Questions and answers for DN2021-022

General comments: The major comment: the TP performance results are mostly presented with respect to offline segment - this is highly biased. As the paper says, the Phase 2 electronics upgrade makes the TP performance comparable to offline performance. Let’s push it to extreme - assuming your electronics are so powerful that they are capable of doing exactly the same thing as offline reconstruction, then the efficiencies and resolutions you measure here are an object with respect to itself. (The efficiencies measured will always be 100%, while the absolute efficiencies might be only 10% due to a bug somewhere). I understand the paper has shown firmware vs emulator comparison, and good spacial and timing resolutions in simulation guarantee good resolutions in firmware. But still, absolute efficiencies are not discussed anywhere.

Indeed you are correct that performance results of differences wrt segments can be biased, in particular when considering correlated TPs. This is actually stated in the paper in this sentence 'It is worth noting that these TP to segment differences are not to be considered as intrinsic TP variable’s resolution measurements, as relevant correlations between TP fit results and variables obtained from the offline segments are possible due to sharing of hits. And this is the reason why we also have results comparing TP with simulated quantities, see next paragraph 'Dedicated studies using as reference simulated quantities have been performed'. We would like to keep both approaches (wrt segments and wrt simulation) and provide the full information in the paper.

- Do I understand correctly that the new TP algorithm would not have been possible without the new capability of 1 ns digitisation? This is not quite clear in the description of the 4 steps (Grouping, Fitting, Correlation, Confirmation).

Indeed, the 1ns digitization is a fundamental feature for achieving the ultimate resolution. The steps of the algorithm are one way to solve this problem that allow to deal with the new architecture of the system and are also responsible of providing the shown efficiencies and rates. Even if the algorithm would be 'formally possible', the resolution performances would significantly change if this digitization step was significantly different. We have a sentence trying to emphasize the relevance of this new feature in the introduction "Given the improved time digitization of Phase-2 DT input hits of 1~ns, the generated Phase-2 TPs are expected to provide a measurement of the time of the collision where the muon was generated both in ns ($t_0$) and in 25 ns bins (bunch crossing number, BX), and of the muon segment parameters (position and direction), with a resolution comparable to what is reachable with the present offline reconstruction software~\cite{MuonDPGpaper}. " To make this more clear, in the conclusions section we have modified "Results obtained by running a software emulator on Phase-2 simulations show good performance in terms of efficiency, resolution, and rates" by adding. "The new capability of around 1 ns digitisation in the TP generation is determinant for achieving these results." We have also added an explicit mention of the new time digitization step in the abstract. "In the new design, the time bin for the digitisation of the chamber signals will be of around 1~ns, and the totality of the signals will be forwarded asynchronously to the service cavern at full resolution."

- The paper mentions many times “ageing effects”, which gives me an impression that the paper were called “DT TP in HL-LHC with ageing effects”, rather than the current title.

Ageing effects are indeed taken into account for some studies presented, but it is not true in all cases. We have tried to make clear in the text when ageing effects are taken into account or not. We prefer to leave the title as it is, as the use of ageing is not a general statement

- Abstract: … and resolutions are close to the ultimate performance of the DT chambers >> and time and spatial resolutions are close to the ultimate performance of the DT chambers

Done

[2] is for GE2/1 and ME0. please add: CMS Technical Design Report for the Muon Endcap GEM Upgrade CMS Collaboration (V. Khachatryan et al.) CERN-LHCC-2015-012, CMS-TDR-013 . ISBN 978-92-9083-396-3

Done

wheels (Wh +/-2), raising >> wheels (Wh+/-2), raising … no space after “Wh”

Done

..that are geometrically matched with a generated muon with pT >20 GeV within … >> “T” is in italic. Use roman.

Done

Fig. 3… the points on Wh0 are the largest for MB1, 2, 3. But a systematic drop (black solid circles) at Wh0 for MB4. Why?

