Fork me on GitHub

Changeset 565 in svn


Ignore:
Timestamp:
Jun 16, 2010, 11:00:59 AM (15 years ago)
Author:
Xavier Rouby
Message:

mineur

File:
1 edited

Legend:

Unmodified
Added
Removed
  • trunk/paper/CommPhysComp/notes.tex

    r562 r565  
    4949calorimeters and a muon system, and possible very forward detectors arranged
    5050along the beamline.
    51 The framework is interfaced to standard file formats (e.g.\ Les Houches Event
    52 File or \texttt{HepMC}) and outputs observables such as isolated leptons,
     51The framework is interfaced to standard file formats and outputs observables such as isolated leptons,
    5352missing transverse energy and collection of jets which can be used for dedicated
    5453analyses. The simulation of the detector response takes into account the effect
     
    119118the various detector inefficiencies, the dead material, the imperfections and
    120119the geometrical details. Their simulation is in general performed by means of
    121 the GEANT~\citep{bib:geant} package and final observables used for analyses
     120the GEANT~\citep{bib:geant} toolkit and final observables used for analyses
    122121usually require sophisticated reconstruction algorithms.
    123122
     
    129128
    130129In this context, a new framework, called \textit{Delphes}~\citep{bib:delphes},
    131 has been developped, for a fast simulation of a general-purpose collider
    132 experiment.
     130has been developed, for a fast simulation of a general-purpose collider
     131experiment~\citep{bib:othersim}.
    133132Using this framework, observables such as cross-sections and efficiencies after
    134 event selection can be estimated for specific reactions.
     133event selection can be estimated for specific processes.
    135134Starting from the output of event generators, the simulation of the detector
    136135response takes into account the subdetector resolutions, by smearing the
     
    159158\begin{figure*}[!ht]
    160159\begin{center}
    161 \includegraphics[scale=0.78]{fig1}
     160\includegraphics[scale=0.60]{fig1}
    162161\caption{Flow chart describing the principles behind \textit{Delphes}. Event
    163162files coming from external Monte Carlo generators are read by a converter stage
     
    267266By default, a track is assumed to be reconstructed with $90\%$ probability if
    268267its transverse momentum $p_T$ is higher than $0.9~\textrm{GeV}/c$ and if its
    269 pseudorapidity $|\eta| \leq 2.5$~\citep{qr:tracks}. No smearing is currently
    270 applied on track parameters. For each track, the positions at vertex
    271 $(\eta,\phi)$ and at the entry point in the calorimeter layers
    272 $(\eta,\phi)_{calo}$ are available.
     268pseudorapidity $|\eta| \leq 2.5$~\citep{qr:tracks}. This reconstruction efficiency is assumed to be uniform and independent of the particle type. No smearing is currently applied on track parameters. For each track, the positions at vertex $(\eta,\phi)$ and at the entry point in the calorimeter layers
     269$(\eta,\phi)_{calo}$ are available while the helix parameters of tracks, especially the impact parameters are not.
    273270
    274271
     
    307304\label{eq:caloresolution}
    308305\end{equation}
    309 where $S$, $N$ and $C$ are the \textit{stochastic}, \textit{noise} and \textit{constant} terms, respectively, and $\oplus$ stands for quadratic additions~\citep{qr:energysmearing}.\\
     306where $S$, $N$ and $C$ are the \textit{stochastic}, \textit{noise} and \textit{constant} terms respectively, $E$ is in GeV and $\oplus$ stands for quadratic additions~\citep{qr:energysmearing}.\\
    310307
    311308In the default parametrisation, ECAL and HCAL are assumed to cover the
     
    350347the energy, determined according to their decay products, that would be
    351348deposited into the \textsc{ECAL} ($E_{\textsc{ECAL}}$) and into the
    352 \textsc{HCAL} ($E_{\textsc{HCAL}}$). Defining $F$ as the fraction of the energy
     349\textsc{HCAL} ($E_{\textsc{HCAL}}$). The positions of these deposits are identical to the one the mother particle would have hit.
     350  Defining $F$ as the fraction of the energy
    353351leading to a \textsc{HCAL} deposit, the two energy values are given by
    354352\begin{equation}
     
    362360where $0 \leq F \leq 1$. The resulting calorimetry energy measurement given
    363361after the application of the smearing is then $E = E_{\textsc{HCAL}} +
    364 E_{\textsc{ECAL}}$. For $K_S^0$ and $\Lambda$ hadrons, the energy fraction is
    365 $F$ is assumed to be $0.7$~\citep{qr:emhadratios}.\\
     362E_{\textsc{ECAL}}$. For $K_S^0$ and $\Lambda$ hadrons, the energy fraction $F$ is assumed to be $0.7$~\citep{qr:emhadratios}.\\
    366363
    367364
     
