joanna was just asking about that I would ask two more things to change on the slides: I have one more smaller comment on NUHM2, concerning the last two sentences on the slide 31. I would put: Existing NUHM2 samles are accessible+compressed. Kinematically differ from Higgsinos! Compressed+compressed samples in preparation. Mass spectrum and cross-sections similar to Higgsinos. Truth level-comparison to Higgsinos kinematical distributions on the way.

[ytatus94]

• 20170608

[08.06.17, 11:18:46 AM] Judita Mamuzic: Hi Yu-Ting, I have created a tex file, and plan to upload the cross-section and mass plots is this ok with you?

[08.06.17, 11:19:16 AM] Yu-Ting: yes

[08.06.17, 11:19:32 AM] Judita Mamuzic: ok, I will upload now, and you can continue from there, filling with you plots do you need some explanations how to co the int note?

[08.06.17, 11:20:28 AM] Yu-Ting: no, I know how to do that thank you

[08.06.17, 11:21:34 AM] Judita Mamuzic: ok, great \so I have created a tex file apndx-nuhm2.tex, where we will be adding the tex, and the foder figures/nuhm2 where we ill be putting the plots Hi Yu-Ting, I made another small formating chage are you able to upload now? I was able from the begining I have created a subsection \subsection{Analysis using a selection with two soft leptons}, be free to edit and change in any way you think is good Since we work on the same file, I will be announcing if I plan to make an upload from now on…

[08.06.17, 11:44:16 AM] Yu-Ting: ok thank you. Joana said she modify the setting so I should be able to commit now but I didn't try to do it yet

[08.06.17, 11:44:41 AM] Judita Mamuzic: ok, if you managed to check out so far, please do svn up from now on, for simplicity, I will not make changes to the Analysis subsection, so we should have no clashes It looks like they have a strategy to quote JIRA tickets, so we will be doing that as well

[08.06.17, 11:46:39 AM] Yu-Ting: You can do any change you want. If there is conflict, then I will correct it.

[08.06.17, 11:46:48 AM] Judita Mamuzic: ok

[08.06.17, 11:47:11 AM] Yu-Ting: I will not upload to the tex svn before I leave the office today. but I may upload plots

[08.06.17, 11:47:36 AM] Judita Mamuzic: yes, that will be good for today, but we have to finish, with all corrections for tomorrow they are sending it on Friday in your own time I dont remember the format, it is a bit better if they are eps files but this is details, and we can change that also later on...

[08.06.17, 11:50:20 AM] Yu-Ting: what do you mean the strategy to quote JIRA ticket?

[08.06.17, 11:50:54 AM] Judita Mamuzic: I mean habit, when they mention a topic, they also add the link to the JIRA ticket where it was discussed

[08.06.17, 11:51:45 AM] Yu-Ting: i see

[08.06.17, 11:52:08 AM] Judita Mamuzic: also, we can make these tweaks later on, for now, lets try to have something :-D ok, ciao for now Hi, I connected, did they ask something about nuhm2?

[08.06.17, 2:59:54 PM] Yu-Ting: not yet now is at page 10

[08.06.17, 3:00:01 PM] Judita Mamuzic: great thanks

[08.06.17, 3:00:09 PM] Yu-Ting: nuhm2 is about page 30

[08.06.17, 3:01:36 PM] Judita Mamuzic: great wohoo :-) now we have to write :-D

[08.06.17, 4:07:02 PM] Yu-Ting: yes I am working on it

[08.06.17, 4:07:07 PM] Judita Mamuzic: great :-D Hi Yu-Ting, we should not bother Mike and Joanna too much with svn questions. They are very busy, and we should keep them available for the difficult topics :-) https://twiki.cern.ch/twiki/bin/view/AtlasComputing/SoftwareDevelopmentWorkBookSVN

[ytatus94]

• 20170609

[09.06.17, 12:11:30 AM] Yu-Ting: Hi Judita, I just finish editing the appendix and commit to the SVN.

[09.06.17, 6:39:25 AM] Judita Mamuzic: Thanks! I will take a look Very nice, did Pat help you? :-) I guess I don’t have to skip my trigger meeting this morning then :-) It’s good to have help :-D

[09.06.17, 2:02:54 PM] Yu-Ting: Hi Judita, can I get the compressed+compressed sample today?

[09.06.17, 2:06:48 PM] Judita Mamuzic: Yes, I will try finish

[ytatus94]

• 20170612

[12.06.17, 7:17:17 PM] Yu-Ting: Hi Judita, do i have to add the EwkHiggsino2016 into LinkDef.h I just finish modifying the EwkHiggsino2016

[12.06.17, 7:18:35 PM] Judita Mamuzic: Hi Yu-Ting, so great to hear from you you just need to rc compile I am preparing the MAdSpin, had to re-write the jo file was not able to add madspin in previous configuration now I am running madspin, but not all is read out, so it is crashing but soon I hope

[12.06.17, 7:20:44 PM] Yu-Ting: the simpleAnalysis also behaves strange. it replys this after I compile it

[yushen@lxplus063 SimpleAnalysis]$ simpleAnalysis
-bash: simpleAnalysis: command not found

[12.06.17, 7:20:47 PM] Judita Mamuzic: the 5000 events job without madspin was also crashing, so we need to have the right joboptions aaha I had that did you source

[12.06.17, 7:21:09 PM] Yu-Ting: The simpleAnalysis doesn't exist in RootCoreBin/bin/x86_64-slc6-gcc49-opt/

[12.06.17, 7:21:23 PM] Judita Mamuzic: so start from a new shell and recompile everything its fast then it will be ok

[12.06.17, 7:21:42 PM] Yu-Ting: I did it from a new shell but the situation still the same

[12.06.17, 7:21:47 PM] Judita Mamuzic: so it was working for you before?

[12.06.17, 7:21:50 PM] Yu-Ting: no never work

[12.06.17, 7:22:00 PM] Judita Mamuzic: hmmm lets chat

[12.06.17, 7:46:51 PM] Yu-Ting: Call 24 minutes 27 seconds

[ytatus94]

• 20170613

[13.06.17, 6:45:47 PM] Yu-Ting: Hi Judita, how is the status about the MC production?

[13.06.17, 6:47:02 PM] Judita Mamuzic: actually, I am very close to havign the MAdSpin working

[13.06.17, 6:48:59 PM] Yu-Ting: ok, do you think we can have it ready by tomorrow?

[13.06.17, 6:49:35 PM] Judita Mamuzic: I am working on it… I will submit 5000 events as soon as it is ready and the resot of your code is it working the SimpleAnalysis and TRUTH3 went OK for the 1000 events? try to complete the chain, and you can already prepare the Higgsino 160_150 point so have everything ready, and then we will just update with 5000 events

[13.06.17, 6:51:41 PM] Yu-Ting: it works but the statistic is very low

[13.06.17, 6:51:51 PM] Judita Mamuzic: yes, I guess so

[13.06.17, 6:51:56 PM] Yu-Ting: FAR is on this Friday, I think we should have some plots to show

[13.06.17, 6:52:03 PM] Judita Mamuzic: yes, it would be good so I will let you know whe the jo is ready, and 5000 events is running there were quite a few problems

[13.06.17, 6:53:22 PM] Yu-Ting: I will upload the plots made by 1000 events to JIRA later today or tomorrow morning

[13.06.17, 6:53:54 PM] Judita Mamuzic: should I look at them, or Pat took a look already? it would be good

[13.06.17, 6:54:31 PM] Yu-Ting: not yet. Some of the distributions look strange so I modify my framework and rerun it now. I will let you see the plots before I upload to JIRA

[13.06.17, 6:56:01 PM] Judita Mamuzic: yes, it would be good, better have one part of the discussion offline

[13.06.17, 7:06:22 PM] Yu-Ting: ddd.tgz Some of the plots still looks strange to me but I cannot find the bugs. So I send all the plots to you.

[13.06.17, 7:07:04 PM] Judita Mamuzic: woow very good

[13.06.17, 7:07:30 PM] Yu-Ting: mll, pull, dphimn1, rll, mtautau are strange to me

[13.06.17, 7:07:35 PM] Judita Mamuzic: perhaps it is enough to take the signal selection not the baseline

[13.06.17, 7:07:43 PM] Yu-Ting: and some other plots too (I don't remember all)

[13.06.17, 7:08:50 PM] Judita Mamuzic: ok, there you go, the answer is mc weight

[13.06.17, 7:08:52 PM] Yu-Ting: all the variables without baseline or signal prefix are basically using baseline, except mt2. mt2 should use signal and I confirmed this with joey

[13.06.17, 7:09:26 PM] Judita Mamuzic: and howcome we dont have the met cut? yes

[13.06.17, 7:09:56 PM] Yu-Ting: I don't know the reason why higgsion has distribution of mc_weight but nuhm2 doesn't

[13.06.17, 7:10:04 PM] Judita Mamuzic: so the let12 seem different ok, mll is strange??

[13.06.17, 7:10:26 PM] Yu-Ting: met cut? I don't understand your question.

[13.06.17, 7:10:30 PM] Judita Mamuzic: looks like a bog ah, put on log and we have low stats ptll also on log it is really nice to see these distributions

[13.06.17, 7:12:31 PM] Yu-Ting: let12 different? which one? there is no let12. bog? do you mean bug? I already correct the bug of mll and still get the similar distribution. I don't know what log file you mentioned. oh~ I see you mean log scale. sorry I misunderstood it

[13.06.17, 7:12:52 PM] Judita Mamuzic: ok, so signal electrons eta has these horns not sure it should be like that there are 2 variables with lepton12 ah, the eta is ok… why are signal jets cut at 20 GeV? ah, it is jets sorry ok, very nice to see these I would not show anything for baseline, its for our check

[13.06.17, 7:15:14 PM] Yu-Ting: jet has pT>20 requirements you can find it from Joey's EB talk.

[13.06.17, 7:15:15 PM] Judita Mamuzic: the signal selection is more important yes yes, I was thinking of lepton pt, then I realised its jet its ok ok, try to fix those problematic plots, and perhaps setup some slides and write out some conclusions and I will try to have the MAdSpin soon, so perhaps you can update it very nice there are similarities, but it is not identical so it is very interesting also, did you manage to do the correct event weighting I think it would be good to have the weighted plots for now...

[13.06.17, 7:17:49 PM] Yu-Ting: I don't know how to do the event weight things for this truth study.

[13.06.17, 7:18:08 PM] Judita Mamuzic: take a look at the script I have sent you that one has event weighting, but you need to adjust it

[13.06.17, 7:18:24 PM] Yu-Ting: I don't know how to access the cross sections from the truth file.

[13.06.17, 7:18:28 PM] Judita Mamuzic: because for NUHM2 we have different processes nono, cross-section is in the SUSYTools

[13.06.17, 7:18:48 PM] Yu-Ting: I have event weight in my results. I just don't know how to use it.

[13.06.17, 7:18:53 PM] Judita Mamuzic: but you use 1 point ok, take a look, you have all the ingredients if you have problems, we will send these, but we have to have also the weighted plots as we will wait long for the sampels but really great to see these distributions put up the slides and first round could be with no weighting but by the end of this week, by Friday, we should have the weighted ones ok, going now MadSpin is waiting for me :-D

[13.06.17, 7:20:56 PM] Yu-Ting: in your code, you don't have anything relate event weight but I retrieve the event weight and mc weight and i don't know how to use them.

[13.06.17, 7:21:25 PM] Judita Mamuzic: yes, truth has no event weighting, so you do it yourself ok, for the Higgsino sample, you have the problem solved in my code, as it is 1 process try that first and then in my code there is no case with different processes so this will have to be added but spend some time first to look at the plots, and draw some conclusions from the distributions, and the cuts we use

[ytatus94]

• 20170614

[14.06.17, 2:36:10 PM] Yu-Ting: ddd.tgz New plots

[14.06.17, 2:41:20 PM] Judita Mamuzic: ok, this is still not the truth higgsino sample

[14.06.17, 2:41:35 PM] Yu-Ting: this is truth higgsinos sample

[14.06.17, 2:41:54 PM] Judita Mamuzic: and what about the mc_weight?

[14.06.17, 2:42:11 PM] Yu-Ting: I used truth higgsinos sample. The one I sent to you also use the truth Higgsino sample I don't know why it has distribution

[14.06.17, 2:42:40 PM] Judita Mamuzic: what is filled in the mc_weight? the SimpleAnalysis does not fill mc_weight so I dont know what is that

[14.06.17, 2:43:21 PM] Yu-Ting: you didn't do it in your code but I did

float  mc_weight        = event->getMCWeights()[0];

I copy this line from other example

[14.06.17, 2:43:43 PM] Judita Mamuzic: I have to go now, have a meeting to me this looks like we can use only the compressed+compressed and we can argue wether to use filter or not for the JIRA, perhaps it is enough to use the 100k samples and something goes wrong with met and Rll is very different and we need to understand ptll where is unfiltered mll? try to put a cut on met 50 GeV, and redo the NUHM2 plots, it goes crazy after normalization this should improve all variables with met because we are using the normalized to 1 and the HT variable should be filled for > 0 GeV, not to have that first bin and mll this is all good news, only Rll has a very different distribution not sure what is the definition, so this one we need to understand

[14.06.17, 3:29:23 PM] Yu-Ting: i will use what I have now to make slides and to see what kind of feedback I will get.

[14.06.17, 3:29:34 PM] Judita Mamuzic: but try to add these improvements that I wrote they will tell you the same the MET cut is important I am on a different meeting, not sure if there is the Oklahoma meeting….

[14.06.17, 4:12:21 PM] Yu-Ting: no today is Wednesday. The OU meeting is on Thursday

[14.06.17, 4:12:32 PM] Judita Mamuzic: Ah! I mixed the days :-D ok is mtautau, the same as mll, only for taus? why does it go into negative?

[14.06.17, 5:59:06 PM] Yu-Ting: it is discussed in HIGGSINO-11 There are 3 way to calculate MTauTau and they finally decide to adopt the second way which allows the negative values

[14.06.17, 6:00:28 PM] Judita Mamuzic: ok, thanks Hi Yu-Ting, I think it would be good to have mll, ptll etc plots with > 0, and then do the normalization, the similar is for MET > 50 GeV then we would have good variables with MET, and perhaps mll would improve and when you finish that, we should ask to change that plot on the slides from Lorenco the mll plot and for those plots, it is sufficient to use e.g. filtered then its easier to compare, no need for all Ah, perhaps extend mll range, perhaps it stops at 30 GeV 10 is very low in any case... so on your slide 10, HT30 should not have that first bin ever filled so there is some problem in your code please check why is that so…

[ytatus94]

• 20170615

[15.06.17, 3:37:20 PM] Yu-Ting: Hi Judita, are you able to atten OU meeting at 4:30pm today? And I notice a thread to discuss MadSpin. I don't know what kind of MadSpin problem you had but you may take a look at this

[15.06.17, 3:45:29 PM] Judita Mamuzic: I will attend thanks it is still not solved, this might help, but looks like rel21, not 19 oh, they switch to 19 Yu-Ting did you try the met cut plots I think it would be good and to ask them to change the plot

[15.06.17, 3:49:21 PM] Yu-Ting: I did it already and I am making slide but I find HTLep12 calculation was wrong so I have to rerun everything again

[15.06.17, 3:50:03 PM] Judita Mamuzic: you don't need all for now the met cut is there? also, you don't meed the small no of ev samples can we chat

[15.06.17, 3:50:59 PM] Yu-Ting: what is ev?

