Referee's report on CALICE proposal =================================== I) Introduction --------------- A list of questions was sent to the proponents, who supplied an answer in writing by 17/06/02. The referee met with two of the proponents (D. Ward and M. Thomson) on 18/06/02 for about two hours. The answers given by the proponents are attached as a separate PDF file. The report below first goes through the various questions one by one and then concludes with a summary and a set of recommendations. II) Comments on answers to questions ------------------------------------ a) Physics and performance Question 1: ---------- It is claimed that the top quark mass can be measured to a precision of 200 MeV from an energy scan, yet the average energy lost to ISR is 25GeV at 500GeV, as well as beamstrahlung effects. How is the accuracy of the top quark measurement consistent with the uncertainty from the latter effects? Comments from referee: none. --------------------- Question 2: ---------- Comparison to LEP hadronic calorimeters is really very naive: this should rather be done to HERA or LHC-type calorimeters, for which resolutions of 30-40%/sqrt(E) are obtained at lower extrapolated cost. What would the gains of the energy-flow approach be in this case? Comments from referee: --------------------- Upon request from the proponents this question was clarified (see below): > 1) Question 2 refers to "resolutions of 30-40%/sqrt(E)" for > HERA and LHC detectors. We assume this is the hadronic jet > resolution, as that is what drives the LC physics case for > the calorimeter we are studying. However, while some of > these detectors achieve this value for single hadron resolutions, > none are within this range for jets. The best we are aware > of is ATLAS, which has a sqrt(E) term of ~50% for jets. Can > you clarify whether this is a single hadron or jet resolution > and provide references where the 30-40% values can be found? Yes, you are right: I got somewhat annoyed with the CALICE proposal which had as only hadron calorimetry reference the LEP calorimeters when there are large-scale calorimeters such as those of H1 and ZEUS operating since many years and LHC calorimeters in construction with test-beam performances for single hadrons much better than the LEP calorimeters which were not really full-fledged hadron calorimeters. So, the question should be rephrased in the following way: What would the gains of the energy-flow approach be with respect to more classical hadronic calorimeters optimised for jet energy resolution based on the calorimetric measurements alone? In a sense, I would like to see new less idealised plots of the W/Z separation for example, in several cases: a) a classical LHC-like hadronic calorimeter (ATLAS is indeed the best-performing one on paper). For example, Figure 1 of your proposal states in the TESLA TDR that the sigma is 6.2 GeV at LEP. It would be 4.5 GeV for ATLAS and I would be interested to know what the ZEUS projection would be. b) the CALICE proposal for its nominal granularity c) the CALICE proposal with reduced EM and/or hadronic granularity to see the impact of cost versus performance. There are several aspects here: 1) My own education as someone who has already refereed CALICE one year ago, at a time when the simulation plots were very idealised. Let me be specific: the W-Z separation plot for nunuWW versus nunuZZ events with four jets and missing energy (figure 4 of your proposal) was done in the TESLA TDR for 30%/sqrtE resolution and ideal energy flow, i.e. not energy flow applied to real GEANT-simulated showers but to a particle-level simulation with 30%/sqrtE energy resolution assumed. Also these plots apply only to the barrel calo, and are hence optimistic at least in the sense that jets in the end-caps will be more collimated. I would personally hope that after one year further work which was recommended that CALICE pursue and which they indeed wanted to pursue, there would exist more detailed plots than this figure 4 which is very crude. I would like to see sigma(Ejj)/Ejj versus Ejj for the pairs of jets used to form the IVB candidates. 2) Technological implications of reduced requirements - does one need silicon or would scintillator suffice? The issue here is to get a significant mip measurement in each layer which might be possible with scintillator but for reduced granularity. - how to minimise the cost of the silicon? --> go from 40 to 30 layers? Only minus is 15%/sqrtE for em showers. Is this a problem? --> gang pads on opposite sides of W plates together? --> go to larger pads? No real evidence has been shown to-date that 1x1 cm2 is needed for physics. - does HADCAL really require 6 10**7 channels in the digital option? Would larger pads be ok? 3) I believe that you have to convince the PPRP members that you are already now putting in significant effort into the software effort in order to establish the UK groups as a major partner in this proto-collaboration. They were not convinced at the May meeting, hence the questions