Actually MB4 receive a lot of radiation in general, so a shielding has been put in place already to mitigate aging effects. However, for MB4 Wh0 in particular, due to integration constrains, the shielding covers only half of the chambers in sectors 1,2,3 and 5,6,7, and it was not possible to shield sector 4 for MB4 Wh0 at all. This can also be seen in Fig 2, upper panel, where the hit efficiency is lowest for MB4 Wh0 S4. It is this hit inefficiency in MB4 Wh0 that translates into the TPG inefficiency you then observe in Fig 3. Actually this behaviour of lower efficiency in MB4 Wh0 is not observed in the plots if we do not consider ageing in the chambers (please see [[https://jleonhol.web.cern.ch/dts/hEff_AM_rossin_noRPC_noAgeing_ext_confok_alignTrue_0.png])[here]].

obtained by running a software emulator on Phase-2 simulations show good performance in terms of efficiency, resolution, and rates. >> time and special resolutions, and rates.

Done

GENERAL

Given one of the internal referees from MUPC is native english speaker, we hope v2 is good enough on this. We can also investigate if we will have a LE when we get a CADI line and go to CWR.

Where you write "simulated and data samples" I would add "real" before "data" as simulated data is data as well.

Done

p.1 muon reconstruction and triggering in CMS is performed --> are performed

Done

the barrel chambers are arranged in stations... in increasing radius --> with increasing radius

Done

HL-LHC start in 2027 ..... I'm afraid it has shifted further meanwhile...

Changed to 2029

p.2 end of introduction "a measurement of the time both in ns and in 25 ns bins" not totally clear... Perhaps you could enhance a bit the gain in performance. What about: "a measurement of the time of the collision generating the muon with a 1 ns granularity (t0): much finer than the 25 ns bins needed for BX identification. The muon segment parameters will have.."

Done

p.3 end of par.2 isn't this phi the standard CMS azimuthal coordinate? why "w.r.t. each sector origin"?

In trigger the phi coordinate is given per sector, wrt each sector origin. Rephrased as "The trigger primitive position is finally translated to CMS global sector coordinates, being φ the azimuthal angle of the TP position with respect to each sector origin (situated in the middle of the sector), and with the muon bending angle defined as φB = ψ−φ."

in addition: "being phi the azimuthal angle of the TP" might be a bit confusing as the TP is a segment and has its own angle... perhaps "being phi the azimuthal coordinate of the TP position" ?

Rephrased as "being φ the azimuthal coordinate of the TP position with respect to each sector origin

par.3 when describing the Rossin's sample it could be worth to stress that +-16 BX is the relevant DT time range within which hits to be fitted are searched for.

Rephrased to "Additional proton-proton interactions are generated within a window of ±16 BX around the central one, fully covering the maximum drift time of ~390 ns, with an average value of 200 pile-up events per bunch crossing."

"in particular neutrons that can produce hits" --> "in particular thermal (or: low energy) neutrons that produce.."

Done

From: "In these samples, DT offline segments" to par.2.1, there is a redundancy of the string "DT" DT segments, DT cell, DT chambers, DT hits...... consider taking some away

Done (thanks! :))

"In the following a scenario" --> "In the following, a scenario"

Done

"After the analysis of the data a degradation" --> "based on this data, a degradation" "is expected only in the most..."--> "is only expected in the most.."

Done

"however, due to the redundancy of the system and the mitigation actions implemented at the current system the optimal performance of the detectors during the HL-LHC is foreseen" --> "however thanks to the redundancy of the system and to the mitigation measures implemented on the current system, a good performance is still expected during the HL-LHC phase."

Done

"in this scenario lowest DT chamber efficiencies" --> "In this scenario, the lowest DT chamber efficiencies"

Done

par 3.1 "Prompt muon efficiencies" --> Efficiency to trigger prompt muons" (?)

Done

The string "TP" also appears many times, perhaps you could simply use trigger in some cases.

We got a comment stating the contrary before... We prefer leaving it like this for the moment

"The numerator of efficiency is defined as the number of trigger primitives with fitted value at the correct BX that match..." --> "The numerator of efficiency is defined as the number of trigger primitives with fitted time at the correct BX and matching..."

Done

End of p.3 "The efficiency drop can be recovered by accepting qual>1" here a question on the expected effect on rates arises spontaneously

We have Jaime's estimations on previous samples (TDR samples) with 250 PU. In IN2019_003, the average rate per sector at 9E34 for all qualities was 0.8MHz, when adding random hits for simulating neutron background. Removing only 3hit-fits (qu>=3 at that time) was around 0.5MHz. This is not far from your estimation on Rossins sample for (new quality definition) qu>2 if extrapolated to 9E34 (350/200*250=437 KHz). I would say it is safe to say that adding qu=1 the rate estimate should remain similar. We could add a sentence on the rate section along these lines if needed "The effect of adding TPs with qu>1 on the rate estimate is not substantial", but not sure if it would be more confusing, and at the moment we dont have a fully robust estimation. Would you like having a look at Rossin's sample for this?