[15.06.17, 3:51:10 PM] Judita Mamuzic: lets chat, ev is events

[15.06.17, 3:51:23 PM] Yu-Ting: ok Call 23 minutes 12 seconds

[ytatus94]

• 20170616

[16.06.17, 10:07:53 AM] Judita Mamuzic: Yu-Ting, the mll distribution actually looks good for NUHM2, it shows the kinematic edge, this is exactly what we should see it is a bit higher than expected ~18 GeV, so this we need to understand

[ytatus94]

• 20170619

[19.06.17, 10:58:55 AM] Judita Mamuzic: Hi Yu-Ting, just to check, how is it going?

[19.06.17, 10:59:36 AM] Yu-Ting: Hi Judita, I am working on it.

[19.06.17, 10:59:39 AM] Judita Mamuzic: great I'm afraid MAdSpin is not yet solved but I am working on it...

[19.06.17, 11:02:03 AM] Yu-Ting: It is okay because I will not include anything related MadSpin in the talk today

[19.06.17, 11:02:23 AM] Judita Mamuzic: ok, they will ask for sure, but we will answer as we did on jira

[19.06.17, 11:03:34 AM] Yu-Ting: They will not ask because I mentioned it at the beginning of my slides

[19.06.17, 11:03:49 AM] Judita Mamuzic: in any case, we need to do it…

[19.06.17, 11:06:06 AM] Yu-Ting: what does PS mean in your reply?

[19.06.17, 11:06:13 AM] Judita Mamuzic: parton shower

[19.06.17, 11:06:22 AM] Yu-Ting: ok, i see

[19.06.17, 11:06:42 AM] Judita Mamuzic: we generate events with MadGraph, and Pythia is doing the PartonShower

[19.06.17, 12:14:14 PM] Yu-Ting: main.pdf Here is the slides for today

[19.06.17, 12:15:40 PM] Judita Mamuzic: thanks, taking a look now

[19.06.17, 12:16:53 PM] Yu-Ting: I will go to lunch now.

[19.06.17, 12:17:14 PM] Judita Mamuzic: ok.... ah, cool, stacked histograms :-) huh, I am not sure we are using the right samples I think they should be all 170_150 ok, I just check with Joanna, all points should be 170_150 so if we are not going to be able to put those, we should not use the 155_150 points so we should go back to N2N1 plots it is wrong to add 170_150 and 155_150 points the naming goes for N2N1 mass difference but perhaps you could run the N1C1p N1C1m 170_150 points, and replace them that would be the best that would mean re-running TRUTH3, SimpleAnalsyis and your framework perhaps its possible OK, I think we should not show these plots, its not correct, either go to previous version, or perhaps you can make the update I can help in something Hi, you are here! do you think you could make it in 1h? we can have a quick chat ok, I will be away for a few min but please replace the plots, because we use normalization, the Higgsino lines are very much off, because we were using a combination of 2 points. So please, either go back to N2N1 plots, and update on JIRA later on, or try to redo with all 170_150 points Yu-Ting, are you back, let me know when you are here, to have a small chat lets chat Call – no answer

[19.06.17, 1:43:35 PM] Yu-Ting: Hi Judita, I will not change my slides anyway because it is impossible to redo everything before the meeting. I will just use the current results. And I asked which sample I have to use on JIRA on last Saturday. I listed I will use N2C1_155_150 and Mike replied me, he didn't say this is wrong. As I know, the naming is N2=155 and C1=150 so the mass difference =5 which closes to the mass difference in NUHM2 (dm=6). If they insist to use 170_150, then I will update the new results on JIRA but not for the meeting. For the old N2N1_170_150 results, they are on the JIRA, everyone can see them. I don't need to present them again.

[19.06.17, 1:43:55 PM] Judita Mamuzic: nooo then at least add this discussion to the slides you can say it is not possible to make the right comparison since for NUHM2 C1 is not in the middle of N2-N1 I missed that Mike said its ok to have 155_150 point I see it, I should have corrected it

[19.06.17, 1:47:55 PM] Yu-Ting: I can raise this in the talk and update if they want to use 170_150 for all N2N1, N2C1p N2C1m

*[19.06.17, 1:48:24 PM] Judita Mamuzic: I guarantee they will raise it, dm is visible in the mll plot and they will ask exactly this the naming is always N2N1 and for higgsino C1 is 1/2 (N2-N1) I am not sure we should show these plots the results we have are much nicer...**

[19.06.17, 1:52:03 PM] Yu-Ting: The EVNT files are converting to TRUTH3 now and it is impossible to get the results before the talk.

[19.06.17, 1:52:18 PM] Judita Mamuzic: how about reverting to previous you can put these to backup

[19.06.17, 1:53:21 PM] Yu-Ting: And from the current plots, we still can see the N2N1 distributions. If just want to compare the shape, then the small differences causing by the normalization doesn't not a big problem.

[19.06.17, 1:53:44 PM] Judita Mamuzic: but you showed this in the previous round that is much better than showing the 155_150 the mll plot you posted to jira yesterday that is what I mean do you have that version? the 155_150 gives N2-N1 mass diference it is really the problem of using the wrong naming

[19.06.17, 1:54:49 PM] Yu-Ting: if only the mll plot, I can download it from JIRA

[19.06.17, 1:54:52 PM] Judita Mamuzic: or be very charming and go with what you have :-D yes, add it, to point out you know that it should be 170_150

[19.06.17, 1:55:57 PM] Yu-Ting: my talk is about 3pm, if i get the new results, then i will update the slides. otherwise, just use the current one.

[19.06.17, 1:56:09 PM] Judita Mamuzic: what does pat say? I think we gain more if you show them next time and continue on Jira until then the results are very nice, plots are excellent, but its the wrong samples and you can see it so the problem was in the naming, because X_Y in the naming is for N2_N1, even when its a N2C1 production, and C1 is fixed to (N2-N1)/2 ask Pat what he thinks, and do as you two decide... we can ask to shift your talk to the end and either replace with just N2N1 for Higgsino or if we are lucky, truth + all the rest will finish should I write a email to Joana? Or you can also write

[19.06.17, 2:09:57 PM] Yu-Ting: If I need to switch the presentation order, then I will let her know later.

[19.06.17, 2:10:06 PM] Judita Mamuzic: ok Ah, you managed! great, I saw your slides

[19.06.17, 2:28:11 PM] Yu-Ting: yes

[19.06.17, 2:28:36 PM] Judita Mamuzic: so the difference is quite large in the jet and lepton multiplicity this will reflect in the sensitivity but the scaling will be possible... so all can change with MadSpin, and we need to see next the normalized to lumi perhaps these 2 you can put to backup... or mention, if you like or if they ask but the kinematic variable shape has fantastic agreement but the labels are missing, are you adding them? very nice! thanks for making the update so fast, much much better like this, good luck! ah, also, for the lepton and jet multiplicity it could be ok, because we include N1N1 production, which has no jets and leptons! so it is ok! Excellent! it was worth the extra work!!! so glad you managed, and you presented it very nice!

[19.06.17, 2:51:15 PM] Yu-Ting: thank you but pat asked me why there are 0 leptons because we have a 2 lepton filter. Actually, I also notice this and I need to figure out the reason why.

[19.06.17, 2:52:05 PM] Judita Mamuzic: ah, it is a good point perhaps the same feature you had for the 0 im Ht etc plots probably you are setting to 0 something that is not passing the selection it would be good to find that bug… but multiplicity you should not just set to 0 because some are real 0 so the production is 2 L filter but the plots are after quality criteria selection so some are possibly with very low pt in this case, N1N1 is actually not passing the 2L filter…so that reasoning was probably not the main reason

[19.06.17, 2:59:45 PM] Yu-Ting: pat said that maybe a tau event and ask me to check.

[19.06.17, 3:01:18 PM] Judita Mamuzic: yes, perhaps you can see if the same event has a tau at the truth level… do you know what to do? is there a vairable like final state, or FS in the TRUTH3?

[19.06.17, 3:05:44 PM] Yu-Ting: The slides with histogram titles are uploaded

[19.06.17, 3:05:53 PM] Judita Mamuzic: perfect! ups, I think you didnt upload the new one its still missing the lables

[19.06.17, 3:07:03 PM] Yu-Ting: you have to refresh your indico page

[19.06.17, 3:07:13 PM] Judita Mamuzic: I did, perhaps indico is lsow

[19.06.17, 3:07:16 PM] Yu-Ting: because I can see the new one on the indico

[19.06.17, 3:15:27 PM] Judita Mamuzic: ah, I see, you label them on top yes, I can see them you can add also the x axis, for e.g. with if statement, if histo name == something, x label is something… you will need that for your thesis ah, make sure to have good book keeping of all these plots later you will need to find them also, the thesis is better to have the eps format, not sure if you already have those, probably yes, as you have your presentation in latex

[19.06.17, 3:20:04 PM] Yu-Ting: eps is not needed because I use pdfLatex. It converts all eps plots to plots

[19.06.17, 3:20:42 PM] Judita Mamuzic: check with your institute, I had to have the eps versions... its lighter later, when you need to redo 100 plots it will be much harder :-D PErhaps we could meet this week to discuss next steps I have some urgent work for tomorrow

[19.06.17, 3:48:18 PM] Yu-Ting: okay I will let Pat knows he is not here now

[19.06.17, 3:48:37 PM] Judita Mamuzic: ok in the meanwhile, you could look into event weighting, and even the mll distribution…that will be a bit tricky :-) and also start preparing everything for the sensitivity plots

[ytatus94]

• 20170622

[22.06.17, 11:18:07 AM] Yu-Ting: Hi Judita, can I find the MadGraph run card, param card, proc card on the SVN? I don't know which one is which.

[22.06.17, 11:19:08 AM] Judita Mamuzic: no, they are generated in the job options, except the param card is the slha file

[22.06.17, 11:21:00 AM] Yu-Ting: so I only can find the control file and slha file on the svn

[22.06.17, 11:21:15 AM] Judita Mamuzic: yes

[22.06.17, 11:21:45 AM] Yu-Ting: than in all of the output files, which is run card and proc card?

[22.06.17, 11:22:21 AM] Judita Mamuzic: run_card.dat and proc_card.dat

[22.06.17, 11:24:29 AM] Yu-Ting: I found run_card.dat and run_card.SM.dat in Chris' directory. Does MadGraph produce two run cards, one for SUSY one for SM?

[22.06.17, 11:25:46 AM] Judita Mamuzic: yes, the run_card.dat has specifications for NUHM2, and the second is run by default this is how I understand it

[22.06.17, 11:26:33 AM] Yu-Ting: ok thank you

[22.06.17, 11:27:17 AM] Judita Mamuzic: np The meeting is ongoing so I will not connect to the Oklahoma meeting

[22.06.17, 4:23:06 PM] Yu-Ting: i see.

[22.06.17, 4:23:55 PM] Judita Mamuzic: I can connect right after

[22.06.17, 4:24:29 PM] Yu-Ting: no one in the ou meeting room now. we are all in eb meeting room

[22.06.17, 4:27:25 PM] Judita Mamuzic: Ok ok, is everybody connecting to the Oklahoma meeting? or we skip

[22.06.17, 4:36:51 PM] Yu-Ting: yes we are in the ou meeting room now

[22.06.17, 5:55:15 PM] Judita Mamuzic: Mike is away until Monday just got a return email

[22.06.17, 5:59:29 PM] Yu-Ting: Ok, I will let Pat know this. He is not in the office now

[22.06.17, 5:59:41 PM] Judita Mamuzic: great, thanks! just so you know…

[ytatus94]

• 20170628

[28.06.17, 9:59:22 AM] Judita Mamuzic: Hi Yu-Ting, I am woking on the jobOptions, will try to send you a working example today Hi Yu-Ting, I managed the MadSpin, for now I have a simple example for N1N2, I made a copy to /afs/cern.ch/user/m/mamuzic/public/ForYuTing/JO the kreci.sh file has the release and command details I will work on adding all productions and filters now

[28.06.17, 3:33:59 PM] Yu-Ting: thank you i will take a look at that file

[28.06.17, 3:34:19 PM] Judita Mamuzic: I saw you managed to run 10k, I guess you are an expert by now ok :-D great you proved the mll shape! let me know if you spot problems I need 1-2 h to have all productions ready it takes a while to test them

[28.06.17, 3:35:08 PM] Yu-Ting: no I couldn't run it by myself. all the mc are produced by Chris

[28.06.17, 3:35:19 PM] Judita Mamuzic: ah, ok, then I will give you instructions chat?

[28.06.17, 3:35:29 PM] Yu-Ting: ok

[28.06.17, 3:35:44 PM] Judita Mamuzic: Call started asetup 19.2.5.21,MCProd,here

[28.06.17, 3:58:09 PM] Yu-Ting: Call ended 22 minutes 26 seconds

[28.06.17, 6:15:15 PM] Judita Mamuzic: Hi Yu-Ting, the test job for N2C1 is still running probably the 10k will not be ready this evening but perhaps, for tomorrow, you could set up a few slides, perhaps its enough to have the mll plots with previous setup, and the new one

[28.06.17, 7:45:11 PM] Yu-Ting: Hi Judita, it is okay. I am also running some testing jobs with 10k events

[28.06.17, 7:45:44 PM] Judita Mamuzic: it is very slow perhaps in 1h I will have 1k test job for N2C1p, and then we can run over night N2C1p 10k and N2C1m 10k but not sure all this will be ready for the meeting tomorrow morning

[28.06.17, 7:47:18 PM] Yu-Ting: could you please give me the job option which you are running now thank you

[28.06.17, 7:47:45 PM] Judita Mamuzic: yes, thats a good idea, you could start 10k for these two already and I will let you know how the test went, and we can stop the 10k jobs

[28.06.17, 7:49:36 PM] Yu-Ting: Since it is running on your side, you don't need to stop running the job. Just let it runs over night

[28.06.17, 7:53:11 PM] Judita Mamuzic: I submitted with 1k, so it should finish fast relatively so the JO are at /afs/cern.ch/user/m/mamuzic/public/ForYuTing/N2C1p and N2C1m only the sign of the C1 is changed in the control file the other files are unchanged, I just added them there for simplicity

[28.06.17, 7:58:44 PM] Yu-Ting: Thank you. I will copy them later. I need to catch the bus now

[28.06.17, 7:58:58 PM] Judita Mamuzic: Sure, have a nice evening P.S. This configuration is with an extra jet huh, found a problem with JO I have sent you the good news is, I was doubling something, so it will be faster... ok, I have made a new copy to /afs/cern.ch/user/m/mamuzic/public/ForYuTing/N2C1p and N2C2m and am running 10k events each this will now finish in ~1h but I will copy it to some public only in the morning... meet you tomorrow at Vidyo around 10

[28.06.17, 11:44:30 PM] Yu-Ting: sorry I didn't check your file yet

[28.06.17, 11:44:54 PM] Judita Mamuzic: np, if it is good, I will manage to generate 10k so all ok :-)

[ytatus94]

• 20170629

[29.06.17, 12:25:48 AM] Judita Mamuzic: the N2C1p and N2C1m are being copied to /afs/cern.ch/work/m/mamuzic/public/ForYuTing/NEW I hope the folder is visible… if not, I copy again tomorrow I added a link to your new mll plot in the agenda that one has everything we need for the discussion, I think

[29.06.17, 9:56:29 AM] Yu-Ting: good morning Judita

[29.06.17, 9:56:46 AM] Judita Mamuzic: hello, good morning :-)

[29.06.17, 9:57:11 AM] Yu-Ting: I just realize that the meeting you mentioned is the meeting with the convenors

[29.06.17, 9:57:27 AM] Judita Mamuzic: yes, one of them Mike was not able to join, then Joanna proposed to put Tommaso instead

[29.06.17, 9:57:44 AM] Yu-Ting: Originally, I thought you mentioned OU Thursday meeting.

