LPC meeting summary 19-09-2016 - final
|
|
LPC meeting summary 19-09-2016 - final |
|
Main purpose of the meeting: Update on luminosity determination in CMS and ATLAS (especially Z-counting); views on luminosity leveling (ATLAS and CMS); a test run with high pileup and discussion on the intensity ramp up after TS2.
Jamie summarised the efficient data taking period from the 5th to the 9th of September. He then pointed out that in an informal meeting the details of the 2.5km β* run have been worked out and agreed upon by the participating experiments. Details can be found in the minutes of the meeting.
The intensity ramp-up after the 2.5km β* run will comprise the following steps:
Finally Jamie pointed out that during this weeks LHCC there will be discussion on the possibility to relax the schedule of the dense programme during the p-Pb run. The spokespersons of the experiments have been asked for input into this discussion. Final decisions will be taken after the LHCC recommendations and discussion in the research board.
Jakob summarised the status of the Z-Counting analysis in CMS. The analysis is currently being done for Z decaying into central muons. (Z-decays into electrons are being investigated). Two fills have been analysed so far: 5198 and 5199. These fills are representative for the observed difference in lumi between ATLAS and CMS.
Comparing the CMS online lumi with the Pixel Cluster Counting offline method (PCC) CMS finds that the online Lumi is within 4% of the offline lumi in these fills. The integrated fill luminosities differ by less than 2% (the PCC offline method resulting in the higher value).
In the analysis the fills are divided into 25/pb data sets. 95% of the recorded luminosity in the relevant fills are used. The "tag and Probe" method is used to determine reconstruction, trigger and identification efficiencies. Simulation is used to estimate pileup effects on this method.
A comparison of the CMS online, CMS offline and CMS Z-counting method on average gives a good agreement between the methods. In a table the Atlas/CMS online lumi ratio, the CMS online/z-counting ratio and the CMS PCC/z-counting ratio are compared for the two investigated fills. Where the difference of the online lumis of ATLAS and CMS is of the order of 10% the internal CMS measurements agree with each other to a level better than 2%.
Jakob concluded that CMS is ready to perform a direct efficiency corrected Z counting comparison with ATLAS in order to investigate the observed difference in delivered luminosity of the two experiments.
In the discussion Jakob mentioned that CMS trust most the offline PCC method and that they
know that the online luminosity is currently underestimated at the start of a fill by about
2%.
CMS did check that for the period in during which ATLAS performed the VdM scan, the ratio of
online lumi to PCC lumi is one (as expected). Therefore it is concluded that the online
luminometer changes scale when going from VdM conditions to high pileup physics run conditions.
Jakob stated that currently CMS still uses the 2015 calibration since it still works on
some apparent inconsistencies found after the 2016 VdM scan. Currently no explanation for
these inconsistencies is available. There is no prediction on when the 2016 calibration will
be available.
Jamie underlined the importance of the problem and experiments should give high priority
to understand the details of their luminosities. He asked for the timescale on which the
z-counting could be applied to all fills of 2016. Jakob replied that the current delay is
estimated to be 1-2 weeks. It was emphasized by Jamie and Witold Kozanecki that observing tendencies
over the entire year (e.g. mu dependence) is important.
Peter Onyisi asked why the eta range was only going to 1.9 and not 2.4. Jakob replied that the
forward muons are complicated to understand and this is work in progress.
Peter asked for the triggers being used. Jakob answered that both a "cocktail" of triggers
and simple specific triggers have been used and the results of both approaches are
consistent.
Michi Hohstetter asked if CMS looked at the various lumi ratios during the lumi scans. Jakob
answered that in these scans not enough statistics per point is acquired. A special trigger
would have to be developed for this. In addition it would have to be investigated if the
scan in this case would have to be extended significantly. Witold estimated that in this case
the scan would become similar to a mu scan and would probably take 30 to 45 minutes.
Peter gave an update on the z-counting analysis in ATLAS which is now done in lumi blocks of 1min. Results were corrected to full acceptance using the Powheg Monte Carlo without additional corrections.
