Performance and Calibration of the ATLAS Tile Calorimeter

Round 1

Reviewer 1 Report

The paper is well written.

The organization of the paper in all its parts is good and allows a quite complete understanding of the matter.

Nevertheless I think that in few parts the text is not completely clear and the authors should spend a few more words and add more information to help reader's understanding.

I'll go into the details now and review the points that need to be fixed.

1. Introduction

lines 20-29: As far as I understand the cell geometry is projective (segmentation in pseudorapidity with |eta|<1.7). Maybe you should say it clearly and I should also add what it means in terms of polar angle.

Because of this fact, I guess that the light collection scheme is quite complicated (I don't think the tails and fibers are projective). Maybe you should add a sentence like: "A suited light collection scheme is realized in order to realize projective cell granularity".

From the scheme of the Figure 1, it looks that there are 11 radial tail layers (made by tiles) and, in the particular case shown, the light of two cells of layers 7 and 9 (numbering from the innermost layer) goes to the same PM. Can you explain a bit more which are the cell layer A, B and D. Maybe, it is enough to add this information in Figure 1, putting a curly brackets that indicate the radial layers.

2.1 Cesium

line 39: you say "at the beginning of the data taking period". Although it is clear among the insiders, I would add "once a year".

line 43: Can you comment in the text or in the Figure 2 caption, the evident drop around |eta|=1.

2.2 Laser

line 46: You say that with the laser runs you also measure possible non-linearity of the PMT response. On the other hand the laser run provides a per-channel constant C_Las that is unique, does not depend on the light pulse amplitude (contrary to what happens for the CIS constants).

Do you correct for PMT non-linearity at this level? Can you be more clear?

2.4 Minimum bias system

This is my curiosity.

This method is equivalent to the Cs method and applies to few cells (the E-cells) where the Cs method is not possible. Due to the statistics, the precision is quite worse.

On the other hand, this calibration can be done almost always during the run period, not once a year as for the Cs.

Since these two calibration are the only method to measure the degradation of the entire readout chain (scintillator, WLS fiber, light collection and PMT), why not apply also this method to all cells, to monitor the degradation continuously?

3. Performances

3.1 Response to isolated muons

lines 107-108: your truncated mean is performed discarding just the 1% of the highest signal, a very low cut. In the double ratio R, is it appreciable the difference between the result with and without this cut?

I imagine that the MC events you use are very similar to the data ones, in particular concerning the muon energy. You should specify it.

Figure 6: the script below the points ...=0.992 \pm 0.005. What is the variable? It should be R.

It sounds strange to me that the ML fit on the 64 points results to be so large (yellow band)

3.2 Response to isolated hadrons

line 120: What do you mean with "Delta R = 0.2" (isolated clusters)?

Figure 7: Can you comment on the strong dependence of the E/p value on pseudorapidity of the track?

Concerning the simulation, do you understand why the E/p values are evidently below the data for |eta|>1.1. Maybe you should comment.

Final remark:

I realize that some of my comments would be clarified by a reading of your reference article.

However I believe that for the sake of completeness, at least some additional information should be provided.

Author Response

Dear reviewer,

  thanks a lot for all your comments, they are very helpful. please find below my answers.

Best regards, Tomas Davidek

• lines 20-29: As far as I understand the cell geometry is projective (segmentation in pseudorapidity with |eta|<1.7). Maybe you should say it clearly and I should also add what it means in terms of polar angle

Answer: the cell geometry is pseudo-projective, as the cells are rectangular in eta. I pointed that out in the text as well as in the Figure 1 caption.

• Because of this fact, I guess that the light collection scheme is quite complicated (I don't think the tails and fibers are projective). Maybe you should add a sentence like: "A suited light collection scheme is realized in order to realize projective cell granularity".

Answer: a statement is added to Figure 1 caption

• From the scheme of the Figure 1, it looks that there are 11 radial tail layers (made by tiles) and, in the particular case shown, the light of two cells of layers 7 and 9 (numbering from the innermost layer) goes to the same PM. Can you explain a bit more which are the cell layer A, B and D. Maybe, it is enough to add this information in Figure 1, putting a curly brackets that indicate the radial layers.