[29.06.17, 9:57:50 AM] Judita Mamuzic: so its a bit different of how we imagined we meet in 3 min https://indico.cern.ch/event/649522/

[29.06.17, 9:58:46 AM] Yu-Ting: yes I checked your previous mail about this meeting

[ytatus94]

• 20170703

[03.07.17, 12:16:27 PM] Yu-Ting: Hi Judita, I prepare slides for today. Basically, it just contains one new plot and the others information are old plots and discussion from JIRA or meeting on last Thursday. I will go to lunch now. If you want me to modify something, then I will do it after lunch.

[03.07.17, 12:20:59 PM] Judita Mamuzic: All OK, very nice summary! thanks for preparing these Hi Yu-Ting can we have a short chat

[03.07.17, 3:39:11 PM] Yu-Ting: yes, when?

[03.07.17, 3:39:13 PM] Judita Mamuzic: now?

[03.07.17, 3:39:17 PM] Yu-Ting: sure

[03.07.17, 3:39:31 PM] Judita Mamuzic: Call started N2C1p, N2C1m, N2N1, C1C1 N2->llN1 N2->C1? N2->qqN1 12% N2->llN1

1) n2->ffx1, x1->ffn1 420-429 2) n2->qqn1 430-434 3) n2->lln1 435-437 4) n2-> vvn1 438-440

We can ignore: 2), 4) N2C1 C1 gives 1 l C1->WN1 at most 1l only 1) ~/public/ForYuTing/N2C1p/MadGraphControl_NUHM2_weak.py msdecaystring="decay n2 > l+ l- n1\ndecay x1+ > f f n1\n" decay n2 > ff x1+\ndecay x1+ > f f n1\n l+ vl l- vl~

[03.07.17, 4:18:36 PM] Yu-Ting: Call ended 39 minutes 6 seconds

[ytatus94]

• 20170704

[04.07.17, 12:49:20 PM] Yu-Ting: Hi Judita, does this affect the result?

PyJobTransforms.transform.execute 2017-07-04 12:43:03,396 CRITICAL Transform executor raised TransformValidationException: Non-zero return code from generate (65); Logfile error in log.generate: "Pythia8             FATAL ../src/GenModule.cxx:51 (StatusCode GenModule::execute()): code 0: this->callGenerator()"

[04.07.17, 1:21:25 PM] Judita Mamuzic: This looks like a fatal error

[04.07.17, 4:01:48 PM] Yu-Ting: Hi Judita, Only n2 > l+ l- n1 works and all the other decays have problems**

[04.07.17, 4:02:09 PM] Judita Mamuzic: I see this was with MAdSpin?

[04.07.17, 4:02:28 PM] Yu-Ting: I tried both MadGraph and MadSpin

[04.07.17, 4:02:30 PM] Judita Mamuzic: perhaps we could try over night MadGraph, no MadSpin oh, great hmmm this is puzzling a bit

[04.07.17, 4:02:47 PM] Yu-Ting: The MadGraph jobs are running more than 22 hours now

[04.07.17, 4:02:49 PM] Judita Mamuzic: it is some techinical reason ok, lets wait for MG these should be good you said last time it took 19h so ok, lets see

[04.07.17, 4:03:13 PM] Yu-Ting: And the output messages just show Permission denied when I text to you And terminate the jobs automatically The MadSpin one only n2 > l+ l- n1 work and the other don't work

[04.07.17, 4:04:54 PM] Judita Mamuzic: ok, it could be some technical reason perhaps we can try the Pythia decays compare Pythia all with Pythia Z->ll ok, I propose we stop here perahps you could start preparing the code for normalisation we still need to see if we have sensitivity with our realistic cross section it is smaller than higgsino, there they probably have factor 2-3 more

[04.07.17, 4:11:25 PM] Yu-Ting: And I have another problem not related above things

[04.07.17, 4:11:30 PM] Judita Mamuzic: yes?

[04.07.17, 4:12:45 PM] Yu-Ting: I modified the job option and make it works for all conditions. So we only need use one JO file and it works for all n1n2, n2c1p, n2c1m, c1c1 but i get this error

generate 15:42:18   File "/cvmfs/atlas.cern.ch/repo/sw/software/x86_64-slc6-gcc47-opt/19.2.5/AtlasProduction/19.2.5.14/InstallArea/jobOptions/EvgenJobTransforms/skeleton.GENtoEVGEN.py", line 225, in <module>
generate 15:42:18     include(jo)
generate 15:42:18   File "./MC15.370621.MGPy8EG_A14N23LO_NUHM2_m12_600_weak_N2N1.py", line 1, in <module>
generate 15:42:18     include('MadGraphControl_NUHM2_weak.py')
generate 15:42:18   File "./MadGraphControl_NUHM2_weak.py", line 185, in <module>
generate 15:42:18     process_dir = new_process(card_loc='proc_card_mg5.dat')
generate 15:42:18   File "/cvmfs/atlas.cern.ch/repo/sw/software/x86_64-slc6-gcc47-opt/19.2.5/MCProd/19.2.5.14.3/InstallArea/python/MadGraphControl/MadGraphUtils.py", line 107, in new_process
generate 15:42:18     raise RuntimeError('No diagrams for this process in dir='+str(thedir))
generate 15:42:18 RuntimeError: No diagrams for this process in dir=
generate 15:42:18 Py:Athena            INFO leaving with code 8: "an unknown exception occurred"
PyJobTransforms.trfExe.execute 2017-07-04 15:42:18,194 INFO generate executor returns 8
PyJobTransforms.trfExe.validate 2017-07-04 15:42:18,197 ERROR Validation of return code failed: Non-zero return code from generate (8) (Error code 65)
PyJobTransforms.trfExe.validate 2017-07-04 15:42:18,250 INFO Scanning logfile log.generate for errors
PyJobTransforms.trfValidation.scanLogFile 2017-07-04 15:42:18,253 WARNING Detected python exception - activating python exception grabber
PyJobTransforms.transform.execute 2017-07-04 15:42:18,253 CRITICAL Transform executor raised TransformValidationException: Non-zero return code from generate (8); Logfile error in log.generate: "RuntimeError: No diagrams for this process in dir="

It is very strange because I didn't touch this part

[04.07.17, 4:14:54 PM] Judita Mamuzic: ok, I am working on the general JO not sure about this error

[04.07.17, 4:59:12 PM] Yu-Ting: ok the problem was fixed. I forgot to output the diagrams

[04.07.17, 5:09:13 PM] Judita Mamuzic: I plan to submit the request tonight

[ytatus94]

• 20170705

[05.07.17, 5:46:19 PM] Yu-Ting: Hi Judita, any progress on your general JO? My version of general JO works for N2N1, N2C1p and N2C1m but not C1C1. I got pythia fatal error for C1C1

[05.07.17, 5:47:03 PM] Judita Mamuzic: I have it running the test job

[05.07.17, 5:47:17 PM] Yu-Ting: I have tested my JO to run MadGraph, MadSpin and Pythia. If I use pythia to decay, then all of them have fatal error.

[05.07.17, 5:47:50 PM] Judita Mamuzic: I will upload my now to JIRA there are many details

[05.07.17, 5:48:10 PM] Yu-Ting: but if I use MadSpin, then only C1C1 has fatal error so you don't have any error message using your JO? your run_card.dat and mine have several differences. But I didn't touch those parameters It is strange

[05.07.17, 6:40:07 PM] Judita Mamuzic: ok…there are many places…

[ytatus94]

• 20170706

[06.07.17, 4:36:06 PM] Yu-Ting: Hi Judita, are you able to attend OU meeting?

[06.07.17, 4:44:07 PM] Judita Mamuzic: Hi Yu-ting, sorry for scilence, im on the susy meeting The test jobs are running, i will sed the root files for 4 production, and then it would be good to make the plots

[06.07.17, 4:45:28 PM] Yu-Ting: it is okay

[06.07.17, 4:45:45 PM] Judita Mamuzic: But we need br for that The jobs i run are for the 600 point Hi @Yu-Ting, the SUSY meeting took a while. Did I miss some conclusions from the Oklahoma meeting?

[06.07.17, 7:02:24 PM] Yu-Ting: no, I just let Pat and Chris know what I am doing now.

[06.07.17, 7:02:56 PM] Judita Mamuzic: thanks :-) very good :-)

[06.07.17, 7:03:18 PM] Yu-Ting: Maybe we can arrange time to talk tomorrow

[06.07.17, 7:04:09 PM] Judita Mamuzic: sounds good, lets plan 10am?

[06.07.17, 7:04:45 PM] Yu-Ting: ok

[06.07.17, 7:05:39 PM] Judita Mamuzic: agreed :-)

[ytatus94]

• 20170707

[07.07.17, 9:24:58 AM] Judita Mamuzic: In r1 10 am

[07.07.17, 9:59:41 AM] Yu-Ting: Hi Judita

[07.07.17, 9:59:59 AM] Judita Mamuzic: hi

[07.07.17, 10:00:01 AM] Yu-Ting: Sorry, I miss the bus and the next one is 10:20

[07.07.17, 10:00:10 AM] Judita Mamuzic: ah, sure np

[07.07.17, 10:00:23 AM] Yu-Ting: So I will be at CERN after 10:35

[07.07.17, 10:00:38 AM] Judita Mamuzic: ok, then we meet at 10:40? R1?

[07.07.17, 10:00:45 AM] Yu-Ting: Sure Sorry for that

[07.07.17, 10:00:56 AM] Judita Mamuzic: ok, great :-) np, I lived in St.Genis 5 years, I know all the troubles with the bus :-) /afs/ific.uv.es/user/m/mamuzic/public/ForYuTing weight = xsec x kfac x gfe x (BR1 X BR2) x LUMI x 1000 / entries https://svnweb.cern.ch/trac/atlasoff/browser/PhysicsAnalysis/SUSYPhys/SUSYTools/trunk/data/mc15_13TeV/MGPy8EG_A14N23LO_Higgsino_2L.txt

[ytatus94]

• 20170710

[10.07.17, 10:51:07 AM] Yu-Ting: Hi Judita, I didn't get your files at weekend

[10.07.17, 10:51:28 AM] Judita Mamuzic: Hi Yu-Ting, by jobs were crashing again I still need to tweak the multipliers, or change the release I found that the problem I get is due to the MG release so go ahead with your study

[10.07.17, 10:52:07 AM] Yu-Ting: And I tried to run your JO, they finished with the same fatal error.

[10.07.17, 10:52:07 AM] Judita Mamuzic: if you like, you could send me the plots? so I can give you some remarks yes, testing time for the JO is quite long, so I am repeating now with next settings

[10.07.17, 10:53:12 AM] Yu-Ting: Because you said your job crashes again, do you have code 0: this->callGenerator() at the end of your log

[10.07.17, 10:53:20 AM] Judita Mamuzic: yes

[10.07.17, 10:53:28 AM] Yu-Ting: I think it is not because of the multiplier it is because of the MET filter

[10.07.17, 10:53:43 AM] Judita Mamuzic: I found on internet that it is due to MG version well, MET filter introduces the cut, so it can be the multiplier I will check if the MET filter is added correctly

[10.07.17, 10:54:32 AM] Yu-Ting: I search my log file and the Filter expression is MultiElecMuTauFilter and MissingEtFilter and MissingEtFilterUpperCut.

[10.07.17, 10:54:34 AM] Judita Mamuzic: but I genereated with 100

[10.07.17, 10:55:05 AM] Yu-Ting: So no event will pass this combined filter because of the MissingEtFilterUpperCut

[10.07.17, 10:55:17 AM] Judita Mamuzic: nono, the upper cut is very high just a artificial one unless I messed it up :-)

[10.07.17, 10:55:59 AM] Yu-Ting: Yes, but the JO treats it as event MET > upper cut not < upper cut

[10.07.17, 10:57:48 AM] Judita Mamuzic: ups, that might be my problem I will check it now I have to send one email

[10.07.17, 10:57:59 AM] Yu-Ting:

24604 generate 00:41:00 MultiElecMuTauF...   INFO Events passed = 83669  Events failed = 84499
24605 generate 00:41:00 MissingEtFilter      INFO Events passed = 27408  Events failed = 56261
24606 generate 00:41:00 MissingEtFilter...   INFO Events passed = 0  Events failed = 27408
24607 generate 00:41:00 Py:EvgenFilterSeq    INFO Filter Expression = MultiElecMuTauFilter and MissingEtFilter and MissingEtFilterUpperCut
24608 generate 00:41:00 Py:EvgenFilterSeq    INFO Filter Efficiency = 0.000000 [0 / 168168]
24609 generate 00:41:00 Py:EvgenFilterSeq    INFO Weighted Filter Efficiency = 0.000000 [0.000000 / 220899.946802]
24610 generate 00:41:00 MetaData: GenFiltEff = 0.000000e+00

[10.07.17, 10:58:02 AM] Judita Mamuzic: and then I go back to JO

[10.07.17, 10:58:27 AM] Yu-Ting: See the line 24606 which is the MissingEtFilterCpperCut

[10.07.17, 10:58:34 AM] Judita Mamuzic: huh, thanks, I will take a look now yes, makes sense what you say

[10.07.17, 10:58:43 AM] Yu-Ting: no event pass and all events fail The only possibility is that the JO treats met > upper cut main.pdf Since you have the same problem, I think I will add this to my slides and the other people may give us feedback.

[10.07.17, 11:04:38 AM] Judita Mamuzic: lets chat can I call?