Agreed to leave the text as it is for the moment and pursue studies in parallel

"the final efficiency is unaffected with a drop of <1%" --> "the final efficiency is unaffected by a drop <1%"

.....but the drop visible in fig.3 is >> 1%...!

The drop in Fig.3 is at the level of chamber. What we are trying to say here is that the output of the full L1 trigger chain (which considers more than 1 chamber and even different detectors) is not as much affected by an efficiency drop in a single DT chamber. We have modify the sentence to: "In addition, previous studies [5] show that the final effect on L1 Muon trigger efficiency is small, dropping <1% in the barrel-endcap overlap region, and less else-where"

Agreed also with Carlo to rephrase as: "In addition, previous studies based on Phase-1 TPs~\cite{L1TDR} show that the final effect of the ageing on L1 Muon trigger efficiency is small, dropping $< $5\% in the barrel-endcap overlap region, and less elsewhere."

Fig.3 is not very well readable in black and white.

This plot has been officially approved already as it is, but if you have specific suggestions we can consider them for the next iteration (we could for example changing markers shapes)

. Plot has been redone

Regarding the drop of this efficiency referred to segments with ageing, it could be worth to specify that the procedure applied for ageing simulation doesn't change the the segments (I myself keep forgetting that hits are only removed in input to the emulator!!!)

Added "for TP generation" here" "In the following, a scenario equivalent to 3000 fb-1 has been considered for TP generation, corresponding to extreme ageing effects in the DT detector at the end of HL-LHC operation.

p.4 par 3.2 "the differences on TP position... wrt corresponding values" --> "the differences of TP position.."

Changed to 'in' for v2, after comments from MUPC

58 urad --> 58 um (text referring to fig.4 bottom)

Done

"The improvement in the bending angle measurement for correlated TPs is due to" --> " The better measurement obtained with correlated TPS is due to"

Done

"these TP to segment differences are not to be considered as intrinsic TP variable resolution measurements" --> "these TP to segment differences are not to be considered as intrinsic measurements of the TP variable resolution"

Done

"two muon pairs with with flat" --> "two muon pairs with flat"

Done

"without any contributions" --> "with no contributions"

Left as is, after review by MUPC.

par.3.3 "The same simulated sample described above" this must be Rossin's sample but it is not the last sample you have been talking of in par.3.2!

Indeed, changed! The same simulated sample used for the efficiency studies described above "

"factors of 2 to 15 smaller" --> "factors 2 to 15 smaller" (? not sure!)

From your plots, MB1 external wheels are at 350 KHz, while MB1 other wheels are <175 KHz (from this, the quoted factor of 2) Other stations, ie M2-MB4 are around 30 KHz at the most (hence the other factor) Probably the meaning of this sentence is not clear enough... we did not want to explicitly write all numbers while still keeping some quantitative statement. We could change this sentence by simply "Rates elsewhere are at least a factor of 2 smaller, with rates for most of the chambers being one order of magnitude below the MB1 external wheels". Do you have other suggestions?

Agreed on "Rates elsewhere are at least a factor of 2 smaller, with rates for most of the chambers being one order of magnitude below the MB1 external wheels"

later: "the estimated rate reduction is about factor of 3 " --> "... is about a factor 3"

Done

"For MB1 external wheels" --> "For MB1 of external wheels"

Done

p.5 fig.5 bottom the y axis title says "resolution" but in the text it was clarified that this is not to be considered as resolution

This is in line with previously approved plots here where the full description was given in the caption. In any case, if you have specific suggestions we can consider them for the next iteration (ex: "Bending Phi Segment - TP)?