about the human resources devoted to the project (a few senior persons at more than 10-20% of their time are indeed required from now). A few additional comments are given below: * The ATLAS mass resolution for W/Z to jj decays with kinematics similar to those considered by TESLA TDR and the proponents (see Figure 4 of the UK Calice proposal) is rather 5 GeV than 8 GeV. * The referee does not agree that electromagnetic resolution is important for CALICE too: - the H to gamma-gamma signal can be easily extracted at TESLA with a much more standard EM resolution of 15-20%/sqrt(E). There is in fact remarkably little discussion of the electromagnetic resolution in the TESLA TDR. - lepton tagging for b-jet identification is indeed an important tool, but its performance depends very little on the EM resolution, since this is a particle identification issue and not a measurement issue. In addition, the vertexing performance proposed for TESLA physics is significantly better than anything achieved nor planned to-date, so it ppears likely that b-tagging through vertexing will really be the dominant tool by far at TESLA. * The resolution on the visible energy of 3 GeV or so obtained for the TESLA TDR studies (see Figure 1 of the UK CALICE proposal) can be decomposed as follows: - 1.6 GeV from photons (and electrons), a term which is dominated by the resolution on photons with energy below 1 GeV or so and overlapping partially with charged hadrons; - 0.6 GeV from charged hadrons (momentum measurement is used); - 2.4 GeV from neutral hadrons (mostly K0's), which account for 10% of the total visible energy, i.e. about 9 GeV. For this component, the resolution is thus only about 80%/sqrt(E). The above numbers appear to indicate that the accent should be put on the hadron calorimeter first and foremost and, to a lesser extent, on an optimisation of the electromagnetic calorimeter to understand the sensitivity of the photon component of the energy resolution to the main parameters and features of the electromagnetic calorimeter. Question 3: ---------- Regarding the overall simulation results, what is gained by having 30 10**6 EM cells rather than 15 or 7 10**6, e.g. by decreasing the sampling frequency which would presumably affect only the EM resolution and/or e.g. by ganging together pads on opposite sides of an absorber layer? Comments from referee: --------------------- * One could seriously put into doubt the claim that the overall readout electronics is only 8% of the total cost of the CALICE calorimeter. However, the UK proponents correctly point out that the results from this first testbeam prototype would in fact be the starting point towards a real optimisation of the CALICE calorimeter. The prototype will be constructed as much as feasible to provide real insight from the data into the issues raised by the very high cost of the initial TESLA proposal. Question 4: ---------- The CALICE calorimeter approach is a global one with little or no emphasis on EM calorimetry with respect to overall calorimetry. At this stage, a lot of effort has already gone into the EM calo conceptual design and simulations and there is only one option actively under study. In contrast, the hadronic calorimetry consists of two competing options, one with digital readout of small gas cells, the other with analogue readout of larger-size tiles. Why have the UK groups chosen to join the EM calo effort rather than the perhaps less populated hadronic calo effort? Comments from referee: --------------------- * The answer from the proponents if fine, since it explains their original motivation in joining this project. * It should be noted as a general area of concern that there is very little expertise in Silicon in the CALICE collaboration. The UK groups have not wished to get involved in this at the present stage (the elementary Silicon layer R&D with all of its problems, some of which are alluded to in the answer of the proponents, is essentially only a French responsibility). Questions 5 and 11: ------------------ The Panel felt it was essential that the UK should play a leading role in the simulation effort. What is the current status of the simulation effort and what are the UK plans to contribute to this in a visible way over the next months? The TDR simulation had many idealised features: it was not based on a full reconstruction of fully simulated GEANT events, but rather on an idealised reconstruction, only the barrel calorimetry had been simulated and one Expects jets to become more collimated in the end-caps, the two hadronic calorimeter solutions were using totally different tools, the choice of G4 at the present moment for performing these hadronic calorimeter simulations is extremely questionable, etc. How is the simulation work shared throughout the UK groups ? Who does what ? Who is leading the simulation effort, and how is it being co-ordinated across groups? What is the UK role in simulation within the wider collaboration - do the UK have an established role? Who in the UK will set