[10.07.17, 11:05:02 AM] Yu-Ting: ok Call 13 minutes 59 seconds

[10.07.17, 11:19:59 AM] Judita Mamuzic: P.S. You have put a question why N2->qqN1 has low eff? its qq we have the 2L filter :-D and I think qq includes also the b that is why its not 0

[10.07.17, 11:21:28 AM] Yu-Ting: I think after consider the filter efficiency then the N2->qqN1 will not contribute more than 1% because the efficiency is very small

[10.07.17, 11:21:53 AM] Judita Mamuzic: so you say it is too high? ah, indeed so it should not contribute much also, perhaps you could make one plot

[10.07.17, 11:22:42 AM] Yu-Ting: N2->qqN1 3.6% is heigher then the N2->WC1 1.8%

[10.07.17, 11:22:52 AM] Judita Mamuzic: stacked, where we see each production contribution

[10.07.17, 11:22:58 AM] Yu-Ting: That is why I put a question mark for that 3.6% but if I add the filter efficiency than it goes away

[10.07.17, 11:23:37 AM] Judita Mamuzic: yes, ok, lets see the numbers after filt eff also, xsec would be good ah, perhaps we need xsec ah, no its the same process also, for the stacked plot usually you put the lowest contribution down and on top the highest so perhaps first see the yields, and then re-arrange the contributions because its a log scale

[10.07.17, 11:27:10 AM] Yu-Ting: I don't know how to control the order of the stack plot...

[10.07.17, 11:27:29 AM] Judita Mamuzic: Its just the order in which you add them it could be it is reverse of the order you add

[10.07.17, 12:27:53 PM] Yu-Ting: main.pdf I will go to lunch now and modify the slides after lunch if it is needed.

[10.07.17, 12:32:47 PM] Judita Mamuzic: ok I would drop the slide with JO bugs this is trivial I need to fix it, there is no reason to get them upset this will be fixed otherwise you could emphesize in the conclusion that N2->llN1 is 99.44% of events (when we use the 2L filter) so it is safe to assume this decay chain the plot looks very nice, good luck today

[10.07.17, 1:20:36 PM] Yu-Ting: done

[10.07.17, 1:21:00 PM] Judita Mamuzic: great, thanks :-) Ah, on slide 4 add that we are using the 2L3 filter

[10.07.17, 1:25:04 PM] Yu-Ting: slide is using my own JO which use 2L2 filter

[10.07.17, 1:25:39 PM] Judita Mamuzic: ok, write it somewhere, all this stands if you use the 2L filter :-)

[10.07.17, 1:26:09 PM] Yu-Ting: I mentioned I use different JO at the top of slide 4

[10.07.17, 1:26:28 PM] Judita Mamuzic: yes, but you need to say if you are using the filter, and then which just to be precise

[10.07.17, 3:22:29 PM] Yu-Ting: I think before you get the working JO, I cannot do anything, right?

[10.07.17, 3:23:08 PM] Judita Mamuzic: oh, right, perhaps you could run the 4 productions, if the test job is ok I think this evening I could have the result of the MET fitler bug fix and if it is ok, you could run 4 productions?

[10.07.17, 3:23:50 PM] Yu-Ting: okay, I can run the 4 productions.

[10.07.17, 3:24:02 PM] Judita Mamuzic: great, so I guess this afternoon there is nothing ]you could start adjusting the code for weighting to lumi next we will need to have the plots of all distributions also, you cold check if the cut flow is fully correct and up to date, we will need the sensitivity plots next

[10.07.17, 3:25:27 PM] Yu-Ting: yes, I know the SR definitions has been changed slightly This message has been removed.

[10.07.17, 3:25:44 PM] Judita Mamuzic: :-) the presentation was very nice, Mike was super happy :-) perhaps they never made this check so nicely :-) also, as we don’t want to wait for the evgen production, we could do the sensitivity plots at the truth level. Since we have only 8 points, perhaps we could generate all 4 productions for all of them. I am not sure how you stand with computing resources, but perhaps we could generate 8 x 4 jobs? if that is feasable, perhaps you could start on preparing some script for the job submission

[10.07.17, 3:32:07 PM] Yu-Ting: Why 8?

[10.07.17, 3:32:12 PM] Judita Mamuzic: all 8 NUHM2 points

[10.07.17, 3:32:18 PM] Yu-Ting: oh~ I see

[10.07.17, 3:32:30 PM] Judita Mamuzic: then we could have the expected exclusion line if the JO is ready and all is OK, no need to wait but lets see so the 32 jobs are not a priority, but if we have time left why wait for all the production, you can be done faster…but it indeed depends on the JO

[10.07.17, 3:34:30 PM] Yu-Ting: by the way, next Monday I am now available

[10.07.17, 3:34:44 PM] Judita Mamuzic: ok, I am also away Friday-Sunday

[10.07.17, 3:35:02 PM] Yu-Ting: i can give you my results if I have them and you can present the new results

[10.07.17, 3:35:22 PM] Judita Mamuzic: ok, that would be the validation, and the JO status ok

[10.07.17, 3:35:40 PM] Yu-Ting: I have afternoon shift form this Thursday to next Thursday.

[10.07.17, 3:35:58 PM] Judita Mamuzic: ok, thanks for letting me know

[10.07.17, 3:36:17 PM] Yu-Ting: They don't allowed to attend vidyo meeting during the shift

[10.07.17, 3:36:22 PM] Judita Mamuzic: what??? that is crazy?? that is some new rule??

[10.07.17, 3:37:01 PM] Yu-Ting: I don't know. They mentioned it when I was in the shift training

[10.07.17, 3:37:24 PM] Judita Mamuzic: wow, that is new :-) but ok, np which shift?

[10.07.17, 3:39:06 PM] Yu-Ting: id

[10.07.17, 3:39:26 PM] Judita Mamuzic: ok, cool, they don’t want anybody to burn the detector :-) Yes! Py:Athena INFO leaving with code 0: "successful run" Hi Yu-Ting, I made a copy of the fixed JO to /afs/ific.uv.es/user/m/mamuzic/public/ForYuTing/JO_METfix indeed it was the METFilter problem, I was missing some includes, but I knew about this problem from before, from some other JO, there were a few discussions I added the log there too and the root file for N2C1p only I will submit all productions over night hope it goes well :-) thanks for pointing to the MET filter bug!

[ytatus94]

• 20170712

[12.07.17, 10:38:48 AM] Judita Mamuzic: Hi Yu-Ting, the 4 jobs have completed for the NUHM2 m1/2=600 point I made a copy to /afs/ific.uv.es/user/m/mamuzic/public/ForYuTing/JO_VALIDATION there are the root files and the logs The efficiency is:

log.generate.C1C1:19:26:29 Py:EvgenFilterSeq    INFO Filter Efficiency = 0.281049 [10000 / 35581]
log.generate.N2C1m:17:02:33 Py:EvgenFilterSeq    INFO Filter Efficiency = 0.295692 [10000 / 33819]
log.generate.N2C1p:17:35:26 Py:EvgenFilterSeq    INFO Filter Efficiency = 0.306645 [10000 / 32611]
log.generate.N2N1:03:02:58 Py:EvgenFilterSeq    INFO Filter Efficiency = 0.305502 [10000 / 32733]

(I think we use just the filter efficiency, not the weighted) I would ask you to run the TRUTH3 and then your code, and to plot all distributions, weighted also with fiter eff. and BR with these you could already calculate the significance for this point, and if time permits, all points later on once we’ve run them I will start to prepare the request for NUHM2, but it would be good if you could post the validation plots to JIRA, and then I could put a link to them. Lets hope all goes well :-)

[ytatus94]

• 20170713

[13.07.17, 10:58:01 AM] Judita Mamuzic: Hi Yu-Ting, could we meet shortly? on Skype

[13.07.17, 11:51:48 AM] Yu-Ting: Hi Judita, yes

[13.07.17, 11:59:30 AM] Judita Mamuzic: Huh, a bit later, I have a meeting soon I guess we meet at the Oklahoma meeting will you join, or your shift starts?

[13.07.17, 12:12:23 PM] Yu-Ting: I will be in the shift

[13.07.17, 12:13:09 PM] Judita Mamuzic: I think I will connect any way, because I skipped last time… for my other analysis we have a FAR soon I was wondering if you are managing the TRUTH3 + plots for the m12 = 600 and if you are submitting the jobs for the rest

[13.07.17, 12:14:07 PM] Yu-Ting: i can try to connect. Although they said it is not allowed, i don't think shifter follows this rule.

[13.07.17, 12:14:08 PM] Judita Mamuzic: of the points haha

[13.07.17, 12:14:46 PM] Yu-Ting: plots were ready yesterday but they are normalized to one not lumi I spent a lot of time to find the needed information of Higgsino in order to normalize to lumi yesterday. But I notice one thing that it is impossible for me to use truth to normalize nuhm2 to lumi because we don't have BR for p p > n2n1, n2c1+-, and c1c1. we only have BR fro n2 > l+l-n1, c1> f f n1

[13.07.17, 12:16:56 PM] Judita Mamuzic: you take the xsec for those in the xsec file, and look for that porcess we looked into that

[13.07.17, 12:20:48 PM] Yu-Ting: If I use cross section for nuhm2, then shall I use index for higgsino? Higgsino has BR for p p > n2n1, n2c1+-, and c1c1. If you want to see the plots normalized to 1, they are put in /afs/cern.ch/user/y/yushen/public/ForJudita/20170712

[13.07.17, 12:25:12 PM] Judita Mamuzic: nono, for NUHM2 read the xsec for index that is for given process production, and BR from the les houches of the correct point perfect! thanks I have to run now

[13.07.17, 12:27:48 PM] Yu-Ting: don't understand

[13.07.17, 12:28:53 PM] Judita Mamuzic: to read the xsec: open the xsec file for the NUHM2 weak production, on the SignalUncertainty page look up the index of the N2N1, C1C1 and N2C1p, N2C1m processes, and read out the xsec to read out the BR for N2 and C1, take the les houches file, for this point it is the m12 = 600, for the other points the corresponding ones, and read out the N2->llN1 and C1->ffN1 (sum up the correct values, like in the first slide of your talk last week)

[13.07.17, 1:15:58 PM] Yu-Ting: no, what I mean is that I don't understand why nuhm2 use cross section for n2n1 c1c1 n2c1pm but higgsino use BR

[13.07.17, 1:16:25 PM] Judita Mamuzic: for them the xsec is already in the ntuple via SUSY tools

[13.07.17, 1:16:34 PM] Yu-Ting: for example, we should use BR(pp>n2c1p)xBR(n2>l+l-n1)xBR(c1>f f n1) if we use Xsec(pp>n2c1p)xBR(n2>l+l-n1)xBR(c1>f f n1) for nuhm2, then the results will be incorrect

[13.07.17, 1:17:06 PM] Judita Mamuzic: xsec(pp>n2c1p)xBR(n2>l+l-n1)xBR(c1>f f n1) Hi Yu-Ting, perhaps you could send to the JIRA plots you sent me, they look didn’t check all of them, but the ones I did, look very nice

[13.07.17, 2:14:18 PM] Yu-Ting: ok, i will upload them to jira

[13.07.17, 2:19:46 PM] Judita Mamuzic: Great! Thanks! Then I can continue with the request :-) P.S. BR for C1->ffN1 is probably 1

[13.07.17, 2:31:12 PM] Yu-Ting: no it is 9.42923324e-02 By the way, I made some alignments for the JO. could you use this when we submit the official request? /afs/cern.ch/user/y/yushen/public/ForJudita/MadGraphControl_NUHM2_weak_MadSpin.py Also change doMETFilter the evt_multiplier*=50.0, otherwise the C1C1 doesn't have enough events if we request 50000 events. But Pat said we will decide how many events to produce after seeing my results.

[13.07.17, 2:37:56 PM] Judita Mamuzic: I will take a look

[13.07.17, 2:38:14 PM] Yu-Ting: I mean the number of events in SR after all the selection.

[13.07.17, 2:38:23 PM] Judita Mamuzic: I would like to submit today, and if I change something, have to re-run I had a multiplier 100 for leptons

[13.07.17, 2:39:13 PM] Yu-Ting: for do2LFilter, it is 100 but for MET it was 30 I think. And I test to produce 50000 events. 30 is not enough. so I change to 50

[13.07.17, 2:40:00 PM] Judita Mamuzic: nono it was 100 for 2L and 50 for MET my 4 jobs finished

[13.07.17, 2:40:57 PM] Yu-Ting: I checked the JO you gave me it was 100 for 2L and 10 for MET anyway, i ran 50000 events and it needs 100 for 2L and 50 for MET 100 for 2L and 10 for MET only work for generating 10k events.

[13.07.17, 2:43:15 PM] Judita Mamuzic: yes, it is 10 for MET for met but they were successful

[13.07.17, 2:43:33 PM] Yu-Ting: how many events you generate

[13.07.17, 2:43:37 PM] Judita Mamuzic: 10k

[13.07.17, 2:44:01 PM] Yu-Ting: yes, that is why I said we need 50 for MET in order to generate 50k events

[13.07.17, 2:44:24 PM] Judita Mamuzic: I don't understand

[13.07.17, 2:44:26 PM] Yu-Ting: I tried to use 10 for MET to generate 50k and it is not enough

[13.07.17, 2:44:33 PM] Judita Mamuzic: they were successful but how, it was successful for me 50 events? or 50k?

[13.07.17, 2:44:56 PM] Yu-Ting: if you use 10 for MET then you will get the same error as I posted on JIRA

[13.07.17, 2:45:07 PM] Judita Mamuzic: but I tell you I didn't

[13.07.17, 2:45:08 PM] Yu-Ting: 50k=50000 events

[13.07.17, 2:45:17 PM] Judita Mamuzic: because I increased it at 2L filter

[13.07.17, 2:45:34 PM] Yu-Ting: you only produce 10k, so you don't have this problem

[13.07.17, 2:45:50 PM] Judita Mamuzic: multiplier you need to have enough events

[13.07.17, 2:45:51 PM] Yu-Ting: I tried 100 for 2L & 10 for MET to generate 10k and it works but 100 for 2L & 10 for MET to generate 50k and it doesn't work 100 for 2L & 50 for MET to generate 50k and it works

[13.07.17, 2:46:28 PM] Judita Mamuzic: this makes no sense to me, when you generate you get more events, and they are generated

[13.07.17, 2:47:10 PM] Yu-Ting: I don't know. It is what happens on me

[13.07.17, 2:47:21 PM] Judita Mamuzic: ok, we can increase MET to 50 are you sure it was not the connection problem?

[13.07.17, 2:47:31 PM] Yu-Ting: so I have to use 50 for MET in order to produce 50k events.

[13.07.17, 2:47:37 PM] Judita Mamuzic: for me the same night 4 jobs failed

[13.07.17, 2:47:48 PM] Yu-Ting: connect problem has different error messages

[13.07.17, 2:47:52 PM] Judita Mamuzic: and next day the same jobs completed without problem ok, we loose nothing if we increase the MET multiplier to 50

[13.07.17, 2:49:06 PM] Yu-Ting: I have to leave now

[13.07.17, 2:49:14 PM] Judita Mamuzic: ok, have fun I will submit today then oh, I need your validation plots :-D so first you need to post the plots, then I submit the request but I will make everything ready in the next 2 hours, and we are good to go I will connect to Oklahoma toady, I will mention all this, and then I think if we are all happy I will submit thanks for making these checks btw, are you submitting all 8 NUHM2 points?

[ytatus94]

• 20170717

[17.07.17, 3:11:21 PM] Judita Mamuzic: Hi Yu-Ting, it went smooth no questions at all :-)

[17.07.17, 3:17:10 PM] Yu-Ting: Hi Judita, thank you for letting me know. Hope the MC production goes well too.

