LPC meeting summary 05-12-2016 - final
Main purpose of the meeting: Status of the investigations of the origins of the luminosity difference in IP1 and IP5.
Jamie summarised the succesfull proton lead run with some first preliminary physics related plots which have been shown during the LHCC.
The EYETS programme has now started with the magnet training of sectors S34 and S45 to nominal energy (14TeV). Beams are expected back in May 2017.
Keith summarised the studies of the lumi ratio on behalf of CMS. The ongoing investigations of CMS go in two directions: LHC beam effects (crossing angles, emittance, β*, orbit drifts), intrumental effects in CMS (calibration, acceptance...) and in adition there could be instrumental effects in ATLAS.
CMS is currently working on the anlysis of Fill 5422 in which the crossing angle was scanned. The VdM scans taken during this fill are being analysed. In addition, in collaboration with ATLAS, a Z-rate comparison which should be independent of the luminosity scale is being worked out for two long physics fills.
Fill 5422 contained four isolated bunches with two different emittances leading to pile-up values of 10 and 25 resp. This allows to study pile-up dependent effects. During the flll the crossing angle was changed from nominal to 0 in 5 steps. Full VdM scans were performed at the beginning (full crossing angle) and the end (0 crossing angle) of the fill. For the intermediate crossing angles CMS performed mini-scans. A preliminary analysis of the online luminosities shows that the lumi ratio depends on the crossing angle (as expected) but also on the the pile-up. This can be seen in the plot of slide 5: At zero crossing angle both the low-emittance and the high emittance bunches show a ATLAS/CMS lumi ratio closer to 1, as expected (the geometric factor is 1 at this angle). (Note that it is a coincidence that the high emittance bunches at zero crossing angle seem to have a lumi ratio of exactly '1', because the CMS VdM calibrations still have a large uncertainties and in addition there could be small β* differences in both IPs.) However the ratios for the low emittance bunches are smaller since the ATLAS online lumi values which were used in this plot, have a known dependency on the pileup which is not corrected for in this plot. The plot also shows the evolution of the geometric factor when going from large to small crossing angles. As expected by the formula for the geometric factor, the effect is larger for low emittance bunches than for the high emittance bunches.
CMS found that the Cap Sigma measurements of all Luminometers and of the PCC offline method (currently only available for the full scans at the start and the end of the fill) are consistent with each other within 1%. They observered a change in Cap Sigma in the crossing plane (X) when going from 140µrad to 0µrad which is consistent with the values expected from the LHC online fits but a change of absolute calibration of 1% - 1.5% is found and currently being investigated.
CMS notes that the ratio of PCC lumi to online lumi depends on the pile-up ranging from 2% at low pileup to 1% at medium pileup.
CMS is providing Z production rates in fills 5198 and 5199 for comparison with equivalent ATLAS data. A preliminary plot showing the ratio of PCC lumi to Z-lumi as a function of the sub-data set (i.e. time) in fill 5198 was shown.
CMS found the calibration constants obtained from the 2016 VdM scans to be 4%-5% higher than those measured in the 2015 scans which is not understood. CMS notes that parts of the 2016 analysis are not yet finalised and that the 2015 analysis is being re-investigated.
Finally CMS remarks that the PCC method does not show severe non-linearities as a function of the pile-up (the inner Pixel layer which suffers from "dynamic" (i.e. pile-up dependent) inefficiencies is not included in the PCC analysis).
In the discussion it was stated that in 2015 the luminosity derived from the PCC method and the one derived from the observed Z-rate agree within 1%. Jamie remarked that during the presentation it was also shown that the 2016 PCC lumi and the one derived from the Z-rate agree. However the calibration constants of the VdM scans are found to be 4% to 5% different for the two years. Keith replied that CMS assumes that either the 2015 or the 2016 calibration are not correct.
Witold gave a summary of the analysis of the crossing angle scan and the luminosity ratio ATLAS/CMS. In the introduction he underlined that the interpretation of the measurements depend critically on the understanding of the μ-dependence of the luminometers. For the anlysis it is important to use specific luminosities. Witold showed that the specific luminosity in the high pile-up bunches decreased with time, probably due to emittance growth (wire scanner and BSRT measurements confirm this). For the low pile-up bunches he found that the specific luminosity increases by 20% when going from 140µrad to 0µrad crossing angle whereas the change for the high pile-up bunches is estimated to be around 30%; the exact numbers depend on when the data are sampled, because of the rapid emittance growth.
ATLAS measured the ratios of CAP Sigma for crossing angle 0 to Cap Sigma for crossing angle 140µrad in both planes. For the X plane the evolution of the ratio for all 4 bunches is consistent with the prediction taking into account the emittance growth reported by the BSRT. However for the low-emittance bunches the ratio in the Y plane is larger than the prediction by up to 7%.
When comparing the Cap Sigma measurment of ATLAS and CMS for all 4 bunches at zero crossing angle a very good agreement is found between the CMS PCC measurements and the ATLAS track-counting measurements. (The caveat of this preliminary result is that the length scale calibration are not yet available and could change the agreement between IP 1 and 5 by a few percent.) Witold showed ratios of Cap Sigma measurements and specific luminosities in the two experiments. Finally he showed the ratios of specific luminosities at crossing angle 0 divided by the same quantity at nominal crossing angle, for all 4 bunches in ATLAS and CMS on slide 16. As expected the ratios are larger for ATLAS where the geometric factor is varies significantly more with the given emittances. The conclusions from this plot are independent of the length scale calibration.
Then Witold showed that ATLAS is observing a time dependence (over the year) of the μ-dependence of the ratio Lumi_Tracks / Lumi_LUCID (slide 19). Integrating this effect over the year, he concludes that the ATLAS offline luminosity is currently overestimated by 3% for a largest part of the year.
On slide 22 the Bunch Lumi ratios (CMS_PCC / ATLAS_Trk) for nominal and zero crossing angle are shown. The lumi ratio at nominal crossing angle results to be 1.11 for all bunches, whereas at zero crossing angle a ratio of 1.03 is found again for all bunches. The deviation from 1 of the latter is believed to originate in imperfect vdM calibrations or β* differences in IP 1 and 5.
Witold then showed a set of slides to seed the discussion on what could be done in 2017 to equalise the delivered luminosity in IP1 and IP5. These thoughts reflect his personal opinion and should not be interpreted as official ATLAS proposals. This discussion will have to continue within the LPC among all experiments concerned and the machine.