    399396The electron, muon and photon collections contains only the true final-state
    400397particles identified via the generator-data. In addition, these particles must
    401 pass fiducial cuts taking into account the magnetic field effects and some
    402 additional reconstruction cuts.
    403 
    404 Consequently, no fake candidates enter these collections. However, when needed,
    405 fake candidates can be added into the collections at the analysis level, when
    406 processing \textit{Delphes} output data. As effects like bremsstrahlung are not
    407 taken into account along the lepton propagation in the tracker, no clustering is
    408 needed for the electron reconstruction in \textit{Delphes}.
     398pass fiducial cuts taking into account the magnetic field effects and have a
     399transverse momentum above some threshold (default: $p_T > 10~\textrm{GeV}/c$). Consequently, no fake candidates enter these collections. As effects like bremsstrahlung are not taken into account along the lepton propagation in the tracker, no clustering is
     400needed for the electron reconstruction in \textit{Delphes}.
     401
     402%In your electron simulation you take only the energy from the electron itself, not from any physics bremsstrahlung that is often collinear with the electron (I mean the real emission bremststrahlung directly from the Z->ee decay, not from the magnetic field in the ID). This leads in Z->ee events to a visible distortion of the line shape. I am aware that there is no easy solution to this problem and just adding all photons in the same cell leads to bias for others reasons. However, the reader should be warned that the Z peak will be slightly shifted due to the missing bremsstrahlungs energy.
     403
    409404
    410405\subsubsection*{Electrons and photons}
    411 Real electron ($e^\pm$) and photon candidates are associated to the final-state
    412 collections if they fall into the acceptance of the tracking system and have a
     406Real electron ($e^\pm$) and photon candidates are identified if they fall into the acceptance of the tracking system and have a
    413407transverse momentum above some threshold (default: $p_T > 10~\textrm{GeV}/c$).
    414408\textit{Delphes} assumes a perfect
     
    454448(2) the lepton transverse momentum~\citep{qr:caloisolation}:
    455449$$ \rho_\ell = \frac{\Sigma_i E_T(i)}{p_T(\ell)}~,~ i\textrm{ in }N \times N
    456 \textrm { grid centred on }\ell.$$
    457 
     450\textrm { grid centred on }\ell.$$
     451
     452Finally it should be mentioned that because photons from physics bremsstrahlung are not added to the electron, they appear in the calorimetric isolation energy. This can lead to unrealistically high isolation energies in the case of very high energetic electrons while in a full simulation most bremsstrahlung photons overlap with the electron and do therefore not enter in the isolation energy.
    458453
    459454\subsection{Jet reconstruction}
     
    464459tools\footnote{A more detailed description of the jet algorithms is given in the
    465460User Manual, in appendix.}. Six different jet reconstruction schemes are
    466 available~\citep{bib:FASTJET,qr:jetalgo}. For all of them, the calorimetric
     461available~\citep{bib:FASTJET,qr:jetalgo}: {\it CDF Jet Clusters}~\citep{bib:jetclu}, {\it CDF MidPoint}~\citep{bib:midpoint}, {\it Seedless Infrared Safe Cone}~\citep{bib:SIScone}, {\it Longitudinally invariant $k_t$ jet}~\citep{bib:ktjet}, {\it Cambridge/Aachen jet}~\citep{bib:aachen} and {\it Anti $k_t$ jet}~\citep{bib:antikt}. For all of them, the calorimetric
    467462cells are used as inputs. Jet algorithms differ in their sensitivity to soft
    468 particles or collinear splittings, and in their computing speed performances.
    469  
    470 \subsubsection*{Cone algorithms}
    471  
    472 \begin{enumerate}
    473  
    474 \item {\it CDF Jet Clusters}~\citep{bib:jetclu}: Cone algorithm forming jets by
    475 combining cells lying within a circle (default radius $\Delta R=0.7$) in the
    476 $(\eta$, $\phi)$ space. Jets are seeded by all cells with transverse energy
    477 $E_T$ above a given threshold (default: $E_T >
    478 1~\textrm{GeV}$)~\citep{qr:jetparams}.
    479  
    480 \item {\it CDF MidPoint}~\citep{bib:midpoint}: Cone algorithm with additional
    481 ``midpoints'' (energy barycentres) in the list of seeds.
    482  
    483 \item {\it Seedless Infrared Safe Cone}~\citep{bib:SIScone}: The
    484 \textsc{SISC}one algorithm is simultaneously insensitive to additional soft
    485 particles and collinear splittings.
    486 \end{enumerate}
    487 
    488 \subsubsection*{Recombination algorithms}
    489 
    490 The next three jet algorithms rely on recombination schemes where calorimeter
    491 cell pairs are successively merged:
    492  
    493 \begin{enumerate}[start=4]
    494 \item {\it Longitudinally invariant $k_t$ jet}~\citep{bib:ktjet},
    495 \item {\it Cambridge/Aachen jet}~\citep{bib:aachen},
    496 \item {\it Anti $k_t$ jet}~\citep{bib:antikt}, where hard jets are exactly
    497 circular in the $(y,\phi)$ plane.
    498 \end{enumerate}
    499 
    500 The recombination algorithms are safe with respect to soft radiations
    501 (\textit{infrared}) and collinear splittings. Their implementations are similar
    502 except for the definition of the \textit{distances} used during the merging
    503 procedure.
    504 
    505 By default, reconstruction uses the CDF cone algorithm. Jets are stored if their
    506 transverse energy is higher than $20~\textrm{GeV}$~\citep{qr:ptcutjet}.
     463particles or collinear splittings, and in their computing speed performances. By default, reconstruction uses the CDF cone algorithm. Jets are stored if their transverse energy is higher than $20~\textrm{GeV}$~\citep{qr:ptcutjet}.
    507464 
    508465
    509466\subsubsection*{Energy flow}
    510467
    511 In jets, several particle can leave their energy into a given calorimetric cell,
     468In jets, several particle can leave their energy in a given calorimetric cell,
    512469which broadens the jet energy resolution. However, the energy of charged
    513470particles associated to jets can be deduced from their associated track, thus
     