[17.07.17, 3:17:35 PM] Judita Mamuzic: yes, I hope it will go smooth :-)

[17.07.17, 3:58:46 PM] Yu-Ting: Hi Judita, I read your slides and I notice this: We had to increase the event multiplier when MET > 50 GeV cut is used to 50 The samples I produced for the validation plots are using multiplier = 10 for MET. I don't think this affect but just let you know and I don't understand your calculation in "EXPECTED YIELDS" page

[17.07.17, 4:23:43 PM] Judita Mamuzic: Hi Yu-Ting, oh, I miss understood your evt_multiplier from your JIRA text, for me the 10 for MET was failing and now that is confusing concerning the formula in the expected yields, I assume 10% acceptance, efficiency after the cut flow for all productions, and the yileds is how many events we expect is there something else?

[17.07.17, 4:30:07 PM] Yu-Ting: can you explain acceptance to me? I don't understand it. I think using 50 for MET evt_multiplier is fine. And I don't know which JIRA text you mentioned. But I believe that is using old JO which has MET50 removes 2L3 problem. I use the new JO and evt_multiplier=10 works for generating 10k events. (But it doesn't work for generating 50k events)

[17.07.17, 4:33:22 PM] Judita Mamuzic: Nev = sigma x L x eff x A and here we have to split this into different productions

[17.07.17, 4:34:10 PM] Yu-Ting: I tried to produce 10k and 50k events. MET evt_multiplier=10 only work for producing 10k events.

[17.07.17, 4:34:16 PM] Judita Mamuzic: so here the sigma x eff is done per production, and then summed up ok, I already requested everything for 50 so that should be ok efficiency will change is the formula more clear now ah, also I am multiplying with BR for N2

[17.07.17, 4:35:40 PM] Yu-Ting: you provide a formula including acceptance but I still don't know what is it.

[17.07.17, 4:36:00 PM] Judita Mamuzic: so that is to account for the number of events after the cut flow the ration before and after the cut flow and I make a very simple assumption in reality if will differ for each point, because the kinematics will change

[17.07.17, 4:39:05 PM] Yu-Ting: Why didn't you consider BR(c1->ffn1) if the decay contains c1?

[17.07.17, 4:39:15 PM] Judita Mamuzic: it is 100% This message has been removed.**

[17.07.17, 4:43:37 PM] Yu-Ting: They are not 100% https://svnweb.cern.ch/trac/atlasoff/browser/Generators/MC15JobOptions/trunk/share/DSID370xxx/susy.370621.NUHM2_m12_600.slha#L531

**[17.07.17, 4:44:40 PM] Judita Mamuzic:

DECAY   1000024  1.21064772E-06   # W1SS+ decays
532 #          BR          NDA       ID1       ID2       ID3       ID4
533       3.33332181E-01    3      1000022         2        -1             # W1SS+  -->  Z1SS   UP     DB         
534       3.33332181E-01    3      1000022         4        -3             # W1SS+  -->  Z1SS   CH     SB         
535       1.11111194E-01    3      1000022       -11        12             # W1SS+  -->  Z1SS   E+     NUE         
536       1.11111194E-01    3      1000022       -13        14             # W1SS+  -->  Z1SS   MU+    NUM         
537       1.11113295E-01    3      1000022       -15        16             # W1SS+  -->  Z1SS   TAU+   NUT

[17.07.17, 4:45:08 PM] Yu-Ting: yes

[17.07.17, 4:45:50 PM] Judita Mamuzic: sum them up

[17.07.17, 4:46:07 PM] Yu-Ting: sorry, yes, it is 100%. I calculated wrong values

[17.07.17, 4:46:40 PM] Judita Mamuzic: we are generating C1->ffN1, and then apply filter, so filter efficiency accounts for qq not passing the criteria

[17.07.17, 4:47:40 PM] Yu-Ting: so filter efficiency accounts for qq not passing the criteria <--- What does this mean? do you mean we have 2L filter and c1->ffn1 doesn't pass this?

[17.07.17, 4:48:28 PM] Judita Mamuzic: so in principle, there are many events in e.g. C1C1 production that should not pass the 2L filter, but filter efficiency accounts for that, but we have generated c1->ffN1

[17.07.17, 4:48:45 PM] Yu-Ting: okay

[17.07.17, 4:48:54 PM] Judita Mamuzic: ok let me know if you need more questions, be free to make an envelope calculation yourself double check is always welcome :-)

[17.07.17, 4:52:37 PM] Yu-Ting: but why the values are much smaller than my number of events in the SR? my values are in the mail

[17.07.17, 4:53:02 PM] Judita Mamuzic: I assumed 1% efficiency did you do the normalization?

[17.07.17, 4:53:35 PM] Yu-Ting: oh, no. I simply count the number of events in the SR This may be the reason

[17.07.17, 4:54:09 PM] Judita Mamuzic: from those number in the table one can get the Acceptance for each production as Npass/Ntot yes, for sure :-D

[ytatus94]

• 20170720

[20.07.17, 4:18:20 PM] Judita Mamuzic: Hi Yu-Ting, do we have a meeting today? btw, did I see you today in B40? I might actually take that offer from Chris to use the OU computing cluster…

[20.07.17, 4:27:25 PM] Yu-Ting: I think we have meeting today But I will not attend the meeting because I am taking control room shift now. And I didn't go to b40 for a long time. The guy you saw is not me.

[20.07.17, 4:29:32 PM] Judita Mamuzic: Ah, ok, I passed by somebody, and for a sec I thought it was you, I guess it wasn’t

[20.07.17, 6:21:09 PM] Yu-Ting: Hi Judita, I think the meeting is finished, right? Anything important?

[20.07.17, 6:21:51 PM] Judita Mamuzic: Ah, nobody was there, so Chris just replied after a while. I asked for resources, but people are on a holiday…so I have to wait…

[20.07.17, 6:23:58 PM] Yu-Ting: ok, I don't know today is a holiday in US sorry

[20.07.17, 6:27:52 PM] Judita Mamuzic: Ah, ok, happy holiday!

[ytatus94]

• 20170721

[21.07.17, 8:07:31 AM] Judita Mamuzic: Hi Yu-Ting, this will be an interesting workshop today at 9AM https://indico.cern.ch/event/647568/ Hi Yu-Ting, the Higgsino analysis got an OK for unblinding, and they want to update the INT note before Monday. It would be good to put all the studies you did for the choice of production and the updated plots with filter efficiency and BR for N2 for 4 NUHM2 productions.

[21.07.17, 3:59:00 PM] Yu-Ting: sorry I was in a meeting I saw Mike's mail

[21.07.17, 3:59:12 PM] Judita Mamuzic: Sure, yes, good news

[21.07.17, 4:00:16 PM] Yu-Ting: by the way, how do we know the nuhm2 mc generation start?

[21.07.17, 4:01:01 PM] Judita Mamuzic: you can subscribe to the JIRA https://its.cern.ch/jira/browse/ATLMCPROD-4894, they will tell us when they have added the production, and then give a panda link

[21.07.17, 4:01:18 PM] Yu-Ting: I saw the JIRA ticket but it is under discussion and not start yet

[21.07.17, 4:01:19 PM] Judita Mamuzic: at the moment I am iterating with them some smaller requests yes, there are a few steps

[21.07.17, 4:01:56 PM] Yu-Ting: ok, they will inform us when the production starts

[21.07.17, 4:01:58 PM] Judita Mamuzic: first I do what he asked, then they upload to svn, then I test again, and then it goes for production yes, but don’t wait for that you can be finished by the time the points are ready so we need to do the analysis at the TRUTH level so I got the OU rights :-D

[21.07.17, 4:03:00 PM] Yu-Ting: The local versions are running I submitted them

[21.07.17, 4:03:08 PM] Judita Mamuzic: Ah, perfect! all 28?

[21.07.17, 4:03:16 PM] Yu-Ting: yes

[21.07.17, 4:03:22 PM] Judita Mamuzic: GREAT!

[21.07.17, 4:03:31 PM] Yu-Ting: but use the JO you provided to JIRA 2 days ago

[21.07.17, 4:03:38 PM] Judita Mamuzic: yes, that's the one thanks! in physics result will not change

[21.07.17, 4:04:07 PM] Yu-Ting: but Calum asked you to modify something in the JIRA ticket

[21.07.17, 4:04:14 PM] Judita Mamuzic: I am doing that I was at the workshop this morning and now at the EW meeting so now I start

[21.07.17, 4:04:43 PM] Yu-Ting: I think the modification will not affect the results a lot. so I use the current JO to run.

[21.07.17, 4:05:07 PM] Judita Mamuzic: no, it is all cosmetics, and I requested the wrong release, how stupid of me wrong copy paste... but I will explain all that

[21.07.17, 4:05:33 PM] Yu-Ting: oh~ for the part to produce run card it would be better to remove the spaces before the MadGraph command

[21.07.17, 4:06:08 PM] Judita Mamuzic: I did that you asked me, I did you were right

[21.07.17, 4:06:25 PM] Yu-Ting: The spaces don't affect the results but just looks ugly the version I downloaded from JIRA still have spaces before the MadGraph commands

[21.07.17, 4:07:10 PM] Judita Mamuzic: yes, I imagined they would appear different, then I had no time to fix back, takes many hours until next jo, but in one iteration, after you suggested, I changed it which Jira?

[21.07.17, 4:07:44 PM] Yu-Ting: the one you uploaded 2 days ago

[21.07.17, 4:07:45 PM] Judita Mamuzic: the 4894?

[21.07.17, 4:07:51 PM] Yu-Ting: yes

[21.07.17, 4:08:53 PM] Judita Mamuzic: ah, true, I removed MAdSpin, not MAdGRaph

[21.07.17, 4:09:00 PM] Yu-Ting: I mean modify to this

[21.07.17, 4:09:02 PM] Judita Mamuzic: ok, no problem yes yes, I know what you meant**

[21.07.17, 4:09:20 PM] Yu-Ting: Screen Shot 2017-07-21 at 4.09.08 PM.png

[21.07.17, 4:09:32 PM] Judita Mamuzic: yes, thanks

[21.07.17, 4:10:13 PM] Yu-Ting: anyway, with spaces are still fine. It doesn't affect the results.

[21.07.17, 4:10:34 PM] Judita Mamuzic: yes, that is why I didn’t bother so much, but it is nicer when its tidy, I agree ok, let me know if you need help or some consulting should I help with the structure for the text? or you will manage, and I will comment after?

[21.07.17, 4:11:45 PM] Yu-Ting: I can work on the note

[21.07.17, 4:12:16 PM] Judita Mamuzic: I will be doing a lot of shopping tomorrow, need to get many presents before I go home :-) but in the evening I will be back ok, great :-)

[21.07.17, 4:12:29 PM] Yu-Ting: I will let you know after I finish modify the note so you can work on what you are working on.

[21.07.17, 4:12:34 PM] Judita Mamuzic: I am just joking, let me know if you need help merci :-)

[21.07.17, 4:12:43 PM] Yu-Ting: are you leaving soon?

[21.07.17, 4:12:48 PM] Judita Mamuzic: yes, 28th we should meet next week discuss a bit the plans after I go

[21.07.17, 4:13:14 PM] Yu-Ting: okay

[21.07.17, 4:13:20 PM] Judita Mamuzic: but I think you kind of got the stream, and it will not be a problem and I am always reachable by email :-)

[21.07.17, 4:13:50 PM] Yu-Ting: thank you

[21.07.17, 4:13:58 PM] Judita Mamuzic: np :-)

[ytatus94]

• 20170725

[25.07.17, 4:36:53 PM] Yu-Ting: Hi Judita, do you know does the MC production team accept to run the same job twice and using different random seeds? we need 20k events. So we can produce 10k events twice using different random seeds and combine the results.

[25.07.17, 4:38:40 PM] Judita Mamuzic: The random sedd is always changed ah, good point, but not sure if that is possible in the same job

[25.07.17, 4:39:24 PM] Yu-Ting: Because the original JO use 100*20 multiplier and it works for producing 10k events. smaller multiplier means smaller number of events produced by MadGraph and save more time. and requiring 10k events work for old multiplier setting.

[25.07.17, 4:41:24 PM] Judita Mamuzic: lets test all for 20k that is what we requested

[25.07.17, 4:41:46 PM] Yu-Ting: We just need to request to produce 10k events twice and combine the root files, then we get 20k events

[25.07.17, 4:41:47 PM] Judita Mamuzic: I will read what you write, I am on a meeting at the moment

[25.07.17, 4:42:25 PM] Yu-Ting: ok, I want to post my question to the JIRA ticket 4894 so the MC production team can reply to me

[25.07.17, 4:42:47 PM] Judita Mamuzic: wait a sec I have in my group Emma and Zach, they are experts I will ask them informally

[25.07.17, 4:45:43 PM] Yu-Ting: ok

[25.07.17, 7:04:10 PM] Judita Mamuzic: Hi yu-ting The solution is actually to request less events, i will take a look again at the estimate

[25.07.17, 7:07:13 PM] Yu-Ting: hi judita, Ximo told me we can do what I want to do. So we just need to produce 10k events twice then we can have 20k events If you finish modify the JO, then please send the new JO to me.

[25.07.17, 7:07:49 PM] Judita Mamuzic: Cool Ok

[25.07.17, 7:08:04 PM] Yu-Ting: I will try to run 10k events using 100 for 2L3 and 20 for MET50 to test all mass points. It should work but I only use old JO to run 600 with 4 decays. it is better to test all of them

[25.07.17, 7:08:35 PM] Judita Mamuzic: Ah ok Huh, im out of office Better would be the new jo Can we start them tomorow?

[25.07.17, 7:09:43 PM] Yu-Ting: https://twiki.cern.ch/twiki/bin/viewauth/AtlasProtected/SUSYMcRequestProcedure#2_5_Extensions_and_additions sure. I have a meeting tonight so I will not submit jobs even you can give me new JO now.

[25.07.17, 7:10:48 PM] Judita Mamuzic: Hihi :-) ok We meet tomorrow at 10 Got to go, ciao :-)

[25.07.17, 7:11:26 PM] Yu-Ting: The above url is about extension the dataset which is what we need.

[25.07.17, 7:11:37 PM] Judita Mamuzic: Cool Ciao

[ytatus94]

• 20170726

[26.07.17, 2:59:25 PM] Judita Mamuzic: Hi Yu-Ting on the point of generating enough events here is just L sum(xsec BR)

300 N_ev= 44942.2073145
350 N_ev= 40903.0797161
400 N_ev= 36977.7549046
500 N_ev= 32996.3111439
600 N_ev= 31062.6941249
700 N_ev= 29935.4163414
800 N_ev= 29191.8236172

this is total number of events, which corresponds to number of events without the filter the filter eff is less than 0.2 for all points, for higher points it is smaller the number of events that we consider before the filter is

N_ev_tot = N_filt / eff (because N_ev_tot = N_filt * eff)

so for requested N_filt = 10k, and eff = 0.2, the N_ev_tot = 50000 as a rule of a thumb, this would be ok-sh for higher points, m12 >= 600 but not for lower m12, where we need to request 20k

[26.07.17, 3:06:02 PM] Yu-Ting: Hi Judita what L you use? 36.1?

