Spsc protodune-sp preliminary tdr review

2.9.1. The DAQ must reduce the data by a significant fraction before sending them to offline. What is the needed reduction factor? How do you get it? The TDR should be more explicit about this crucial point.
Data reduction will be achieved via two main mechanisms. On one hand the raw data of the TPC will be reduced through lossless compression. The compression factor will depend on the noise level, but we are assume a conservative factor of four. In addition, we have the flexibility to prescale the trigger rate to fit within the RCE output and storage capacity. ProtoDUNE-SP is aiming at recording data at 25 Hz during the SPS spill only, thus reducing the continuous readout rate by a factor ~20. Last but not least, a transient storage system will allow to decouple the output capacity of 20 gbps from peaks in the DAQ storage rates.
2.9.6. What are the adequate resources needed to acquire and store data from the PDS and the beam instrumentation? The description in the TDR is vague on these points.
PDS data is part of the DAQ data stream together with TPC data (while beam instrumentation data and CRT data are separate data streams). PDS data represents a negligible fraction of the TPC data in the main DAQ data stream, provided only short waveform data are read-out on an external trigger. In specific calibration runs, longer waveform data may be recorded, thus increasing the data throughput. Enough provisioning will be made to be able to accommodate up to 20 times more data than currently estimated during physics data taking.
Table 2.8. 16 weeks at 50% DAQ efficiency ==> 56 days. Average data rate 576 Mbyte/s * 56 days = 3 PByte + Monte Carlo + reconstructed data … this corresponds to about 10% of the annual data volume of all 4 LHC experiments. The TDR should specify what computing provisions have been made to handle this data.
Beam data are collected during the two 4.8s spills every 48s SPS supercycle (and not continuously). The computing provision is being discussed by the dedicated CERN/Fermilab/DUNE interface group, which consists of stakeholders from protoDUNE-SP, protoDUNE-DP, DUNE computing, CERN IT and the FNAL computing sector.

2.10 Cryostat and feedthroughs
It is not clear in the TDR whether the design of the LAr recirculation cryopump is a common item between ProtoDUNE-SP and DP. The referees note that minor but time-consuming problems were experienced on this item by WA105 in the 1x3x3 m3 prototype. What is the schedule contingency for this in the installation schedule?
The design of the LAr recirculation pumps for protoDUNE SP and protoDUNE DP is identical. It is however different than the one of the WA105 1x1x3 prototype. In the case of protoDUNE SP/DP the LAr pumps are located outside the cryostat, in a dedicated valve box (each cryostat is equipped with identical but totally independent LAr pump and recirculation circuit). In the case of the WA105 1x1x3 the LAr pump is located inside a pump tower located in the cryostat itself. There was a problem with the foot valve and the pump tower. These items are specific to the WA105 1x1x3 installation and are not present in the protoDUNE installations. At this point in time, the protoDUNE-SP cryogenic system remains on schedule for being ready for beam operation in July 2018.
2.10 and 2.12 One of the most time consuming operation in cryostat design is the validation of structural and seismic response and the preparation of a risk analysis in case of major failures (LAr boiling). This documentation must be compliant to EU rules and approved by the CERN GLIMOS. What is the status of this task?
The cryostat is a responsibility of the CERN Neutrino Platform cryostat group.

2.11 Installation
A detailed schedule should be provided before the annual review.
The detailed schedule is provided in a separate document
Is cleaning after transportation foreseen for the CPA panels? CPA manipulations to assemble the CPA column may compromise the cleanliness of the plane and impact on the LAr purity (or at least on purification time).
After transportation to CERN, the components will be manipulated inside the clean room, with all measures to maintain the cleanliness of the received parts. It is worth noting that most of the impurities affecting electron drift in LAr come from outgassing from the warm surfaces in gas ullage, which then diffuse into the LAr volume. Outgassing from the cold surfaces immersed in the LAr, such as the CPA planes, is strongly suppressed at cryogenic temperatures.