up the simulation work based on GEANT4? Will GEANT3 and FLUKA also be considered? Comments from referee: --------------------- * The proponents will concentrate on the simulation of the testbeam prototype, where little effort had been put in to-date. This is certainly a valid approach, particularly if the intentions expressed during the discussion with the proponents can indeed become a reality, namely the comparison of different physics packages available for hadronic showers in dense calorimeters. * The proponents however clearly underestimate the magnitude of the task and the short time available: - Geant4 does not have a working microscopic hadron physics package today. It is not clear when such a package will realistically be available. The opinion expressed that the only viable approach today is Geant4 is characteristic of how difficult it is for anyone to assess the status of software projects as opposed to hardware. The only experiment today which runs Geant4 in production is BaBar, and this is only because BaBar is not using this tool for hadronic physics issues. CDF and D0 are Geant3-based, even though their reconstruction software is OO/C++. The same is true for e.g. ATLAS. - Geant3 has an obsolete version of Fluka with well-known defects and limitations. - Fluka stand-alone is possibly the best package on the market at present (it is certainly the one used by the LHC experiments for their radiation calculations), but it is hard to get to the source code. - Integration of Geant4 and Fluka has been discussed since many years, with little progress to show because of various political issues ... - The energy flow studies required for a realistic assessment of the CALICE approach cannot be realised without a detailed and flexible simulation package. This will take more than one year for a few people to develop and a structured software approach would already be very useful at this stage. There are many issues related to this: C++ and Fortran living together, how to build Geant3, Geant4 and Fluka geometries from the same input data, etc... * It would therefore be very desirable to see very soon a more detailed workplan for the simulation work and for the testbeam measurements themselves, since they are an essential input ingredient into the simulation workplan. It is not clear today where the prototype will be measured, nor which basic beam types and beam energies are considered to be the most important. Simulation results from single hadrons and perhaps from "jets" emulated from boosted interactions in a thin target will be needed soon to guide these choices. * The level of effort available appears marginal today and this issue is discussed further in Question 10. Question 6: ---------- Are there competing technologies to Si-W, for example compensating calorimeters? What is the scope of other LC calorimeter studies throughout the world ? Where would the UK stand if Si-W was not the final chosen option? Comments from referee: none. --------------------- Question 7: ---------- Is there an issue of radiation damage to the silicon? Will the possibility of running cheaper lower-grade silicon at cryogenic temperatures be considered? Comments from referee: --------------------- * The radiation damage issues for the conceptual TESLA detector have been assessed reasonably well at this stage, including beam-induced backgrounds. The levels arising in the calorimeter itself are indeed several orders of magnitude below those expected at the LHC. Even taking into account the uncertainties on the calculations at this early stage, there appears to be no real motivation for considering the extra complication of running silicon at low temperatures. Question 8: ---------- What is the status of elementary R&D on one layer of Si/W: connections, FE electronics, mechanics, signal/noise, etc? There was little experience with Si calorimetry in CALICE ~1 year ago: has this changed? Comments from referee: --------------------- * As already stated, the UK groups are not actively involved in the "technical" prototype of the CALICE collaboration and could not provide much detailed information on the recent progress in this field. * The technical details of the elementary silicon layer are not yet finalised, but the readout chip has been frozen: it will be the current version of a chip developed for OPERA. The interface between the detector and the readout electronics to be developed in the UK is thus well defined. Question 9: ---------- Regarding the prototype electronics design: will the on-board FPGAs provide intelligent processing of the data, or are they just passive? Comments from referee: none. --------------------- Question 10: ----------- How will the UK provide the strong academic leadership necessary to provide effective coordination of a relatively large number of people spending small fractions of their time on the project? This is a major concern of the Panel. Comments from referee: --------------------- * It is rather daunting to see the commitments of most of the proponents