[26.07.17, 3:09:08 PM] Judita Mamuzic: yes times 1000

[26.07.17, 3:09:35 PM] Yu-Ting: Is the BR for N2 to l+l- N1?

[26.07.17, 3:10:13 PM] Judita Mamuzic: br = 1 if "N2" in prod: br = 0.107

[26.07.17, 3:12:37 PM] Yu-Ting: and I don't understand how to know 10k is find for m12<600 from the N_ev you calculated

[26.07.17, 3:13:51 PM] Judita Mamuzic: its reverse, it is not fine for m12 < 600

[26.07.17, 3:14:44 PM] Yu-Ting: where to find the N2 in prod br=0.107

[26.07.17, 3:15:15 PM] Judita Mamuzic: I took this from your slides for N2->llN1

[26.07.17, 3:18:41 PM] Yu-Ting: oh~ that's why I feel I saw this number before :D But I forget where But I still don't understand how to know 10k events for m12<600 are not fine All the calculated value in your list are less then 50000

[26.07.17, 3:21:52 PM] Judita Mamuzic: so we should generate more

[26.07.17, 3:21:54 PM] Yu-Ting: anyway, if you finish modify the JO, then please send it to me. I can submit jobs for multiplier=100 (2L3), 20(MET50), and 10k events to test today. if 10k generation works, then we can require extension of dataset. for example, 10k+10k=20k or we can require more.

[26.07.17, 3:24:04 PM] Judita Mamuzic:

300 N_ev= 5010.35328165
350 N_ev= 4795.66380824
400 N_ev= 4385.20406426
500 N_ev= 3965.72262222
600 N_ev= 3758.60384536
700 N_ev= 3636.55594495
800 N_ev= 3555.32455169

so this is xsec x br x LUMI x 1000 x eff

[26.07.17, 3:24:40 PM] Judita Mamuzic: this is how much we need using the filter eff of the 600 point the filt eff is lower for higher points so we need less ok, lets do as you say I will prepare and send in the next 15 min

[26.07.17, 3:27:00 PM] Yu-Ting: ok. Do you use the filter efficiency I sent in my mail?

**[26.07.17, 3:27:39 PM] Judita Mamuzic: not sure, I forgot

    effs = {}
    effs["N2C1p"] = 0.176
    effs["N2C1m"] = 0.176
    effs["N2N1"] = 0.165
    effs["C1C1"] = 0.107

[26.07.17, 3:29:56 PM] Yu-Ting: then they are not the values in my mail. Where do these number come from? If you find the mail I sent on July/24 in titled "We have to increase the evt_multiplier for NUHM2 MC production", then you can see the efficiencies.

[26.07.17, 3:31:12 PM] Judita Mamuzic: I might have used mine

[26.07.17, 3:31:22 PM] Yu-Ting: it is okay.

[26.07.17, 3:31:25 PM] Judita Mamuzic: with 2L multiplier 100 and MET 20

[26.07.17, 3:32:30 PM] Yu-Ting: the efficiencies do not change a lot using MET multiplier 20 or 50. The first mail contains the efficiencies for MET multiplier 20 anyway, I think your values are fine.

[26.07.17, 3:33:03 PM] Judita Mamuzic: great

[26.07.17, 3:33:52 PM] Yu-Ting: I just need to test 10k events using the new JO and make sure it works well for all samples. Then we can submit the requests again. I personally prefer to request the same number of events for all samples. So I don't need to remember the reason why we have to request different number of events for different mass points.

[26.07.17, 3:35:55 PM] Judita Mamuzic: but they fail for 20k

[26.07.17, 3:35:59 PM] Yu-Ting: Because I might forget the reason after several weeks later... :D I said 20k events can be split to 10k + 10k

[26.07.17, 3:36:18 PM] Judita Mamuzic: what you need in the analysis is the efficiency, and this will be in the xsec file

[26.07.17, 3:36:40 PM] Yu-Ting: so if the new JO producing 10k are fine then it is okay If you worry about the efficiency, then we can find the efficiency from the log file at AMI. It doesn't matter how many extension we require as long as their information can be found on AMI. We always can calculate the efficiency for all of them.

[26.07.17, 3:42:22 PM] Judita Mamuzic: we should not have 2 requests they take time better go with 1 right away ok, the JO is at /afs/cern.ch/user/m/mamuzic/public/ForYuTing/JO_NEWFILTERS there is a folder with examples of JO and renames les houches files

[26.07.17, 3:43:28 PM] Yu-Ting: Don't worry about the request until I get the 10k testing results.

[26.07.17, 3:43:45 PM] Judita Mamuzic: in the JO files one shold change include('MC15JobOptions/MadGraphControl_NUHM2_weak_MadSpin.py') to include('MadGraphControl_NUHM2_weak_MadSpin.py') to run locally one example of a command to run is in the kreci.sh so now you have to have the right DSID the command is like

Generate_tf.py --ecmEnergy=13000. --runNumber=123456 --firstEvent=1 --maxEvents=20000 --randomSeed=123 --jobConfig=MC15.123456.MGPy8EG_A14N23LO_NUHM2_m12_600_weak_N2C1p_2LMET50_MadSpin.py --outputEVNTFile=user.mamuzic.N2C1p.pool.root

but there I renames the les houches file perhaps in your script, you can add the DSIDs so now for each JO there is a different les houches file they asked me to duplicate them (how strange in my oppinion) the control JO is for 2L multiplier 100, and MET multiplier for 20, but I increased the min number of events, 5000 for C1C1 and 2000 for the rest the JO was successful for m12 = 600 (but I tested before I increased the minevents)

[26.07.17, 3:56:32 PM] Yu-Ting: #'ktdurham':runArgs.qcut , <---- shall we still comment out this line? Calum asked this in the JIRA ticket 4894

[26.07.17, 3:56:57 PM] Judita Mamuzic: Yes, it crashes if I uncomment

[26.07.17, 3:57:23 PM] Yu-Ting: I see.

[26.07.17, 3:57:29 PM] Judita Mamuzic: it is set in runArgs.qcut = 15. # low matching scale, following DM group recommendations

if njets==0: runArgs.qcut = 0

[26.07.17, 3:58:03 PM] Yu-Ting: ok, I check your files, I think they are fine. I will prepare everything to run the MC production.

[26.07.17, 3:58:22 PM] Judita Mamuzic: yes, you will need to make a bit of modifications for the job submission

[26.07.17, 3:58:28 PM] Yu-Ting: And Ximo told me that

[26.07.17, 3:58:42 PM] Judita Mamuzic: now each has a different dsid, and the correct les houches is called from DSID yes?

[26.07.17, 4:01:09 PM] Yu-Ting: it doesn't matter the value of --maxEevents we set. It is only for the local test. If our JO works for 10k events, we still can request 20k in the request using the working JO. The MC production team will manage that. And the final efficiency on AMI will be the correct one.

[26.07.17, 4:01:20 PM] Judita Mamuzic: aaaah amazing! then you can run with -1 this sets it to 5000 it will be faster just do that woooow

[26.07.17, 4:01:48 PM] Yu-Ting: -1?

[26.07.17, 4:01:48 PM] Judita Mamuzic: yeei yes, set it to --maxEvents=-1 this runs 5000 then the JO will be faster

[26.07.17, 4:02:38 PM] Yu-Ting: I see.

[26.07.17, 4:02:42 PM] Judita Mamuzic: then it is not our responisbitilty ok, just make the 10k jobs lets see what we get tomorrow :-D but -1 will also be fine

[26.07.17, 4:03:41 PM] Yu-Ting: maybe this is the reason why Mike asked us trying to set maxEvents=5000 in our Generate_tf.py command.

[26.07.17, 4:04:10 PM] Judita Mamuzic: when was that?

[26.07.17, 4:04:16 PM] Yu-Ting: Because he knows if the JO works, then the MC production team will manage the rest and provide the correct number of events. yesterday, you replied him on JIRA ticket HIGGSINO-28

[26.07.17, 4:05:53 PM] Judita Mamuzic: ok, I will re-submit the test jobs for -1 (default = 5000) and if this works, all is ok for m12 = 600 you could do that too

[26.07.17, 4:06:16 PM] Yu-Ting: okay, but I will try 10k for all

[26.07.17, 4:06:21 PM] Judita Mamuzic: ok, as you prefer both is good Can you hold the JO submit for a few min I will try to reduce the C1C1 multiplier to have the same as Higgsino

[26.07.17, 4:10:13 PM] Yu-Ting: sure, I need time to build all the correct folder structure and link to the correct files. it takes ~30mins

[26.07.17, 4:10:39 PM] Judita Mamuzic: ok, thanks!

[26.07.17, 4:10:40 PM] Yu-Ting: to avoid mistakes

[26.07.17, 4:12:29 PM] Judita Mamuzic: ok, I think it is OK like this we could try to reduce the multipliers, but it didn’t work really brefore so lets keep them so JO is unchanged I put minevents for C1C1 5000 as before, they have 2000, and no limit for other prod, while I put 2000 for other prod

[26.07.17, 4:14:02 PM] Yu-Ting: anyway you can do any modification you want before I build all the folder structure to run MC production. I will let you know when I am ready to run.

[26.07.17, 4:14:19 PM] Judita Mamuzic: ok, thanks!

[26.07.17, 4:14:28 PM] Yu-Ting: and you can provide me JO if you modify it.

[26.07.17, 4:14:53 PM] Judita Mamuzic: ok, if I change it, I will let you know, for now I think it is good like the one I sent ok, I have submitted with maxevents=-1, for m12=600

[26.07.17, 4:44:57 PM] Yu-Ting: ok, I am ready to submit jobs

[26.07.17, 4:45:39 PM] Judita Mamuzic: I think for the sake of test, perhaps you can run with -1 as well because we can do the same trick locally, and run more with different seeds the 5000 will finish faster I am thinking that would be better

[26.07.17, 4:47:44 PM] Yu-Ting: ok i can change to 5000

[26.07.17, 4:47:50 PM] Judita Mamuzic: yes, thanks the priority is to have the JO submitted, and this information, that they deal with higher stats, and we just request 20k is like a perfection then all we need is to be sure the 5k jobs run/complete for all points

[ytatus94]

• 20170727

[27.07.17, 12:47:12 PM] Yu-Ting: New jobs are failed but not for the same reason as before The reason is my afs folder is full I need to clean more space for the jobs But it is very strange because I use maxEvent=-1, which means 5000. And MG will use 500010020=10^7 to run. I tried 1.6E8 to run two days ago and the disk does't full. I clean the disk before I started the new test yesterday. So I don't know the reason why this time the jobs occupy more disk space.

[27.07.17, 2:30:30 PM] Judita Mamuzic: that is strange I got to my computer just now, I will see if my were succesful give me a few min

[27.07.17, 2:31:52 PM] Yu-Ting: That is fine. I rerun all jobs on other machine now. But I might not have any results today.

[27.07.17, 2:33:09 PM] Judita Mamuzic: can you change this line?

evgenConfig.minevents = 2000
if proc == "C1C1": evgenConfig.minevents = 500

it was 5000, but put 500 3 of mine failed I will investigate more

[27.07.17, 2:34:38 PM] Yu-Ting: All failed jobs are C1C1?

[27.07.17, 2:35:08 PM] Judita Mamuzic: no, only N2N1 is still running, C1C1, N2C1p/m failed

[27.07.17, 2:35:59 PM] Yu-Ting: Then that is very strange

[27.07.17, 2:38:23 PM] Judita Mamuzic: very strange, this was the only difference

[27.07.17, 2:42:23 PM] Yu-Ting: yes. there should be something wrong in the new JO

[27.07.17, 2:45:56 PM] Judita Mamuzic: yes, need to find the problem huh ah, I know what it is in the one which was successful for m12=600, I didn’t have any minevents probably 2000 is never produced I will put 500 for both, that is 10% and that is our efficiency or a bit more... also, its the 19.2.5.14.3 release only these 2 differences but probalby the minevents OK, I am starting the m12=600 jobs with

evgenConfig.minevents = 500
if proc == "C1C1": evgenConfig.minevents = 100

and release asetup MCProd,19.2.5.21.1,here I think perhaps its worth to wait for these I think I would submit the request the moemnt this is succesful and then later we can ask for changes to the JO perhaps you can submit just a few C1C1 and high m12 jobs, I can do that as well.. but it makes no sense to do all again, if they will fail again... so sorry about multiple iterations btw, do we have a meeting today? if not, I would need to meet with some people...

[27.07.17, 4:09:07 PM] Yu-Ting: can you send me all the log.generate which you said C1C1 N2C1p/m were failed I don't know will we have meeting today but I will connect. If you want to meet some people, then just do it. I can present your status to Pat

[27.07.17, 4:10:29 PM] Judita Mamuzic: I didn’t get a reminder, and last time everybody was on a holiday, so perhaps its the same today

[27.07.17, 4:10:48 PM] Yu-Ting: yes I didn't get a reminder too

[27.07.17, 4:10:50 PM] Judita Mamuzic: I will take my phone, perhaps you could send me a chat, I will be a few minlate? perhaps we ask Chris?

[27.07.17, 4:11:25 PM] Yu-Ting: okay

[27.07.17, 4:12:31 PM] Judita Mamuzic: will you send a email to Chris? I can send too, I am in the office a few more min

[27.07.17, 4:13:22 PM] Yu-Ting: ok

[27.07.17, 4:13:28 PM] Judita Mamuzic: merci :-) oh sorry I deleted the last one I deleted the last round with the logs

[27.07.17, 4:17:52 PM] Yu-Ting: so you don't have any log files

[27.07.17, 4:17:52 PM] Judita Mamuzic: you wanted to compare I dont have the last round I have the one which were good, with 10k events and older release

[27.07.17, 4:18:16 PM] Yu-Ting: ok

[27.07.17, 4:18:51 PM] Judita Mamuzic: I have one more round, but I don’t remember what was there, ah, yes, very large multiplier, they still didn’t finish…

[27.07.17, 4:18:58 PM] Yu-Ting: and shall we use 19.2.5.21.1?

[27.07.17, 4:19:06 PM] Judita Mamuzic: I am trying with this one

[27.07.17, 4:19:09 PM] Yu-Ting: I prefer to use 19.2.5.14.3 because all of my previous tests are using this version.

[27.07.17, 4:19:26 PM] Judita Mamuzic: yes, so that one worked for 600 yes, perhaps we need to stick to that

[27.07.17, 4:19:59 PM] Yu-Ting: only the jobs running new JO you gave me yesterday are using the 19.2.5.21.1

[27.07.17, 4:20:04 PM] Judita Mamuzic: so perhaps resubmit all with 5000, new JO and old release I am testing the new release, and if that is OK, probalby it is ok to switch and we can request the new the problem is really the multipliers and events

[27.07.17, 4:20:37 PM] Yu-Ting: I don't thinek minevents affects the results. you think they crashed due to the release? because in my old testing files, all samples can produce > 15000 events

[27.07.17, 4:21:44 PM] Judita Mamuzic: oh, here is the reminder

[27.07.17, 4:21:45 PM] Yu-Ting: and if Muhammad told me is correct, then the minevents only works when the number of events < we required. i.e. if we require 20k, and only produce 15k, if we set minevents=2000, then Athena will use 2000 to study.