Peter showed that using the latest ATLAS calibration the pileup dependence of the analysis is slightly reduced (improvement in μ linearity). The expected non-linearity effect still needs to be determined from Monte Carlo. Also was shown the stability of the analysis over time. Some outliers in the relevant distribution (slide 5 and also on slide 7) are known to be due to trigger problems. Finally the comparison of the z-counting luminosity to the ATLAS offline luminosity was shown to be in good overall agreement.
The further steps to proceed were discussed. All parties involved agreed that if we want to draw conclusions relevant for next year from this exercise, there is need for a thorough offline analysis of at least some fills.
Peter stated the ATLAS numbers should be signed off by the ATLAS lumi group. In a follow up meeting the fiducial volume and the treatment of the systematics should be agreed upon by CMS and ATLAS. The LPC should follow up with the management of the collaborations to make sure that there are no hurdles due to the fact that z-counting is closer to "physics" than the "conventional" online lumi measurement methods. Jakob agreed to organise a kick-off meeting to discuss the remaining technical items for the comparison among the experiments. Witold recommended to agree on handful of fills to exercise the comparison before applying it to the entire 2016 data sample.
Alessandro stated that switching between standard and low pileup mode for ATLAS is currently a long procedure and cannot be done during a fill without significant loss of data (e.g. at the end of a physics fill). Atlas will take data during the 150b fill with a μ between 0.15 and 1.0. During the 600b fill the μ will be adjusted to 0.3 (it has been verified that this is compatible with the planned pressure tests during this fill). During the 1200b fill the μ will be probably adjusted to 0.5 to 1.0. ATLAS would like to know if CMS is interested in low μ running during the 600b fill after MD4 since they would like to plan in function of the CMS requests.
ATLAS thinks that before leveling becomes a realistic option some things need to be addressed: Among these are the difference of the luminosities seen by CMS and ATLAS and the instabilities of the luminosities in IP1. Reyes noted that in IP1 the lumi leveling implemented by ATLAS in the past would be used (LHCf needs to be leveled to a low luminosity during their p-Pb run). Also the scenarios in which leveling will be applied need to be clarified as the algorithm used to perform the leveling.
ATLAS is interested in a high pileup fill and has not new requests concerning this topic (see slides).
CMS will remove the main bottleneck for higher luminosities during the EYETS with the Pixel
upgrade. In order to better investigate in which scenario leveling would be desirable for
CMS, more input data as a guideline is desirable as soon as possible. These include the
foreseen number of bunches in the machine, BCMS or "standard" running schemes, how hard is
the limit of the triplet cooling limitation (1.7e34), What is the expected peak lumi and pileup
profile for 2017 and beyond (optimistic estimates)? In the discussion it was stated that the
triplet cooling limitation is a number with large error bars, but it is clear that for short
periods the limit can be exceeded without quenching the magnets (this could be interesting
to find the peak luminosities in the experiments if leveling down a constant fraction from
this value is desired.)
CMS would like to know the possible options of leveling techniques and would like to learn
more details about the various methods. Important aspects are length of the luminous region and
the pileup density in the various schemes. Also the expected integrated lumi, the optimal fill
lengths, and the lumi profile per fill are important aspects to consider.
According to CMS it will be always difficult to guarantee the same delivered luminosity to
CMS and ATLAS. CMS will always provide the best lumi estimate possible. CMS asks if
β* leveling would guarantee "proportional" leveling from first principles.
Some of these questions will not be easy to answer and there will need to be an iterative
procedure to converge on the different possible options (β*, crossing angle,
injection-scheme etc…) since it is not possible to do detailed studies for every option
to try to answer these questions.
Concerning the high pileup test fill CMS would request at least one train with at least
12 bunches with a peak pileup at least 20% than usual (ideal would be 55-60). It would be
good to let the pileup decay to today's "standard" peak values during the fill. The fill
should happen just after the MD4.
CMS also stated that towards the end of 2017 they would like to have a high pileup fill
with conditions similar to Run 4. They also investigate if a fill with 50ns bunch spacing
or individual bunches could be useful for this exercise.
In the slides CMS lists the requests for the ramp up after the 2.5km β* run.
ALICE stated that they would like to do a my scan after the 2.5km β* run. In the slides ALICE presents the target lumis during the various ramp-up fills.