2.12 Cryogenics and LAr purification systems
2.10 and 2.12 One of the most time consuming operations in cryostat design is the validation of structural and seismic response and the preparation of a risk analysis in case of major failures (LAr boiling). This documentation must be compliant to EU rules and approved by the CERN GLIMOS. What is the status of this task?
The cryostat is a responsibility of the CERN Neutrino Platform cryostat group.


3 Software and Computing
3.6.2. What is the justification for having many reconstruction efforts running in parallel? What kind of coordination exists among them all? What are the plans (if any) to end up with a single unified reconstruction / simulation framework for DUNE?
This reflects the current state of the art in LArTPC reconstruction. For example, a number of complementary paths are being pursued by MicroBooNE experiment and at this stage it not clear which is the optimal solution, or even if there is a single solution that is good for all applications. However, all existing efforts are brought together under the single unified framework provided by LArSoft. Both PMA and Pandora are being developed within this framework. Furthermore, the developments for protoDUNE-SP build strongly on those being pursued for DUNE and for MicroBooNE. At this point, the framework is fixed, but different algorithms are being investigated.


Figure 3.3. there are quite long tails in the reconstruction of single muons. Do these correspond to high-momentum muons? Are there plans to improve this?
Just as for all LArTPC reconstruction software, the Pandora reconstruction package is under constant development. Substantial improvements have been made and developments to address issues specific to protoDUNE-SP represent ongoing work.
3.6.2. How small is the “small inefficiency to match segments from different APA volumes”?
We do not believe that this is an inherent problem, the comment referred to a snapshot of the reconstruction at the time of writing the TDR.
3.6.2. What is the progress on reconstructing electromagnetic & hadronic showers in 2D & 3D?
Separation of hadronic and electromagnetic component and identification of the primary interaction vertex and of daughter particles are the primary goals in current hadronic shower reconstruction efforts by the protoDUNE offline team. For EM shower reconstruction, the main goals are the identification and collection of single shower fragments, separation of multiple showers and identification of the photon conversion point.
There has been a great deal of recent progress in EM shower reconstruction. In parallel to the mainstream approach based on development of pattern recognition algorithms, a new modern machine-learning approach (based specifically on Convolutional Neural Networks - CNN) has been developed and fully embedded into the LArSoft package. The CNN has been shown to effectively recognize and separate EM-like vs. track-like activity in protoDUNE-SP MC events. The figure below, shows a protoDUNE-SP MC event with a beam electron overlaid with cosmic-ray muons. The corresponding CNN output, shows the recognized EM-like component in the event (blue) vs the track-like component (red).



4 Space and infrastructure
What space is required for Q/A at CERN?
Details are given in the report on QC/QA
Has planning taken into account joint use of space with WA105?
As far as we are aware there are no conflicts in the use of space in the EHN1 hall. The organisation of the overall logistics is being coordinated by the Neutrino Platform.

Is a major re-cleaning of parts planned for?
There are no major (re)cleaning procedures for parts upon receiving at CERN. Detector components are expected to be shipped clean and protected in special boxes and/or bags or envelopes. When required, parts in their container will be moved through the SAS into the Clean Room and the connected Cryostat for assembly. Components will be manipulated inside the clean room, with all measures to maintain the cleanliness of the received parts.

5 Test beam specifications
5.1 What is the minimum momentum of the beamline? What is the rep rate?
The numbers are given in Tables 1.1, 1.2 and 5.3.