Answer: a statement is added to Figure 1 caption

• line 39: you say "at the beginning of the data taking period". Although it is clear among the insiders, I would add "once a year".

Answer: The HV equalization was performed only once at the beginning of the whole Run-2, then the changes are tracked with Cs constants. A sentence is introduced in the text to make it clear.

• line 43: Can you comment in the text or in the Figure 2 caption, the evident drop around |eta|=1.

Answer: comment added to the Figure 2 caption

• Do you correct for PMT non-linearity at this level? Can you be more clear?

Answer: no, we found that basically all PMTs have good linearity, a footnote was added to make it clear.

• This method is equivalent to the Cs method and applies to few cells (the E-cells) where the Cs method is not possible. Due to the statistics, the precision is quite worse.

  On the other hand, this calibration can be done almost always during the run period, not once a year as for the Cs.

  Since these two calibration are the only method to measure the degradation of the entire readout chain (scintillator, WLS fiber, light collection and PMT), why not apply also this method to all cells, to monitor the degradation continuously?

Answer: the min bias suffers from worse precision than Cs, moreover the results from Min Bias analysis come later after rather complex analysis, eventual correction are only available for data reprocessing. Therefore, we primarily rely on Cs.

• lines 107-108: your truncated mean is performed discarding just the 1% of the highest signal, a very low cut. In the double ratio R, is it appreciable the difference between the result with and without this cut?

Answer: the full mean is sometimes affected by fluctuations in energy loss (e.g. by delta-electrons), we found that the truncated mean is more stable

• I imagine that the MC events you use are very similar to the data ones, in particular concerning the muon energy. You should specify it.

Answer: the text says that the analysis targets muons from W decays, of course the corresponding MC events (W->mu+nu) are used for comparison.

• Figure 6: the script below the points ...=0.992 \pm 0.005. What is the variable? It should be R.

Answer: description added in the figure caption

• line 120: What do you mean with "Delta R = 0.2" (isolated clusters)?

Answer: I see, the confusion probably comes from the use of R  for the double ratio. Therefore, a footnote was added specifying the cluster size definition.

• Figure 7: Can you comment on the strong dependence of the E/p value on pseudorapidity of the track?

  Concerning the simulation, do you understand why the E/p values are evidently below the data for |eta|>1.1. Maybe you should comment.

Answer: statements were added to the text

Reviewer 2 Report

This article, as the title suggests, discusses the ATLAS tile calorimeter’s calibration and some on the performance.  In general, the article is well written and presents its topic clearly.  I recommend publishing the article.  I do, however, have a few questions and suggestions that might improve the article should the authors choose to do so.

1.   Lines 21-22, What is the \eta division between the “long barrel” and the two “extended barrels”?

2.     Lines 23-24, What are the interaction lengths of the individual sections?  Are they equal, or are they adjusted for better reconstruction? 

3.     Sec. 2.2 what is the laser pulse width and intensity?  How does it compare with the same numbers for a particle response?

4.     Lines 74-79, The discussion of the energy calibration to actual hadron energy needs to be made clear.  The main factor linking the light calibration to an electron particle energy is C_{TB}.  What was the test beam particle, and what energies?  Since ultimately, this is a hadronic calorimeter, how is the electron response linked to the hadronic response? Is there any test beam data, or is it just done as described in Sec. 3.2, where the response to isolated hadrons is discussed?  The E_{observed}/p ratio of Fig. 7 is not near 1, which is explained as the calorimeter not being compensating.  Thus, the labels should specifically note “observed” or measured, as it is not the actual hadron’s energy.  Otherwise, one could naively interrupt this plot as measuring \beta = 0.5-0.7 particles.

5.     Figure 6, define \alpha_{BC2}, and S_{BC2}.

Copy editing comments:

6.     Line 67, not a new paragraph.

7.     Figure 5 left, there is a lot of white space in the figure.  The caption does not explain the difference in the bunch crossing timing.  Is that the line near 26 ns?

8.     Figure 5 right.  The caption needs to tell me at which 6 channels I should be looking.  Are those the EBC09 Ch 30-35?  I count 4 channels, not 6?

Author Response

Dear reviewer,

   thanks a lot for all your comments, they are very helpful. Please find below the answers.