[27.07.17, 4:22:26 PM] Judita Mamuzic: yes, it is smaller, I think with efficiency we have of 10%, it was too high to put it 2000 and 5000, for requested 5000 so now its 500 and 100 yes, but I was running 5000

[27.07.17, 4:23:01 PM] Yu-Ting: so I believe minevents is not the reason. The main reason should be the version.

[27.07.17, 4:23:10 PM] Judita Mamuzic: I see well, then we have the logs for 10k events with the old release we might dare to submit

[27.07.17, 4:23:51 PM] Yu-Ting: If I use old JO, old version, and only require 10k events, then all samples should work

[27.07.17, 4:24:26 PM] Judita Mamuzic: when you say old, it is the same as now, but no minevents statement at all

[27.07.17, 4:24:26 PM] Yu-Ting: I have some testing jobs to do this

[27.07.17, 4:24:31 PM] Judita Mamuzic: perfect! ok, then lets do that

[27.07.17, 4:24:40 PM] Yu-Ting: they are running now

[27.07.17, 4:24:47 PM] Judita Mamuzic: for that I have the logs, and we are good to go perfect ok, then we might have the result this evening, and then I can submit I can prepare all with my logs, and mention you that you did the test for all

[27.07.17, 4:25:23 PM] Yu-Ting: Chris sends the meeting reminder

[27.07.17, 4:25:46 PM] Judita Mamuzic: we got two :-)

[27.07.17, 11:06:57 PM] Yu-Ting: Hi Judita, all jobs are still runing.

[27.07.17, 11:07:14 PM] Judita Mamuzic: hi, thanks for letting me know

[27.07.17, 11:07:19 PM] Judita Mamuzic: lets see in the morning

[27.07.17, 11:07:26 PM] Yu-Ting: Only 700_N2C1m using new JO old release finish [27.07.17, 11:07:27 PM] Judita Mamuzic: if not, then in the afternoon [27.07.17, 11:07:36 PM] Judita Mamuzic: perfect [27.07.17, 11:07:47 PM] Judita Mamuzic: so the new release run longer? [27.07.17, 11:07:53 PM] Yu-Ting: But I check the log file, then I realise we cannot use evgenConfig.minevents = 2000 [27.07.17, 11:08:06 PM] Judita Mamuzic: I have set less [27.07.17, 11:08:16 PM] Yu-Ting: Although I ask to produce maxEvents=5000 [27.07.17, 11:08:18 PM] Judita Mamuzic: huh, I wrote how much [27.07.17, 11:08:36 PM] Yu-Ting: But it seems Athena only use 2000 even though 5000 is possible [27.07.17, 11:09:03 PM] Yu-Ting: And today Muhammad also told me the same thing. [27.07.17, 11:09:03 PM] Judita Mamuzic: I think it will use 2000 if 2000 is created [27.07.17, 11:09:10 PM] Judita Mamuzic: until 2000 is created [27.07.17, 11:10:03 PM] Yu-Ting: Original Athena follows this: if MG produces >5000, then use 5000, otherwise use 2000 [27.07.17, 11:10:20 PM] Judita Mamuzic: mine are alos still running [27.07.17, 11:10:57 PM] Yu-Ting: But based on my 700 N2C1m results and what Muhammad told me this afternoon, the Athena use 2000 if we set evgenConfig.minevents = 2000 in JO [27.07.17, 11:10:58 PM] Judita Mamuzic: my last job is [27.07.17, 11:10:59 PM] Judita Mamuzic: evgenConfig.minevents = 500 if proc == "C1C1": evgenConfig.minevents = 100 [27.07.17, 11:11:07 PM] Yu-Ting: It ignore 5000 in the Generate_tf.py [27.07.17, 11:11:43 PM] Judita Mamuzic: but the effieciency is low, and it will not produce 5000, but instead it will not fail, but run until it reaches 2000 [27.07.17, 11:12:13 PM] Yu-Ting: Then your job will only produce 500 for N2N1 N2C1+/-, and 100 for C1C1. It will not use the maxEvents value set in the Generate_tf.py option [27.07.17, 11:12:58 PM] Judita Mamuzic: not sure, I think it will run 5000, and get what is produced, and if there is less produced, it will take 500 [27.07.17, 11:13:09 PM] Yu-Ting: no [27.07.17, 11:13:26 PM] Yu-Ting: what you said is what I original thought. [27.07.17, 11:13:50 PM] Yu-Ting: But the 700 N2C1m log file shows it only produce the values specified by evgenConfig.minevents [27.07.17, 11:14:02 PM] Yu-Ting: it doesn't use the value of maxEvents [27.07.17, 11:14:56 PM] Judita Mamuzic: but if efficiency is 10%, and you ask 5000, it should produce 500. Then if you set minevents 2000, it will run until it produces 2000. If you run 20000, it should produce 2000, and it will give you 2000 [27.07.17, 11:15:16 PM] Yu-Ting: no [27.07.17, 11:15:56 PM] Yu-Ting: the efficiency is not related to 5000 [27.07.17, 11:16:23 PM] Judita Mamuzic: so if it should produce less, it will give you more if you requested too little, but if you requested enough, it will give you how much you got [27.07.17, 11:17:01 PM] Judita Mamuzic: higgsino also uses minevents, and they requested 20000 [27.07.17, 11:17:19 PM] Judita Mamuzic: but lets see what we get in the morning [27.07.17, 11:17:34 PM] Judita Mamuzic: I had a chat with Joanna, about the HistFitter implementation [27.07.17, 11:17:46 PM] Judita Mamuzic: she said that there are instruction on the Higgsino Git [27.07.17, 11:18:42 PM] Judita Mamuzic: so for NUHM2 you would need to follow those isntructions, how to run it, how to prepare the macro [27.07.17, 11:18:52 PM] Judita Mamuzic: we need to find the Higgsino Git [27.07.17, 11:19:03 PM] Judita Mamuzic: it is probably one of the threads [27.07.17, 11:19:50 PM] Yu-Ting: I think we should remove evgenConfig.minevents [27.07.17, 11:20:07 PM] Judita Mamuzic: ok, the jobs without, for old release worked for m12 600 [27.07.17, 11:20:09 PM] Yu-Ting: Actually the old JO with old release is fine [27.07.17, 11:20:12 PM] Judita Mamuzic: so that works for sure [27.07.17, 11:20:17 PM] Judita Mamuzic: yes [27.07.17, 11:20:49 PM] Judita Mamuzic: so the point 700 you are running old release and no minevents? [27.07.17, 11:21:35 PM] Yu-Ting: okay, the old jo + old release, I use maxEvent=5000 [27.07.17, 11:22:13 PM] Yu-Ting: and 600 C1C1, N2C1+, N2C1-, and 700 N2C1- are finished. [27.07.17, 11:22:32 PM] Judita Mamuzic: great! [27.07.17, 11:22:53 PM] Judita Mamuzic: that is without minevents? [27.07.17, 11:23:12 PM] Yu-Ting: old jo doesn't have minevents [27.07.17, 11:23:33 PM] Yu-Ting: this is an example in the log file from 600 C1C1 [27.07.17, 11:23:34 PM] Yu-Ting: generate 21:50:02 Py:EvgenFilterSeq INFO Filter Efficiency = 0.106943 [5000 / 46754] [27.07.17, 11:23:39 PM] Judita Mamuzic: ok, great [27.07.17, 11:23:52 PM] Yu-Ting: efficiency [minevents/ ???] [27.07.17, 11:24:05 PM] Judita Mamuzic: 5000 is maxevents [27.07.17, 11:24:21 PM] Yu-Ting: if we set minevents, then Athena use minevents, otherwise Athena use maxEvents [27.07.17, 11:24:30 PM] Judita Mamuzic: nono [27.07.17, 11:24:31 PM] Yu-Ting: the ??? means [27.07.17, 11:25:06 PM] Yu-Ting: there are only ??? events passed / maxEvents10020 [27.07.17, 11:25:15 PM] Judita Mamuzic: athena uses minevents if less than maxevents is produced [27.07.17, 11:25:24 PM] Judita Mamuzic: well, this is the same what you said [27.07.17, 11:26:12 PM] Yu-Ting: I don't think what you said is correct. [27.07.17, 11:26:38 PM] Yu-Ting: In the new jo log file, athena use 2000 only even though it has 5000 [27.07.17, 11:27:07 PM] Judita Mamuzic: for the same point? [27.07.17, 11:27:11 PM] Judita Mamuzic: and prod? [27.07.17, 11:27:56 PM] Yu-Ting: this is the log in 700 N2C1m: 22:54:08 Py:EvgenFilterSeq INFO Filter Efficiency = 0.161434 [2000 / 12389] [27.07.17, 11:28:00 PM] Yu-Ting: it only uses 2000 [27.07.17, 11:30:30 PM] Judita Mamuzic: lets see my jobs in the morning, but I would say we concluded it is safe to use no minevents, 5000 maxevents, and old release [27.07.17, 11:30:32 PM] Judita Mamuzic: right? [27.07.17, 11:30:37 PM] Judita Mamuzic: so I could proceed with that [27.07.17, 11:32:01 PM] Yu-Ting: yes, from the log files, using old jo + old release should be fine. [27.07.17, 11:32:18 PM] Yu-Ting: I will let you know all of the results tomorrow morning [27.07.17, 11:32:26 PM] Judita Mamuzic: ok, then I will send them that in the morning [27.07.17, 11:32:32 PM] Judita Mamuzic: ok [27.07.17, 11:32:35 PM] Yu-Ting: and you can decide which jo and release you want to request [27.07.17, 11:32:51 PM] Yu-Ting: but I think you will not get the results until you arrive your home. [27.07.17, 11:33:05 PM] Judita Mamuzic: haha, it could be, it is perhaps slower [27.07.17, 11:33:08 PM] Judita Mamuzic: and that is not good [27.07.17, 11:33:17 PM] Judita Mamuzic: so then better use the old [27.07.17, 11:33:30 PM] Yu-Ting: if you want to submit the jobs before your flight, then I guess old jo + old release should work. [27.07.17, 11:33:41 PM] Judita Mamuzic: I have run that [27.07.17, 11:33:51 PM] Judita Mamuzic: with new jo, but without minevents [27.07.17, 11:34:13 PM] Yu-Ting: I couldn't give you results before your flight just because you have to arrive airport about 7am if your flight is 9am [27.07.17, 11:34:17 PM] Judita Mamuzic: you tested the same for 20k, and they failed, so we halted all, but now we know 5000 is sufficent, so we are kind of done [27.07.17, 11:34:20 PM] Judita Mamuzic: I hope [27.07.17, 11:34:35 PM] Judita Mamuzic: nono, flight is at 11:20 [27.07.17, 11:34:43 PM] Yu-Ting: that means I have to let you know my results before 7am and it is too early for me [27.07.17, 11:35:39 PM] Judita Mamuzic: ok, would say we stay as we said, maxevents 5000, no minevents, old release [27.07.17, 11:35:45 PM] Judita Mamuzic: and thats it [27.07.17, 11:36:36 PM] Yu-Ting: Screen Shot 2017-07-27 at 11.36.06 PM.png [27.07.17, 11:37:14 PM] Yu-Ting: This table was multiplier=100, 20 for 2L3 and MET50, and request 20k maxEvents [27.07.17, 11:37:54 PM] Judita Mamuzic: I think you request here 20k events [27.07.17, 11:38:21 PM] Yu-Ting: you can see all the numerator > 10k, so that is why I think if we require 10k or 5k using the old jo + old release should work [27.07.17, 11:38:51 PM] Yu-Ting: The blank cells mean the job failed but not for the reason of number of events. [27.07.17, 11:38:59 PM] Judita Mamuzic: you just said m12 700 points are finishing for the setup we agreed [27.07.17, 11:39:05 PM] Judita Mamuzic: so that point was critical [27.07.17, 11:39:16 PM] Judita Mamuzic: and for 5000 maxevents its ok now [27.07.17, 11:39:22 PM] Yu-Ting: Because nothing in the log files for those failed jobs, I don't know what happened. [27.07.17, 11:39:53 PM] Yu-Ting: but the failed jobs work again in my testing results when I use multiplier=20 for MET [27.07.17, 11:40:10 PM] Judita Mamuzic: that is amazing great news [27.07.17, 11:40:19 PM] Judita Mamuzic: ok, I have no more doubt [27.07.17, 11:40:19 PM] Yu-Ting: Screen Shot 2017-07-27 at 11.40.02 PM.png [27.07.17, 11:41:41 PM] Yu-Ting: From the second table we can see the failed jobs in the first table have 20k in the numerator. [27.07.17, 11:42:15 PM] Yu-Ting: Yes, I request maxEvent=20k when I ran these two table. [27.07.17, 11:42:51 PM] Yu-Ting: anyway, I will let you know the results as soon as possible. [27.07.17, 11:43:15 PM] Judita Mamuzic: ok, but if I mnage in the morning, I will submit, I think we have the setup [27.07.17, 11:43:16 PM] Yu-Ting: hope this is the last test and we can submit the final request by tomorrow. [27.07.17, 11:43:46 PM] Judita Mamuzic: yes, me too :-) [27.07.17, 11:44:01 PM] Judita Mamuzic: ok, good night :-D [27.07.17, 11:44:18 PM] Yu-Ting: good night

[ytatus94]

• 20170728 [28.07.17, 8:41:15 AM] Judita Mamuzic: I submitted as we said yesterday

[ytatus94]

• 20170731

[31.07.17, 12:50:46 PM] Yu-Ting: Hi Judita, my slides for today's meeting is on the indico https://indico.cern.ch/event/655272/ [31.07.17, 12:51:09 PM] Yu-Ting: I will go to lunch. If you have any suggestions, then I will read it after lunch. [31.07.17, 1:06:43 PM] Judita Mamuzic: Hi Yu-Ting, very nice results [31.07.17, 1:07:40 PM] Judita Mamuzic: for the page 4, about the acceptance, we use the number of generated events (i.e. 5000), because the efficiency already accounts for the 100 x 20 of generated events [31.07.17, 1:09:17 PM] Judita Mamuzic: on the page 5, I guess the difference comes for the acceptance, where you used the truth level results, while I just used an assumption [31.07.17, 1:10:02 PM] Judita Mamuzic: make sure not to use acceptance twice [31.07.17, 1:10:21 PM] Judita Mamuzic: ie, either you use the value after the cuts, or you use acceptance [31.07.17, 1:15:19 PM] Judita Mamuzic: Very nice! I will not be able to connect, good luck! Very nice results, looking forward to see the sensitivity plots! :-) [31.07.17, 1:31:46 PM] Judita Mamuzic: If they ask you why we need 20k events per points, the reason is that we also need a nice distribution for mll and mt2, and don’t want to suffer from large statistics [31.07.17, 1:32:26 PM] Yu-Ting: I don't understand about "not to use acceptance twice". Am I doing something wrong and calculate it twice but I don't notice it? [31.07.17, 1:32:53 PM] Yu-Ting: ie, either you use the value after the cuts, or you use acceptance <--- Also I don't understand this. [31.07.17, 1:37:43 PM] Judita Mamuzic: P.S. pg. 5, there is a typo in the formula, filter efficiency is inside the sum [31.07.17, 1:39:05 PM] Yu-Ting: I use calculate each term separately [31.07.17, 1:39:21 PM] Yu-Ting: so the efficiency inside or outside the sum doesn't matter [31.07.17, 1:40:00 PM] Judita Mamuzic: ok... [31.07.17, 1:40:04 PM] Judita Mamuzic: good luck [31.07.17, 1:40:07 PM] Judita Mamuzic: you will manage? [31.07.17, 1:40:11 PM] Yu-Ting: yes [31.07.17, 1:40:13 PM] Judita Mamuzic: I don’t need to connect? [31.07.17, 1:40:20 PM] Yu-Ting: yes [31.07.17, 1:40:30 PM] Judita Mamuzic: perfect! good luck! :-) [31.07.17, 1:40:37 PM] Yu-Ting: I don't think there is question for this talk [31.07.17, 1:41:26 PM] Judita Mamuzic: ah, you never know :-) [31.07.17, 1:42:12 PM] Judita Mamuzic: ok, ciao and let me know how it went :-) [31.07.17, 1:42:19 PM] Yu-Ting: ciao [31.07.17, 1:42:24 PM] Yu-Ting: enjoy your vacation [31.07.17, 1:42:29 PM] Judita Mamuzic: merci :-D [31.07.17, 3:21:47 PM] Yu-Ting: These are about today's presentation:

• We can use the Table 8 Higgsino optimise SR to do the study.
• We have to use the weighted filter efficiency.
• For the acceptance, instead of using 5k, we have to use the weighted events
• BR(C1->l vl n1) ~ 0.1 (Mike will give me the value they use) [31.07.17, 3:22:41 PM] Yu-Ting: Also Mike asked me can we use the truth to do the study before we get the MC production samples. [31.07.17, 3:22:57 PM] Judita Mamuzic: Yes, this is what we are doing already [31.07.17, 3:23:24 PM] Judita Mamuzic: so you are calculating the singificance at the truth level [31.07.17, 3:23:27 PM] Yu-Ting: I think he means the exclusion plots and HistFitter things [31.07.17, 3:23:40 PM] Judita Mamuzic: yes, this is the next step [31.07.17, 3:23:45 PM] Judita Mamuzic: and yes, we plan exactly that [31.07.17, 3:23:53 PM] Judita Mamuzic: because it will take 1 month for the points [31.07.17, 3:24:07 PM] Yu-Ting: anyway, I will re-calculate everything using the weighted events. [31.07.17, 3:24:58 PM] Judita Mamuzic: ok [31.07.17, 3:25:16 PM] Judita Mamuzic: I think he is probably right for the weighted filter efficiency [31.07.17, 3:25:41 PM] Judita Mamuzic: but not sure what did he mean about the weighted events, in principle, you are doing the same [31.07.17, 3:25:56 PM] Judita Mamuzic: because you did everuthing per production, so you had the right weights [31.07.17, 3:26:11 PM] Judita Mamuzic: so I expect no difference (except the filter eff) [31.07.17, 3:26:37 PM] Judita Mamuzic: but go ahead, and do as he proposed [31.07.17, 3:26:59 PM] Yu-Ting: the weighted events are not 5000 [31.07.17, 3:27:14 PM] Yu-Ting: we have filter efficiency = 5000/xxxx [31.07.17, 3:27:29 PM] Yu-Ting: and weighted filter efficiency = aaaaa/bbbbb [31.07.17, 3:27:42 PM] Yu-Ting: so we have to use aaaaa instead of using 5000 [31.07.17, 3:27:53 PM] Yu-Ting: the number can be found from the log file [31.07.17, 3:28:10 PM] Judita Mamuzic: ups, I think you are right, sorry for not spotting this

[ytatus94]

• 20170801

[01.08.17, 6:57:21 PM] Judita Mamuzic: Hi Yu-Ting are you at work, can we have a short chat? let me know when you can :-)

[01.08.17, 7:00:14 PM] Yu-Ting: yes I am working now

[01.08.17, 7:00:37 PM] Judita Mamuzic: ok, could we meet shortly? I jusst neet to get my headphones ok, can I call?

[01.08.17, 7:01:18 PM] Yu-Ting: ok Call 16 minutes 15 seconds

[ytatus94]

• 20170807

[07.08.17, 10:32:53 AM] Judita Mamuzic: Hi Yu-Ting, I am back in the office, perhaps we could meet for a chat when you have a bit of time…there is a lot of progress, but there are still a few questions concerning the event by event weighting… Hi Yu-Ting, should we meet shortly after the meeting?

[07.08.17, 2:31:14 PM] Yu-Ting: ok

[07.08.17, 2:31:22 PM] Judita Mamuzic: ok, I call Call started

[07.08.17, 2:40:18 PM] Yu-Ting: [https://gitlab.cern.ch/atlas-phys-susy-higgsino/SusySkimHiggsino/tree/master/data/samples/SUSY16_Signal}(https://gitlab.cern.ch/atlas-phys-susy-higgsino/SusySkimHiggsino/tree/master/data/samples/SUSY16_Signal) https://its.cern.ch/jira/projects/HIGGSINO/issues/HIGGSINO-42?filter=allopenissues

*[07.08.17, 2:45:20 PM] Judita Mamuzic: `rucio ls user.jreicher:SusySkimHiggsino_v1.8_SUSY16_Bkgs_tree`**

[07.08.17, 2:45:37 PM] Yu-Ting: list-dids rucio list-dids user.jreicher:SusySkimHiggsino_v*_SUSY16_Signal_tree

**[07.08.17, 3:03:46 PM] Judita Mamuzic: https://en.wikipedia.org/wiki/Log-normal_distribution https://root.cern.ch/doc/v608/classRooLognormal.html Two potential problems:

1. Grid samples are missing all productions, only N2N1 there. So the fit funcition is determined from N2N1, but applied to all prod, and this coauses problems.
2. Different dM causes a problem in mll re-weighting, possible empty bins

   if pow(m,4) -pow(m,2) < 0
   var = 1
   var = 0
   if(x>fabs(m2) - fabs(m1)) var=1;
   if((pow(m,4) -pow(m,2)) < 0) var=1;

   [07.08.17, 3:47:33 PM] Yu-Ting: Call ended 1 hour 15 minutes 59 seconds

[ytatus94]

• 20170813

[13.08.17, 10:10:09 AM] Judita Mamuzic: Hi Yu-Ting, I’m here, are you at P1? I will be online until 15h today, later on only in the evening. LEt me know when you are available.

[13.08.17, 10:26:19 AM] Yu-Ting: Yes, I am in ACR now I am login and will reply you when I am available

[13.08.17, 10:27:34 AM] Judita Mamuzic: Hi Yu-Ting, great! OK, good luck with the combined run, let me know when you have a minute …

[13.08.17, 3:25:08 PM] Yu-Ting: Hi Judita, I uploaded the results to JIRA but I need to take a nap now. You can take a look at my new results first and we can talk in the evening.

[13.08.17, 3:25:52 PM] Judita Mamuzic: Hi @Yu-Ting, I took a look already, they are quite off they need to agree perfectly we need to find a way to have perfect agreement for all distributions the difference could come from not using the tails, but also from the function implementation we need to clear out all the points and actually hope the NUHM2 will be ready fast... The further we go, I would trust the production more… but we have to have this as a place holder we need to tweak it, and find the reasons why ok, I will be here another 30 min then I go and will be away until 20h I think

[13.08.17, 10:43:09 PM] Yu-Ting: Hi Judita, I am available now

[13.08.17, 10:43:21 PM] Judita Mamuzic: haha, me too ok, can I call, it will be faster

[13.08.17, 10:43:40 PM] Yu-Ting: okay

[ytatus94]

• 20170814

[14.08.17, 10:16:06 AM] Yu-Ting: Hi Judita, I uploaded the new slides to JIRA

[14.08.17, 10:16:30 AM] Judita Mamuzic: Yes, thanks! I am looking at them already Hi Yu-Ting, looks like they didn’t provide the agenda for today, I guess they will do so later today

[14.08.17, 10:42:14 AM] Yu-Ting:

Dear All, Our next Higgsino LSP meeting is scheduled for Monday, August 14th, at 14h00 CERN time. The agenda will grow here: https://indico.cern.ch/event/655274/ Let us know if you'd like to present an update or raise a topic for discussion. Thanks! Mike and Joana PS: A reminder that our SUSY group approval meeting will be held this Friday, August 11th, at 16h00 CERN time. The agenda can be seen here: https://indico.cern.ch/event/658796/ Sheena has kindly agreed to give the talk and she already circulated a draft of her slides for comments (https://its.cern.ch/jira/browse/HIGGSINO-47) Joana sent on Aug/10

[14.08.17, 10:42:55 AM] Judita Mamuzic: https://indico.cern.ch/event/655274/ but they didn’t make the entries for people to upload slides, this is what I meant

[14.08.17, 10:43:31 AM] Yu-Ting: Oh~ I didn't connect to that indico page

[14.08.17, 10:44:10 AM] Judita Mamuzic: you can also write to Jonana and/or Mike and ask for a entry

[14.08.17, 10:44:11 AM] Yu-Ting: but you can send a mail to let Joana and Mike know we will update yes

[14.08.17, 10:44:20 AM] Judita Mamuzic: ok, I can do that who will do it?

[14.08.17, 10:44:58 AM] Yu-Ting: ok, you can do that because I won't give a presentation today. :)

[14.08.17, 10:45:22 AM] Judita Mamuzic: ok, I will ask, but its on behalf of you… :-) Yu-Ting, are you doing some weighting with the cross-section I think we need to do that because the different contributions for different productions are not equal in your combine method, are you taking into account the cross-section? or you are just adding them up? or we need to do each production separately that would work too

[14.08.17, 10:56:51 AM] Yu-Ting: no I didn't add cross section when I combine the histogram. I add them based on their area. [14.08.17, 10:57:56 AM] Yu-Ting: area h1 = 10 and area h2 =5 then the combined h = 10/15 h1_normalized + 5/15 h2_normalized [14.08.17, 10:57:59 AM] Judita Mamuzic: we need to take the cross-section into account [14.08.17, 10:58:09 AM] Yu-Ting: how to do that? [14.08.17, 10:59:45 AM] Judita Mamuzic: we need to scale each histogram by the cross-section for that point and production (for Higgsino and NUHM2), and then add them up [14.08.17, 11:00:14 AM] Judita Mamuzic: or we need to run the code for each production, and get event weights for each production [14.08.17, 11:03:34 AM] Yu-Ting: TRUTH3 cannot get the event weight information [14.08.17, 11:04:23 AM] Yu-Ting: how to scale histogram by Xsec? [14.08.17, 11:04:26 AM] Judita Mamuzic: you need to do it yourself [14.08.17, 11:04:38 AM] Judita Mamuzic: we discussed this before, we converged [14.08.17, 11:04:54 AM] Judita Mamuzic: you can avoid that by producing the event weights per production [14.08.17, 11:05:08 AM] Yu-Ting: I don't understand [14.08.17, 11:05:14 AM] Judita Mamuzic: you did produce mll plots before with event weights [14.08.17, 11:05:47 AM] Judita Mamuzic: in the note you have one mll plot, where you were investigating C1 decays, those are normalised to luminosity [14.08.17, 11:05:51 AM] Judita Mamuzic: so you did this before [14.08.17, 11:11:48 AM] Yu-Ting: no all of my plots are normalised to equal area [14.08.17, 12:28:28 PM] Judita Mamuzic: Ah, Yu-Ting, there is a difference: Lorentzo was saying you should fix the paramters to dM, but he didn’t say we should discard mll > dM [14.08.17, 12:28:45 PM] Judita Mamuzic: I think the fit function can go to higher values than mll [14.08.17, 12:28:50 PM] Judita Mamuzic: we should try that I think [14.08.17, 12:28:59 PM] Judita Mamuzic: lets see first how this goes on the meeting today [14.08.17, 12:29:21 PM] Judita Mamuzic: they might say it is all OK for the 400, 700 and 800 point, and this can be sufficient for the NUHM2 line [14.08.17, 12:31:25 PM] Yu-Ting: no we can not use mll > dm as I already said. [14.08.17, 12:31:44 PM] Yu-Ting: weight = func_NUHM2 / func_Higgsino [14.08.17, 12:32:14 PM] Yu-Ting: when dm_NUHM2 < mll < dm_Higgsino, func_NUHM2 = 0 and func_Higgsino != 0 [14.08.17, 12:32:23 PM] Yu-Ting: 0 / non-zero = 0 [14.08.17, 12:32:39 PM] Judita Mamuzic: ah, yes [14.08.17, 12:32:40 PM] Judita Mamuzic: ok [14.08.17, 12:33:23 PM] Judita Mamuzic: but when you leave the fit function use the whole dm, it will be func_NUHM2 != 0 [14.08.17, 12:34:07 PM] Judita Mamuzic: we should check when func_NUHM2 == 0 for that case [14.08.17, 12:34:32 PM] Yu-Ting: I have set the range for the function from 0 to dm [14.08.17, 12:35:04 PM] Judita Mamuzic: yes, it depends on the definition of the function [14.08.17, 12:35:29 PM] Judita Mamuzic: if it is similar like yours Gaus+Exp, I it can just have a better fit, and we could use the full range [14.08.17, 12:36:03 PM] Judita Mamuzic: but in any case, tails are less important than the mll < dM [14.08.17, 12:36:16 PM] Yu-Ting: the Gabs + Exp cannot manage the tail well especially at the dm part [14.08.17, 12:36:35 PM] Yu-Ting: If you check my previous slides, then you can see large deviations. [14.08.17, 12:36:48 PM] Judita Mamuzic: yes, I know, but the Lorenzo’s finction could have a better function [14.08.17, 12:36:52 PM] Judita Mamuzic: that was my point [14.08.17, 12:37:31 PM] Judita Mamuzic: I am not suggesting to go back to Gaus+Exp, but to use the Lorenzo’s funcition, with fixed parameters, but not setting the mll = 0 for mll < dm [14.08.17, 12:38:12 PM] Yu-Ting: ok, then how will you do when dm_NUHM2 < mll < dm_Higgsino [14.08.17, 12:38:34 PM] Judita Mamuzic: we should see if the fit function is > 0 [14.08.17, 12:38:42 PM] Yu-Ting: in this region, the Higgsino is signal but the NUHM2 is tail. [14.08.17, 12:38:43 PM] Judita Mamuzic: and why it kicks below 0 [14.08.17, 12:38:52 PM] Yu-Ting: It is unreasonable to reweight the signal to the tail [14.08.17, 12:39:45 PM] Judita Mamuzic: in principle, for large dm_NUHM2 - dm_Higgsino, you are right, but for the small mass diffence it affects a small range [14.08.17, 12:40:03 PM] Judita Mamuzic: we already have a few bins at the end of NUHM2 distribution that are a bit off [14.08.17, 12:47:31 PM] Judita Mamuzic: I have added the mll calculation functions plots to your slides