Plot redone

par.4 "the performance of the system with real data inside the prototyping boards" --> "the performance of the system with real data produced by the prototype boards"

Done

I would take away "and considerations of resource usage" ("practical constrains" is probably enough)

Done

"confirmed qualities, r-z view or super-primitive building" --> "confirmed qualities, r-z view and super-primitive building"

Done

I don't understand: "modifying the corresponding parameters for each chamber accordingly"

In order to asses" --> "assess"

Done

"..and avoid any effect related to late arrival of hits or ...buffers overflow at this stage. " --> "... and avoid, at this stage, any effect related to late arriving hits or... buffer overflows."

Done

p.6 "For any given primitive in output from the emulator" --> "For any given primitive output by the emulator"

Done

Sometime you could drop "software" from "software emulator".

Affter review by MUPC we prefer leaving it as is

par.5 "The time digitisation performed at the OBDT" --> "the time digitisation performed by the OBDT"

Done

p.7 "and is referenced to the Bunch Crossing 0" --> "and is referred..." ??

After review by MUPC we prefer leaving it as is

Fig.10 caption: "Difference he AM TP.... and..." --> "Difference between the AM TP ... and..."

Done

THANK YOU FOR ACKNOWLEDGING ME !!

OF COURSE!!

Comments to figures and captions:

Fig 2, caption-> superscript in cm-2s-1

Done

Fig 4: here we have the old issue about propagation of signals along the wire. What is done here for segments?

We think this is in for segments, but we are using your code, so probably you know better!

Decided to exchange this figure with the TP time distribution, w/o making the difference wrt offline segment.

Fig 6: [mrad]-> (mrad) Remove space and change brackets to round.

To be done

Done

The trend in upper plot must be explained (phi resolution improving from MB1 to MB4)

Added: The observed trend on phi resolution is primarily given by the increasing radial position of the stations from MB1 to MB4. For a given local position resolution in a chamber, the radial position of the station determines de angular variable resolution.

Comments to text:

Abstract:

- high luminosity-> high-luminosity

Done

- occupancy and trigger rates-> latency as well, right?

Done "Due to the increase of occupancy and trigger latency and rates."

- trigger primitives -> define for first time

Rephrased as "The new backend system will be in charge of building the trigger primitives of each chamber out of this asynchronous stream of hits. These trigger primitives contain the information at the chamber level about the potential muon candidate’s position, direction, and collision time, and are used as input in the L1 CMS trigger. The resolutions achieved for the trigger primitive parameters are comparable to the ones of the current offline reconstruction." Removed the definition of TPs from Introduction

1. Introduction:

- steel layers-> or iron?

Steel according to this

- 'The barrel chambers are aranged...' -> Add something about GEMs, as GE1/1 are part of CMS now

Done in v2 already

- Ar/Co2 -> Ar:Co2

Done

- staggered-> half-staggered

Done

- and provides -> by providing?

Done

- at around -> up to

Done

- cm-2s-1-> superscript missig

Done

- time bin of 1 ns-> time bin of around 1 ns

Done

- exploit the chamber intrinsic time resolution -> exploit the detector intrinsic time resolution

Done

2. Description of the AM algorithm:

- fake/ghost-> maybe change ghost to duplicate? (ghost might be jargon)

Done

- depending on the number of hits used in the fit -> and permutation?

Even if we understand the suggestion we think adding 'and permutation' would make the text less clear... did not manage to think in a short and easy way to explain this further, but we hope it is clear what is meant, specially when looking at the table. We did not modify the text for the moment

Rephrased "At the end of the process, a quality code is assigned as described in the table below."

3. Simulation and Monte Carlo samples

- For the results presented here a sample containing-> add comma

Done

- scenario equivalent to 3000 fb-1-> mention safety factor?

The safety factor is mentioned already in the caption to Fig.2. Do you think it is necessary to add this explanation in the text body as well? (or where you asking something else?)

Added description in main text and slightly rephrasing caption to Fig 2 to avoid repetitions.

-This scenario is based on measurements of the detector performance -> This is a conservative scenario, based on measurements of the detector performance

Done

- After the analysis of the data-> add comma

Took Francesca's suggestion here

- previous studies [5] show that the final L1 Muon trigger efficiency is unaffected with a drop of <1% in the barrel-endcap overalp region, smaller elsewhere. -> Not clear where exactly one can see this info, add a pointer

We have modified the sentence to make it more clear. "In addition, previous studies [5] show that the final effect on L1 Muon trigger efficiency is small, dropping <1% in the barrel-endcap overlap region, and less elsewhere" For pointing to results we could refer to DN2017-04 where estimations of effects of ageing in muon reconstruction efficiency are shown (Note that muon reconstruction studies have been taken as proxy for L1 trigger studies in several contexts).