    530487parent $b$ quark. For $c$-jets and light jets (i.e.\ originating in $u$, $d$,
    531488$s$ quarks or in gluons), a fake $b$-tagging efficiency of $10 \%$ and $1 \%$
    532 is assumed respectively~\citep{qr:btag}. Therefore, in current version of
     489is assumed respectively~\citep{qr:btag}. In current version of
    533490\textit{Delphes}, the displacement of secondary vertices is not taken into
    534 account. As such, the $b$-tagging efficiency is below the expected $40\%$.
     491account.
    535492
    536493\subsection{Identification of hadronic \texorpdfstring{$\tau$}{\texttau} decays}
    537494
    538495Jets originating from $\tau$-decays are identified using a procedure consistent
    539 with the one applied in a full detector simulation~\citep{bib:cmsjetresolution}.
     496with the one applied in a full detector reconstruction~\citep{bib:cmsjetresolution}.
    540497The tagging relies on two properties of the $\tau$ lepton. First, $77\%$ of the
    541 $\tau$ hadronic decays contain only one charged hadron associated to a few
     498$\tau$ hadronic decays contain only one charged hadron in combination with a few
    542499neutrals (\textit{1-prong}). Secondly, the particles arisen from the
    543500$\tau$ lepton produce narrow jets in the calorimeter (this is defined as the jet
     
    602559of tracks associated to particles with significant transverse momenta is one and
    603560only one in a cone of radius $R^\textrm{tracks}$ ($3-$prong $\tau$-jets are
    604 rejected). This cone should be entirely incorporated into the tracker to be
     561rejected)~\footnote{In release 1.9, the 3-prong tau decays are also included.}. This cone should be entirely incorporated into the tracker to be
    605562taken into account. Default values of these parameters are given in
    606563Tab.~\ref{tab:tauRef}.
     
    644601However, as muon candidates, tracks and calorimetric cells are available in the
    645602output file, the missing transverse energy can always be reprocessed a
    646 posteriori with more specialised algorithms.
     603posteriori with more specialised algorithms. Moreover, the degradations of the missing transverse energy performance due to noise is not simulated.
    647604
    648605\section{Trigger emulation}
     
    657614
    658615A trigger emulation is included in \textit{Delphes}, using a fully
    659 parametrisable \textit{trigger table} \citep{qr:triggercard}. When enabled, this
    660 trigger is applied on analysis-object data. In a real experiment, the online
    661 selection is often divided into several steps (or \textit{levels}).
    662 corresponding to the different trigger levels.
    663 First-level triggers are fast and simple but based only on partial data as not
    664 all detector front-ends are readable within the decision latency.
    665 Higher level triggers are more complex, of finer-but-not-final quality and
    666 based on full detector data.
    667 
    668 Real triggers are thus intrinsically based on reconstructed data with a worse
    669 resolution than final analysis information. On the contrary, the same
    670 information is used in \textit{Delphes} for the trigger emulation and for final
    671 analyses.
     616parametrisable \textit{trigger table} \citep{qr:triggercard}. While triggers in real experiments are intrinsically based on reconstructed data with a worse
     617resolution than final analysis information, in \textit{Delphes} the same information is used for the trigger emulation and for final analyses.
    672618
    673619\section{\label{sec:vfd}Very forward detector simulation}
     