5.1. Inputs from the physics group led to the choice of an injection angle for the beam of about 13 degrees. Specify these inputs. What are the key parameters used to optimize for beam configuration #3?
The layout of the H4 beam line and the position of the cryostat were almost entirely determined by space constraints at the EHN1-extension. Three possible beam configurations are available, by steering the beam toward corresponding beam injection points on the front face of the cryostat:

1. 
   Config#1 pointing to the left drift volume of the TPC (Jura side);
   
2. 
   Config#2 pointing to the middle with the beam direction crossing the central cathode;
   
3. 
   Config#3 pointing to the right drift volume (Saleve side).
   

Config#1 and #3 are designed for beam events collection and reconstruction with maximum containment of electromagnetic showers and hadron cascades on either side of the TPC. Config#2 was conceived for specific studies (effects of cathode material on crossing particle and energy loss useful for neutrino event reconstruction). Track reconstruction algorithms tend to loose efficiency for tracks lying in planes parallel to the TPC wire-planes (which result in ambiguities in 2D to 3D track matching for hits with same drift-coordinate. Config#3 has a 12 degree angle of incidence with respect to the wire planes of the right-side TPC, larger than in Config#2. Config#3 was thus selected as the primary beam injection point with: full instrumentation, a beam window through the insulation, a beam plug inside cryostat to minimize the amount of material along the beam direction, upstream the TPC active volume.

5.2.2. For momenta above 3 GeV/c, the particle rates are above the DAQ capabilities. What is the plan for dealing with this?
The DAQ trigger rate is set at 25 Hz. This comes as reasonable compromise between increasing technical complexity (and related costs), expected available beam particle rate at H4 and target size of the beam event sample to be collected in the run of ProtoDUNE-SP. The beam trigger rate will be prescaled if it exceeds the maximum sustainable DAQ rate. The Run plan (Table 1.1) is based on this 25 Hz DAQ trigger rate assumption.



5.2.2 Since the secondary production target is 37m from cryostat, what is the expected muon beam halo rate and spectrum from this target (not the primary considered in 5.2.3, which is ~600m away)? From not only the H4 beamline but nearby? What are the corresponding versions of Table 5.2 and 5.3 for all particles impinging on the cryostat, not just the R<10 cm beam plug?
The fluence of punch-through particles at the detector front side will depend on the shielding (thickness and position) around and at the end of the H4 tertiary beamline. Radiation shielding design assessment is currently under development through detailed MC simulations (CERN RP and Neutrino Platform).
5.2.2. Up to 250 beam particles per spill are shown in Table 5.3. What it the impact of this pile-up on the systematics measurements aimed at here?
The highest rate (250 Hz) is expected for the 7 GeV/c (positive polarity) beam setting. Even at this highest rate, the probability of pile-up of two or more events in the same 2ms drift frame is 9%. For all other settings the probability is <6% and it is negligible for the lower momentum settings. Given the relatively low rates, recorded events with pile-up of two or more beam events will be discarded at off-line data analysis level.
What is the number of cosmic rays incident on the cryostat per spill-- is the whole 4.8s the spill time for beam incident on the production target for ProtoDUNE-SP?
A cosmic rate of 10 kHz is assumed. This is based on standard cosmic muon rate calculations. More accurate estimate, based on detailed MC simulations, are being developed. The 10 kHz rate corresponds to an average of 20 muons per 2 ms drift time (which is the relevant timescale here).
At what beam momentum setting is the DAQ operating range exceeded when the cosmic ray background is included? The text implies the maximum DAQ rate is ~200 particles per spill (at 4 GeV/c of beam particles).
ProtoDUNE-SP will operate in triggered mode. During the spill (4.8 s) beam event triggers are generated by the coincidence of beam counters along the beam line. Beam triggers will drive the DAQ up to the max sustainable rate (25 Hz), and will be prescaled when the trigger rate exceed the DAQ rate, as expected at the higher beam momentum settings. Cosmic-ray muon triggers can also be generated by the cosmic muon-tagger (CRT), either during the spill or in the period between spills. The CRT will allow to select specific classes of cosmic muon triggers, e.g. long muon tracks crossing the entire detector at large zenith angles. In all cases the trigger rate will be throttle to the level for sustainable DAQ operation.