Best regards, Tomas Davidek

• Lines 21-22, What is the \eta division between the “long barrel” and the two “extended barrels”?

Answer: since the cells are pseudo-projective in eta  (now stated in the updated version), the boundary between long barrel and extended barrel depends on the radius. For instance, the boundary at the inner radius is eta=1.0, while at outer radius it is eta~0.7. I believe people interested in these details can find them in the references.

• Lines 23-24, What are the interaction lengths of the individual sections? Are they equal, or are they adjusted for better reconstruction? 

Answer: individual thicknesses are now added in the text

• Sec. 2.2 what is the laser pulse width and intensity? How does it compare with the same numbers for a particle response?

Answer: the laser intensity can be tuned and it spans basically the whole dynamic range of the PMTs. The laser pulse shape is very similar to that of collision signal, I added a footnote pointing that out.

• Lines 74-79, The discussion of the energy calibration to actual hadron energy needs to be made clear. The main factor linking the light calibration to an electron particle energy is C_{TB}. What was the test beam particle, and what energies? Since ultimately, this is a hadronic calorimeter, how is the electron response linked to the hadronic response? Is there any test beam data, or is it just done as described in Sec. 3.2, where the response to isolated hadrons is discussed? The E_{observed}/p ratio of Fig. 7 is not near 1, which is explained as the calorimeter not being compensating. Thus, the labels should specifically note “observed” or measured, as it is not the actual hadron’s energy. Otherwise, one could naively interrupt this plot as measuring \beta = 0.5-0.7 particles.

Answer: the calorimeter is calibrated at the EM scale, and all results presented in this proceedings are at this scale. I have added a sentence in the corresponding section to make it clear. Analyses dealing with jets of course use calibrated jets, but their calibration is pretty complicated and beyond the scope of this proceedings.

• Figure 6, define \alpha_{BC2}, and S_{BC2}.

Answer: done

• Figure 5 left, there is a lot of white space in the figure. The caption does not explain the difference in the bunch crossing timing. Is that the line near 26 ns?

Answer: yes, the bunch-crossing offset corresponds to 25 ns, added in the figure caption to make it clear

• Figure 5 right. The caption needs to tell me at which 6 channels I should be looking. Are those the EBC09 Ch 30-35? I count 4 channels, not 6?

Answer: done

Reviewer 3 Report

The referee congratulates with the author for the very interesting results, well presented throughout the text. The paper on the performance and calibration of the ATLAS Tile Calorimeter reads nicely, the results are of high quality and clearly presented. I have suggested in few places some modifications to hopefully help the readers to appreciate it even more. In the annotated pdf there are some English language suggestions but also some more relevant questions and comments that may be useful to address.

Comments for author File: Comments.pdf

Author Response

Dear reviewer,

  thanks a lot for all your comments, they are very helpful. Basically, all your language/style suggestions are applied in the new version, below I comment on the other points you have raised.

Best regards, Tomas Davidek

• It is not clear how the Minimum bias system can be used for calibration unless you specify (I guess) that the minbias signal is symmetric in azimuth, right?

Answer: yes, this is exactly the point, I made it clear in the text

• what is meant by optics? I imagine the whole system of scintillator+wavelength shifters+light propagation?

Answer: basically yes; it is now explicitly specified in the introduction

• What is the precision of the time calibration?

Answer: I added the statement on the time calibration stability, it is better than 1 ns

• Section 3.2: The response in data seems consistently lower than MC for |eta| < 1.5 and the difference seems not covered by systematic uncertainties: perhaps this may be explained/commented explicitly in the text

Answer: comment added to the text

• For clarity, the Thin Gap Chambers deserve a succint description

Answer: I put there a reference to the ATLAS detector paper, I think the description of a component of another sub-system is beyond the scope of this proceedings

• Section 3.4: the inefficiency obtained with Z->mu+mu should be explained in few words

Answer: the point is that there are thin gaps between individual Tilecal modules in azimuth, by construction. Therefore, the if a muon passes through the boundary between two modules, it might not be detected in Tilecal. I've added few words to make it clearer.

• The points in figure 8 plots seem without error bars: are those included?

Answer: the symbols are larger than the error bars, a statement was added in the figure caption.