over the net few years: - Birmingham is in BaBar - Cambridge is almost finished with OPAL, but is in MINOS. - IC is in BaBar - UCL is in ATLAS and ZEUS upgrade - Manchester is in ATLAS, BaBar and H1 upgrade. * The proponents have responded to some extent to this concern, but will this be enough? - for the electronics effort, the experience of the groups and their technical expertise, combined with the relatively standard hardware to be built, tested and commissioned, appears to provide sufficient guarantees. - for the simulation effort, where the software to be used is not plug and go, where the transition to C++ may be an added burden for a while, and where deep thought and design are needed from the start, the level of effort available today is sub-critical. This aspect of the project is vital for the longer term and should be monitored in terms of its progress after one year or so. Question 12: ----------- In more detail, how is the electronics work broken down amongst institutes, including TD. Who does what ? Justify why Extra TD effort is necessary. Who is leading the electronics effort, and how is it being co-ordinated across groups? Comments from referee: none. --------------------- Question 13: ----------- The travel costs are felt to be high, in particular what are the "beam-time expenses". Comments from referee: none. --------------------- Question 14: ----------- The Panel was not convinced about the award of RAs (1.6 FTE's), and was confused about how the RA's were to be shared. How would the RAs be distributed across institutes, and why are they essential to the work? A workplan with deliverables would need to be provided. Are there RAs in post already, and are they being applied for as assurance that they might be lost in the next RG round? Or are they really new posts? Comments from referee: none. --------------------- III) Summary and recommendations -------------------------------- The proponents of Proposal 317 to the PPRP wish to join one of the most exciting R&D proposals for the Next Linear Collider. The calorimetry technique proposed for this next generation of collider experiments is quite novel, since it basically proposed to perform tracking of particles within the calorimeter itself. The proponents see their involvement with CALICE proceeding possibly in two stages: 1) First stage: 2002-2004 a) participation in the development and construction of the physics prototype to be used for data taking in testbeam in 2004. As latecomers in the collaboration, the UK proponents have accepted to produce more or less standard readout and trigger electronics for this prototype. b) direct involvement in simulation studies of this physics prototype with the goal of providing the collaboration with the necessary tools to define the measurements to be done in beam and to assess their results. 2) Second stage: 2004-onwards Based on the results of the first stage and on the evolution of the basic R&D on the electromagnetic and hadronic calorimeter technologies, the UK groups would actively participate in the real optimisation of the overall CALICE calorimeter, including aspects such as cost, cooling, physics performance, etc., which they cannot address at present given their current commitments. This strategy is probably very sound given the large uncertainties involved in the definition of the Next Linear Collider project. If the UK roadmap sees this project as an important one for UK High Energy Physics in the longer-term future, then Proposal 317 should be recommended for approval by the PPRP with a few recommendations and caveats: 1) Do the PPRP and SC agree with the emerging trend that scientists having more or less completed their activities at LEP move either towards Tevatron (already a strong component in the UK) or towards the NLC? The referee may well be biased, but there could be a real danger here to the successful realisation of the LHC programme. 2) The level of effort planned by the proponents for simulation tasks is insufficient to meet the goals of providing the collaboration with the required variety of simulation tools. The magnitude of this task is underestimated by the proponents and the progress should be reviewed in one year or so. 3) The physics studies indicate that the hadronic calorimeter performance drives the energy flow algorithm performance advertised in the TESLA TDR. The initial UK simulation effort should study this further to define the basic requirements for the beam measurements in 2004. 4) A detailed plan for the beam studies needs to be developed quite soon, both in terms of overall logistics and planning and of coordination with simulation work. 5) The UK groups should as much as feasible provide input also to the final technical choices to be made for the elementary silicon layer in the physics prototype. 6) The academic leadership of the UK collaboration should be strengthened as soon as feasible beyond what is stated in Proposal 317 and even beyond the hopes expressed in the answers to the questions of the PPRP.