    801747To be able to reach these detectors, particles must have a charge identical to
    802748the beam particles, and a momentum very close to the nominal value of the beam
    803 particules. These taggers are near-beam detectors located a few millimetres from
     749particles. These taggers are near-beam detectors located a few millimetres from
    804750the true beam trajectory and this distance defines their acceptance
    805751(Tab.~\ref{tab:fdetacceptance}). For instance, roman pots at $220~\textrm{m}$
    806752from the  \textsc{IP} and $2~\textrm{mm}$ from the beam will detect all forward
    807 protons with an energy between $120$ and $900~\textrm{GeV}$~\citep{bib:hector}.
     753protons with an energy loss between $120$ and $900~\textrm{GeV}$~\citep{bib:hector}.
    808754In \textit{Delphes}, extra hits coming from the beam-gas events or
    809755secondary particles hitting the beampipe in front of the detectors are not taken
     
    847793defined according to the detector specifications. Similarly, the $b$-tagging
    848794efficiency (for real $b$-jets) and misidentification rates (for fake $b$-jets)
    849 are taken directly from the expected values of the experiment. Unlike these
    850 simple objects, jets and missing transverse energy should be carefully
     795are taken directly from the expected values of the experiment. Unlike these objects, jets and missing transverse energy should be carefully
    851796cross-checked.
    852797
     
    897842generator-level particles $E_T^\textrm{MC}$, in a \textsc{CMS}-like detector.
    898843The jets events are reconstructed with the JetCLU clustering algorithm with a
    899 cone radius of $0.7$. The maximum separation between the reconstructed and
     844cone radius of $0.7$ and no energy flow correction. The maximum separation between the reconstructed and
    900845\textsc{MC}-jets is $\Delta R= 0.25$. Dotted line is the fit result for
    901846comparison to the \textsc{CMS} resolution~\citep{bib:cmsjetresolution}, in blue.
     
    940885energy of the closest jet of generator-level particles $E^\textrm{MC}$, in an
    941886\textsc{ATLAS}-like detector. The jets are reconstructed with the $k_T$
    942 algorithm with a radius $R=0.6$. The maximal matching distance between
     887algorithm with a radius $R=0.6$ and no energy flow correction. The maximal
     888matching distance between
    943889\textsc{MC}- and reconstructed jets is $\Delta R=0.2$. Only central jets are
    944890considered ($|\eta|<0.5$). Dotted line is the fit result for comparison to the
     
    961907The samples used to study the \textsc{MET} performance are identical to those
    962908used for the jet validation. It is worth noting that the contribution to
    963 $E_T^\textrm{miss}$ from muons is negligible in the studied sample.
     909$E_T^\textrm{miss}$ from muons is negligible in the studied sample. Indeed for e.g.$W\rightarrow \mu \nu$ samples the missing transverse energy should be corrected for the presence of muons.
    964910The input samples are divided in five bins of scalar $E_T$ sums $(\Sigma E_T)$.
    965911This sum, called \textit{total visible transverse energy}, is defined as the
     
    10551001coverage of the different detector subsystems.
    10561002As an example, the generic detector geometry assumed in this paper is shown in
    1057 Fig.~\ref{fig:GenDet3} and~\ref{fig:GenDet2}. The extensions of the central
     1003Fig.~\ref{fig:GenDet3}. The extensions of the central
    10581004tracking system, the central calorimeters and both forward calorimeters are
    10591005visible. Note that only the geometrical coverage is depicted and that the
    10601006calorimeter segmentation is not taken into account in the drawing of the
    10611007detector.
    1062  
    1063 \begin{figure}[!ht]
    1064 \begin{center}
    1065 \includegraphics[width=\columnwidth]{fig11}
    1066 \caption{Layout of the generic detector geometry assumed in \textit{Delphes}.
    1067 Open 3D-view of the detector with solid volumes. Same colour codes as for
    1068 Fig.~\ref{fig:GenDet3} are applied. Additional forward detectors are not
    1069 depicted.}
    1070 \label{fig:GenDet2}
    1071 \end{center}
    1072 \end{figure}
    10731008 
    10741009Deeper understanding of interesting physics processes is possible by displaying
     
