Hi Gunther,
This one, in particular, has always bothered
me:
As an example, the Jet resolution UPP in my notes is "Jet resolution of < 150%/sqrt(E)" (did we specify at which E, as the resolution does not go as sqrt(E), but has a constant term from the AuAu underlying event?). Our simulations however show a resolution that is much better than that. E.g. 15% at 50GeV from simulations vs 150%/sqrt(50) = 21% from the UPP.
Forgive me if this gets a bit long...I'm
kind of walking through things in detail...
First, the statement above was originally
(before UPP's) that we require a " jet energy
resolution from the combined calorimetry system of
< 100%/sqrt(E) in quadrature with a small constant
term." In subsequent presentations I have "defined"
small as < 15%, but the important point is the
stochastic term as this dominates the energy
resolution for low energy jets.
The origin of this statement, as noted, is
that we want the intrinsic energy resolution of the
calorimeter system to be smaller than the fluctuations
introduced by the subtraction of the underlying event
in heavy ions. In this sense it really is some sort
of a limit - it is either smaller or it isn't - so the
fact that this has been blown up to 150%/sqrt(E)
confuses the issue a bit. If I Iook at the plots
Dennis and the CO group sent around on 4/24
(especially the lefthand plot on slide 3) according to
simulations we're clearly in that limit, as evidenced
by the fact that the energy resolution in p+p is
substantially smaller than for jets of the same energy
embedded in HI's.
In previous reviews the committee hasn't
really understand/concentrated on the subtraction of
the underlying events in heavy ions. They tend to
focus on the performance specification in terms of the
jet energy resolution and ask if we can achieve that,
especially with the inner HCAL replaced with an Al
frame. Songkyo has approached this by asking what is
the jet energy resolution we could achieve in p+p
using a semi-realistic calibration procedure that uses
sample of gamma+jet events (with the expected
statistics). With the CDR configuration she can get
~19% at 25GeV (just making the 100%/sqrt(E) spec) for
R=0.4 jets and closer to ~14% @ 25GeV with an
instrumented SS310 inner HCAL. So, approaching the
specification from this direction, given the original
100%/sqrt(E) specification the CDR configuration
barely makes it and it is comfortably better than the
expanded 150%/sqrt(E) spec.
(for Songkyo's plots see https://indico.bnl.gov/event/4012/
, slide 7 shows the resolutions)
However, if I Iook at the plots Dennis sent
around on 4/24, on slide 3 he shows a resolution in
p+p for R=0.4 jets that is about 0.14 at 25GeV for the
CDR configuration, while Songkyo has an energy
resolution of ~0.19.
Question to Dennis/Rosi - This makes me
worried. Shouldn't these numbers be very similar for
p+p jets between these two approches? I'm concerned
that what might be shown in the HI plots is the
instrumented SS310 instead of the uninstrumented Al
frame? Is that possible?
I realize that what the CO group and Songkyo
are showing approach similar things from different
directions, but I would like to better understand the
difference before CD-1. The plots from the CO group
would seem to argue that even with the inner HCAL
replaced by a frame we are still in a regime where we
are dominated by the fluctuations in the underlying
event. Songkyo's work would seem to seem to argue
that we are closer to the edge. If the combination of
the calorimetry and our ability to calibrate it in the
end results in worse performance, we should
acknowledge that, especially it is close to what we
have previously stated as "the limit".
John
On 4/27/2018 8:39 AM,
Gunther M Roland wrote:
Friends,,
Maria just reminded us that we will have a dry-run of the CD-1 review plenary talks next Wednesday. I am not sure we will quite have the final plots at that time, but we need to figure out what the line of argument is going to be.
We have a nice set of new simulations for the MIE configuration for e.g. jet energy resolution and various physics plots. I'm now thinking of how to make the connection to the UPPs for the OPA talk, and I'm finding that not so easy.
As an example, the Jet resolution UPP in my notes is "Jet resolution of < 150%/sqrt(E)" (did we specify at which E, as the resolution does not go as sqrt(E), but has a constant term from the AuAu underlying event?). Our simulations however show a resolution that is much better than that. E.g. 15% at 50GeV from simulations vs 150%/sqrt(50) = 21% from the UPP.
So, my questions/suggestions:
- I assume it is ok if the UPP is somewhat relaxed compared to the performance in simulations?
- We probably should add a line on the performance plots reflecting the UPP performance
- We should add a set of points corresponding to the UPP performance to the physics plots
- for the physics plots, one needs some to provide some scale of why we want to reach a certain performance. I'm planning to argue that we want comparable performance to that at LHC (e.g. gamma-jet balance uncertainties, Upsilon resolution). Any better ideas?
- for the t-shirt plot, it would be nice to have for comparison also the current RHIC status (applies only to jets/hadron RAA) and a backup plot of how much the MIE vs reference acceptance costs us
Question to Dennis/Rosi: Any chance of getting a photon resolution plot from simulation to compare to the photon UPP of <15%/sqrt(E) photon resolution? I think it would be good to fold this into the physics plot.
Question to Tony: Where can I find the latest Upsilon mass plot for pp? The UPP (Y(1s) mass resolution < 100MeV) probably needs to refer not just to the system, but also specific running conditions: single central Au+Auevent, max pileup, average pileup? It would be good to have a comparison of the resolution for pp, single central Au+Au and central Au+Au at max pileup (beginning of store) to see where we stand.
Cheers,
Gunther