5.2.3. Why is the muon halo study done with muons only >4 GeV/c? Presumably the muon detection threshold in ProtoDUNE-SP is far below this value.
The muon halo refers to muons from the decay in flight of 80 GeV pions of the secondary pion beam at EHN1 (which originates from protons on the primary target 600 m upstream). The majority of these halo muons are high energy muons. The fraction below 4GeV/c are assumed to be removed by shielding blocks to be located upstream the detector.
5.2.3. Given that the primary beam muon halo rate is ~400 muons/m2/spill incident on the cryostat, what is the pile-up rate from this source? Given the spill length of 4.8s it is not at all clear that the expected rate in time is only a few/m2/spill. Re: “The rate of beam events with out-of-time halo muon tracks captured in the 2.5 ms time frame of the recorded event is expected to be limited” (If I integrate the red square in Fig. 5.5 by eye I get a number of order 100).
The muon intensity is not a flat 400µ/m2/spill. In the protoDUNE-SP TDR the highest flux of 400 µ/m2/spill is only expected in the upper left corner of the cryostat (as shown in Fig.5.5). The integrated rate of halo muons impinging on the front side of the TPC active volume (red rectangle in Figure 5.5) is expected to be significantly lower. For example, from Figure 5.5, the estimated total rate (in spill) of halo muons crossing the TPC drift volume on the right (Saleve side) is expected to be 12 Hz over this surface. This is the relevant region for the beam (Config#3). The cumulative probability of pile-up one or more muons in the 2ms frame of a triggered event is rather small ~ 2.5%.

5.2.3 The study described here does not support the conclusion that the beam halo rate is negligible. Re: “Figure 5.5 is very preliminary though, not only because of low statistics, but also because shielding around the low-energy beamline is not included in the simulation, and the muons produced in neighboring beamlines in EHN1, including the H2 beamline that feeds ProtoDUNE-DP, are not considered here. Based on these results, the estimated contribution to the total data volume from beam halo is negligible.” The impact on the DAQ could potentially be severe; this study should be updated with the shielding, the H2 beamline, and muons of all detectable momenta.
The shielding, residual beam halo and punch through particle spectra and fluences at the detector front face are all currently subject of extensive ongoing MC studies by CERN-RP, the Neutrino Platform and ProtoDUNE-SP.
What number of events is needed to constrain the systematics and/or validate the reconstruction? Putting all numbers together (Table 5.3 combined with the discussion following Fig. 5.10), with 100% efficiency I get that the number of protons per spill at 1 GeV/c is 0-1, with no proton flux below 1 GeV/c. What beam exposure is required to meet the physics goals of the beam test?
Table 1.1 gives the relevant numbers. The assume beam plan will give a sample of 700k 1 GeV/c protons. This is considered sufficient for the physics goals of the beam test. Note, that to record this relatively large sample, the 1GeV/c beam setting is significantly longer than the other momentum settings.

