[Hps-analysis] tPass1

[Alessandra Filippi]

Hi all,
just a few more words to substantiate better what I was mentioning yesterday
at the analysis meeting (sorry, I can't make it today for the software meeting
due to the critical time slot).
I compared the output quality for one of the processed runs for pass0 and
pass1: there is no big difference between the two as far as the reconstruction
efficiency is concerned - maybe slightly better for pass1, for which the
alignment is better as well (not excitingly, but improved).
However, as you can see in the first page of the pdf in attachment, the 
plots of the momenta distributions for all tracks without any particular
selection, there is (in both the cases) a striking inefficiency between top
(black) and bottom (red), that is the reason why I was alarmed yesterday; 
top tracks are about 30% less. This issue was also present in pass0, but at 
that time I did not realize as I was just looking at residuals and 
alignment quality.

I cannot say at the moment which is its origin; in all my tests (that 
however were run with an older -wrong?- version of the db) I never noticed 
such difference in top vs bottom inefficiency. As I was saying yesterday, 
when running my tests reading the new db and old/updated code it looks 
like I'm loosing tracks overall (not only on the top), but I'm still inquiring 
on this.

Something else which is weird, but maybe has an obvious explanation, is
the pattern of the predicted track hits on svt layers for positive and
negative tracks (which has the same behaviour for both pass0 and pass1, 
again). I report the plots on the last two pages of the pdf. While the 
positrons have a normal behaviour, for the electrons there is a sort of 
"hole" in sensors of the slot side (that has some shadow also on the first 
three). Is this normal? Is it just the shadow of the calorimeter hole? 
(that narrow and as high as to span the full height of both top and bottom 
sensors??)
This pattern is peculiar of 2016 data. Probably it would be useful to 
check it with MC data.

Finally, something else that I do notice in running my gbl and refit stuff 
on the gbl ascii file is that now there are a lot more faulty occurrences
with crazy curvatures and residuals. I don't know yet whether there could 
be any connection among all these issues.
cheers
    Alessandra





On Tue, 13 Jun 2017, Graf, Norman A. wrote:

> Hello All,
>
> You are encouraged to analyze the output of tpass1 to help catch any errors or omissions in this next pass over the 2016 data. Details of the existing test pass, and instructions on how to run it if you would like to process any particular data samples you might find useful, can be found on the following confluence page:
>
> https://urldefense.proofpoint.com/v2/url?u=https-3A__confluence.slac.stanford.edu_pages_viewpage.action-3FpageId-3D223224540&d=DwIFAw&c=lz9TcOasaINaaC3U7FbMev2lsutwpI4--09aP8Lu18s&r=J4PP6Zl8IyGHpsqWaKegORCYw8hoCHePTw5O95a5lqQ&m=vyg1pN9LzLiWvuxUKzlsKrKSc741i1TbskQgz96_-2U&s=zzDqeVDiz7HQxMN4Y-Zj0aVwtOegCHODKbmk96n3lT0&e=
>
> Norman
> _______________________________________________
> Hps-analysis mailing list
> Hps-analysis at jlab.org
> https://mailman.jlab.org/mailman/listinfo/hps-analysis
>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: pass1Quality_momenta.pdf
Type: application/pdf
Size: 201108 bytes
Desc: 
URL: <https://mailman.jlab.org/pipermail/hps-analysis/attachments/20170614/db899f1e/attachment-0001.pdf>