[Hps-analysis] HPS Analysis Meeting June 26 @ 10am/1pm PDT/EDT

[Graham, Mathew Thomas]

Sorry, I missed this email.  

So slide 5 has both the true and the recon Esum/Ebeam>0.8; Slide 4 is just the recon, right? 


> The A' efficiency at the target is an input to the analysis - you use the radiative fraction to get the number of reconstructed radiative tridents, then you use the efficiency at the target to get the number of radiative tridents in your luminosity, then you convert to number of A'. For this trick to work, the numerator of your radiative fraction needs to be defined the same way as the denominator of your efficiency at the target.

> 
> In my opinion, maintaining this apples-to-apples-ness, and avoiding systematics, is a little more complicated if you do a mass cut and your signal definition is "A' with reconstructed mass within 3 sigma" or something like that.

Yeah true, but I think that in fact the mass binning is more like 2 sigma, isn’t it? 

> The only place this falls down is at the end: even if we don't think of low-x A' as part of our signal, they will appear in our data if the A' exists. But: we're not fitting our signal so the signal shape is not very important, the low-x component is way subdominant and will have smaller Z (reduced boost, recoil acceptance), and this is a background that only appears if HPS finds the A', and in that scenario what are we complaining about?

Do these plots have the e+e- truth selected or do they contain recoil pairs?  I’m guessing they must contain recoils based on how big the tail is.   

> 
> Nan - this is a nice study. I hadn't thought about the mass and Z dependence.
> 
> On Tue, 26 Jun 2018, Solt, Matthew Reagan wrote:
> 
>> Thanks for looking into this Sho.
>> 
>> 
>> I think bottom line, I just apply the truth x cut for the efficiency curves. I don't think this is cheating since we care about the truth for these.
>> 
>> 
>> Nan (a rotation student), did a quick and dirty study similar to this. He cut on the residual of both electron and positron momentum, decay z, and mass at 3 sigma for all these. Then he took the ratio between the efficiency curves (as a function of z) of those curves and the curves without those cuts. Results are below (ratio vs z for several A' masses). I originally was curious in the affect of the recoil electron in the reconstruction (those cuts should get rid of most of them), not the radiative tail as discussed this morning. The results are what I expected, a very large affect for low mass (which always has a large tail in mass), and a slightly larger affect neat the target where reconstructing with the recoil electron is more likely. In our mass and z region of interest (~>30 mm and mass > 35 MeV), it seems like a 5-10% effect.
>> 
>> 
>> [cid:572ab039-93d9-433d-ab75-634925e2a259] at
>> 
>> Same plot but with error bars
>> 
>> 
>> [cid:700bafb6-2e6f-4e9f-a468-583c70a63e60]
>> 
>> 
>> More details are here if you want to see some of the tails he cut out: https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_nanw0321_HPS_tree_master_Plots_vertex-5Fcut&d=DwIGaQ&c=lz9TcOasaINaaC3U7FbMev2lsutwpI4--09aP8Lu18s&r=J4PP6Zl8IyGHpsqWaKegORCYw8hoCHePTw5O95a5lqQ&m=zy4qkq1g34EsAiNpE-VFlnSpCitGuZptIHR5-T463v8&s=nD5O0Tu1qOL5IZiefRZfs0GkUb0If2pFrsl5j0MemY4&e= 
>> 
>> 
>> Matt Solt
>> 
>> ________________________________
>> From: Hps-analysis <hps-analysis-bounces at jlab.org> on behalf of Sho Uemura <meeg at slac.stanford.edu>
>> Sent: Tuesday, June 26, 2018 1:44:52 PM
>> To: Graham, Mathew Thomas
>> Cc: hps-analysis at jlab.org
>> Subject: Re: [Hps-analysis] HPS Analysis Meeting June 26 @ 10am/1pm PDT/EDT
>> 
>> I read the note, it looks good so far. I've made a couple of typo fixes
>> and I'll put more substantive comments in a text file (in the Dropbox,
>> comments_sho/comments.txt), but the only things I have there right now are
>> things I screwed up in my thesis.
>> 
>> I looked at the low-mass tail (pyroot macro and plots attached). See the
>> comments in the macro for detail, but summary: I'd misremembered what the
>> cut is that cleans it up: it's not the radiative cut on the recon vertex,
>> it's a x>0.8 cut on the truth A'. So, that's obviously cheating (in my
>> thesis I say I'm doing it, but completely sweep it under the rug). But the
>> tail I saw (might not be the same tail you see) is definitely recoil e-
>> misidentification in low-X A' events.
>> 
>> On Tue, 26 Jun 2018, Graham, Mathew Thomas wrote:
>> 
>>> 
>>> Hi All,
>>> 
>>> We?ll meet tomorrow for an update on things in the run-up to ICHEP.  We?ve got both the 2015 bump-hunt and vertexing analysis aiming for it and this Wednesday is the nominal deadline for presenting the results to the collaboration.
>>> 
>>> https://urldefense.proofpoint.com/v2/url?u=https-3A__confluence.slac.stanford.edu_display_hpsg_June-2B26-252C-2B2018-2BDAWG-2BMeeting&d=DwIGaQ&c=lz9TcOasaINaaC3U7FbMev2lsutwpI4--09aP8Lu18s&r=J4PP6Zl8IyGHpsqWaKegORCYw8hoCHePTw5O95a5lqQ&m=k4vmYUNmn_n-5PP-EHertr6XxvPSqcKO5XvWw2_iXGE&s=tfApseM4CjHLmQsSyzN2_qiQXCcyyhPiyBz4KuhbuJs&e= <https://urldefense.proofpoint.com/v2/url?u=https-3A__confluence.slac.stanford.edu_display_hpsg_June-2B26-2C-2B2018-2BDAWG-2BMeeting&d=DwIGaQ&c=lz9TcOasaINaaC3U7FbMev2lsutwpI4--09aP8Lu18s&r=J4PP6Zl8IyGHpsqWaKegORCYw8hoCHePTw5O95a5lqQ&m=k4vmYUNmn_n-5PP-EHertr6XxvPSqcKO5XvWw2_iXGE&s=0pcsXe14RWQBSJw6vopXDE59LApRfHspxGQ9MmL_W5o&e=>
>>> 
>>> Thanks, Matt & Nathan
>>> 
>>> 
>>> *   Browser: https://urldefense.proofpoint.com/v2/url?u=https-3A__bluejeans.com_722990411&d=DwIGaQ&c=lz9TcOasaINaaC3U7FbMev2lsutwpI4--09aP8Lu18s&r=J4PP6Zl8IyGHpsqWaKegORCYw8hoCHePTw5O95a5lqQ&m=k4vmYUNmn_n-5PP-EHertr6XxvPSqcKO5XvWw2_iXGE&s=zv9zYKaJUIsEnSy69p9nhR411Ne36jaUOe7bumSNm8k&e=
>>> *   Phone:  8882402560  722990411  ##  *4
>>> *   Details:  https://urldefense.proofpoint.com/v2/url?u=https-3A__confluence.slac.stanford.edu_display_hpsg_DAWG-2BMeetings&d=DwIGaQ&c=lz9TcOasaINaaC3U7FbMev2lsutwpI4--09aP8Lu18s&r=J4PP6Zl8IyGHpsqWaKegORCYw8hoCHePTw5O95a5lqQ&m=k4vmYUNmn_n-5PP-EHertr6XxvPSqcKO5XvWw2_iXGE&s=k9XGeUQtlLm2joswzEwU0A6I-8sNf26yGFyjfTUwUpo&e=
>>> 
>>