Agreed with Carlo to rephrase as: "In addition, previous studies based on Phase-1 TPs~\cite{L1TDR} show that the final effect of the ageing on L1 Muon trigger efficiency is small, dropping $< $5\% in the barrel-endcap overlap region, and less elsewhere."

3.2. Comparisons with segments and resolution

- x0(lower pannel) -> add space

Done

- TPs in Wh+1-> why Wh+1?

We used the same chamber as in previously approved studies here

- It is worth noting-> must be stressed

Done

- are possible -> are to be expected

Done

- simulated quantities have been performed-> add 'as well'

Done

- The sample used consisted on -> is this the standard sample for L1 TDR?

No, it is a privately produced sample

- The positions of simulated muon hits in the inner and outermost layers -> Is this Nicola's analysis?

No, this is an independent analysis performed by C. Carrillo, and presented at DT meeting here

3.3. Rate studies

- of the order of 0.5 MHz at 10 x 10E34 cm-2s-1-> superscript cm-2

Done

- Rates in other stations are factors of 2 to 15 smaller -> Is 15 in MB3? written like this the sentence is puzzling

Francesca brought up the same point. We replied this to her "From your plots, MB1 external wheels are at 350 KHz, while MB1 other wheels are <175 KHz (from this, the quoted factor of 2) Other stations, ie M2-MB4 are around 30 KHz at the most (hence the other factor) Probably the meaning of this sentence is not clear enough... we did not want to explicitly write all numbers while still keeping some quantitative statement. We could change this sentence by simply "Rates elsewhere are at least a factor of 2 smaller, with rates for most of the chambers being one order of magnitude below the MB1 external wheels" We are happy to get suggestions how to write this better if needed

Agreed to rephrase as "Rates elsewhere are at least a factor of 2 smaller, with rates for most of the chambers being one order of magnitude below the MB1 external wheels."

- regions more affected by radiation effects -> as expected by the hit efficiency loss

Done

4. Firmware implementation

- sharing same fitted hits with the same laterality assignment-> add 'and used in the following plots'

Done

5. Operation of the AM trigger in the CMS DT Slice Test

- (Wheel +2, Sector 12) -> (Wh+2, S12)

Done

- The time digitization performed at the OBDT provides a resolution of 0.8 ns -> this is sampling frequency, the intrinsic DT resolution is larger than that

Done, changed to "The time digitization performed at the OBDT has a time bin of 0.8 ns"

- Fig. 9 shows the 2D distribution of the Phase-2 TP qualities-> are there cuts and selections applied?

The selection on segments is similar to the one used for efficiency. Differences are that we do not ask for t0 < 15 ns , neither we request 4 hits in theta view.

Added more information on this in the text "Fig.~\ref{fig:STTPQualNhitsSeg} shows the 2D distribution of the Phase-2 TP qualities, as obtained by the AM algorithm in MB4, as a function of the number of hits associated with the offline reconstructed segment. Offline segments must have at least 4 hits in the r-$\varphi$ view. Only the segment with more associated hits, and the TP with the highest quality are considered."

- An example of the the timing performance -> An example of the time resolution performance

Prefer keeping 'timing' and not to use the word resolution, to avoid falling in your next comment...

- the time resolution -> wrt segments

Done

7. Acknowledgments

Francesca is INFN

Remove 'Preliminary' everywhere. Jaime: To be done

Francesca:

Fig.3 is not very well readable in black and white. This plot has been officially approved already as it is, but if you have specific suggestions we can consider them for the next iteration (we could for example changing markers shapes) Jaime: done

fig.5 bottom the y axis title says "resolution" but in the text it was clarified that this is not to be considered as resolution This is in line with previously approved plots here where the full description was given in the caption. In any case, if you have specific suggestions we can consider them for the next iteration (ex: "Sigma of Bending Phi TP - Bending Phi Segment )? Jaime: done

Carlo:

Fig 6: [mrad]-> (mrad) Remove space and change brackets to round Camilo: done

Jorge:

Fig 6: Change y-axis upper panel to start at zero Camilo: done