Clarification Comments:
1.3.2. What charge-sign determination methods are referred to here? Re: “Samples of stopping muons with Michel electrons from muon decay (or without them, in the case of negative muon capture) will be used for energy calibrations in the low-energy range of the SN neutrino events and for the development of charge-sign determination methods”
mu+ decay at rest 100% into e+, mu- are instead ~75% captured in Ar and ~25% decay into Michel e-. Capture is followed by low energy gamma and p/n emission. Studies of the activity at the end of a recorded stopping muon track can allow to determine the sign of the muon with some statistical confidence, based on the recognition of a Michel electron or of some product of the capture process.
1.4 compare p and pi at same momenta. Why 1 and 2 GeV/c for p but 0.4 and 0.7 GeV/c for pi? Text discusses momenta, but Fig 1.2 show KE. These should be consistent.
The figure in question shows the distribution of the kinetic energy at the interaction point for a samples of p and pi with different incident momenta. For example, a proton of 1 GeV/c interacting immediately after entering the LAr volume has kinetic energy of just 0.43 GeV/c2 (equivalent to 1GeV/c momentum). A proton which traverses a layer of LAr before interacting loses energy and its kinetic energy at the interaction point is lowered.
2.2.1. What is the electronics sampling uncertainty? Re: “In practice, the resolution on the drift-coordinate (x) of a vertex or hit will be better than that on its location in the y-z plane, due to the combination of drift-velocity and electronics sampling-rate uncertainties.”
Each wire in APA corresponds to a Cold Electrons read-out channel. Signal digitization is performed every 500 ns (2 MHz sampling rate). The drift velocity (x coordinate) is 1.6mm/us at the nominal Electric Field of 500 V/cm. The granularity along x coordinate is thus 0.8mm. In the other coordinates the granularity is determined by the wire-to-wire distance (wire pitch) of 5mm for all the wire planes of the TPC. In the 3D reconstruction, the finer granularity allows for a better accuracy in the x (drift) coordinate. The resolution (sig_x) primarily depends on the uncertainty on the electron drift velocity in LAr. The uncertainty on the drift time, depending on the sampling rate, also contributes to the resolution in the determination of the drift coordinate.
2.2.3. How is the electron deflection arranged? Re: “An electron deflection technique is used to ensure that electrons drawn towards a joint between two APAs will be deflected to one or the other, and not lost.”

The electron deflection technique uses one or two conductive strips placed over and above the gap between two APAs, and bias the strips more negatively than the natural location potential to push the incoming electrons away from this gap. This is the same principle that allows the electrons to move around the induction plane wires. The deflector is implemented as a row of FR4 boards with copper strips on their edges. Although one deflector board was installed in the 35 t TPC, we did not record sufficient data to evaluate its performance.

2.2.5. How is the angle of the combs with respect the wire planes maintained? Has this been verified after thermal cycling?
The wire support combs are glued on the APA frame using a G10 strip to serve as a right angle bracket. Custom fixtures are used to keep the stack of combs perpendicular with respect to the wire planes. This structure has been used in prior two generations of APA prototypes. Multiple thermal cycles were performed with the APAs used in the 35 t TPC and no issues were observed.
2.4.5. How does attaching the FC to the cathode planes prevent damage to the APAs during assembly? Re: “On the top and bottom of the TPC, hinges connect each FC module to two CPA columns. These hinges allow the FC modules to be pre-attached to the CPAs during installation, preventing accidental damage to the APA wire planes when FC modules are raised and connected to the APAs.”
The distance between the APA row and CPA row is fixed by the installation rails, and the field cage modules are supposed to fill the gap between the two walls with tight tolerance. If we lift the top or bottom field cage modules freely between the APA and CPA rows with the APAs already installed, there is great risk of the FC module bump into the wires and cause damage. By fixing the field cage modules on one end to the CPAs, it is much less likely for the FC modules to have accidental lateral movement to damage the APAs.
5.2. What is the partial beam window mentioned here: “The remaining two beam positions will have partial installation of the beam window system.”
There is only one “Beam Window” on the front face of the cryostat, at the beam injection point for beam Config#3. A hole is made on the external 1cm thick Stainless Steel wall of the cryostat and the first thick layer of polyurethane foam is removed to allow the vacuum beam pipe to penetrate up to the first membrane of the cryostat. For the beam Config#1 (and #2) there is no beam window, however a hole is made in the external Stainless Steel wall of the cryostat, sealed by the end-cap flange of the beam pipe.
11. It is not clear in the TDR whether the tests currently performed in Ash River are an installation demonstration procedure for DUNE only of for ProtoDUNE-SP, as well.
The tests currently performed in Ash River were specifically for the installation procedure for ProtoDUNE-SP.

We could not find any reference on the external muon tagger (to be built with the Double Chooz scintillators). This should be clarified in the TDR.
Now we have a concrete design, a short section dedicated to the external muon tagger will be added to the TDR