    10781013mouse action. As an illustration, an associated photoproduction of a $W$ boson
    10791014and a $t$ quark~\citep{bib:wtphotoproduction} is shown in Fig.~\ref{fig:wt}.
    1080 
    1081 % This corresponds to a $pp(\gamma p \rightarrow Wt)pX$ process, where the $Wt$
    1082 % couple is induced by an incoming photon emitted by one of the colliding
    1083 % proton. This leading proton survives after photon emission and is present in
    1084 % the final state. As the energy and virtuality of the emitted photon are low,
    1085 % the surviving proton does not leave the beam and escapes from the central
    1086 % detector without being detected. The experimental signature is a lack of
    1087 % hadronic activity in the forward hemisphere where the surviving proton
    1088 % escapes.
    1089 % The $t$ quark decays into a $W$ boson and a $b$ quark. Both $W$ bosons decay
    1090 % into leptons ($W \rightarrow \mu \nu_\mu$ and $W \rightarrow e \nu_e$). The
    1091 % balance between the missing transverse energy and the charged lepton pair is
    1092 % clear, as well as the presence of an empty forward region. It is interesting
    1093 % to notice that the reconstruction algorithms build a fake $\tau$-jet around
    1094 % the electron.
    10951015
    10961016\begin{figure}[!ht]
     
    11081028\end{figure}
    11091029
    1110 For comparison, Fig.~\ref{fig:gg} depicts an inclusive gluon pair production
    1111 $pp \rightarrow ggX$. The event final state contains more jets, in particular
    1112 along the beam axis, which is expected as the interacting protons are destroyed
    1113 by the collision.
    1114 
    1115 \begin{figure}[!ht]
    1116 \begin{center}
    1117 \includegraphics[width=0.6\columnwidth]{fig13}
    1118 \caption{Example of inclusive gluon pair production $pp \rightarrow ggX$. Many
    1119 jets are present in the event, in particular along the beam axis (black line).}
    1120 \label{fig:gg}
    1121 \end{center}
    1122 \end{figure}
    1123 
    1124 
    11251030\section{Conclusion and perspectives}
    11261031
    11271032We have described here the major features of the \textit{Delphes} framework,
    1128 introduced for the fast simulation of a collider experiment. This framework is a
     1033intended for the fast simulation of a collider experiment. This framework is a
    11291034tool meant for feasibility studies in phenomenology, gauging the observability
    11301035of model predictions in collider experiments.
     
    11591064 
    11601065\bibitem{bib:delphes} \textit{Delphes}, \href{http://www.fynu.ucl.ac.be/delphes.html}{www.fynu.ucl.ac.be/delphes.html}
     1066\bibitem{bib:othersim} Other less sophisticated software for fast simulation: \textit{AcerDET} E. Richter-Was, \href{http://arxiv.org/abs/hep-ph/0207355}{arXiv:hep-ph/0207355v1}; \textit{PGS}, John Conway et al, \href{http://www.physics.ucdavis.edu/~conway/research/software/pgs/pgs4-general.htm}{http://www.physics.ucdavis.edu/~conway/}
     1067
    11611068\bibitem{bib:Root} %\textsc{ROOT}, \textit{An Object Oriented Data Analysis Framework},
    11621069R. Brun, F. Rademakers, Nucl. Inst. \& Meth. in \textbf{Phys. Res. A} \href{http://dx.doi.org/10.1016/S0168-9002(97)00048-X}{389 (1997) 81-86}.
     
    12541161\bibitem[s]{qr:ptcutjet} See the \texttt{PTCUT\_jet }variable in the detector card.
    12551162
    1256 \bibitem[t]{qr:jetparams} See the \texttt{JET\_coneradius} and \texttt{JET\_seed} variables in the detector card. The existing FastJet code has been modified to allow easy modification of the cell pattern in $(\eta, \phi)$ space.
    1257 In following versions of \textit{Delphes}, a new dedicated plug-in will be created on this purpose.
     1163%\bibitem[t]{qr:jetparams} See the \texttt{JET\_coneradius} and \texttt{JET\_seed} variables in the detector card. The existing FastJet code has been modified to allow easy modification of the cell pattern in $(\eta, \phi)$ space. In following versions of \textit{Delphes}, a new dedicated plug-in will be created on this purpose.
    12581164
    12591165\bibitem[u]{qr:energyflow} Set \texttt{JET\_Eflow} to $1$ or $0$ in the detector card in order to switch on or off the energy flow for jet reconstruction.
Note: See TracChangeset for help on using the changeset viewer.