Skip to content

Commit

Permalink
implemented Gloria's suggestions
Browse files Browse the repository at this point in the history
  • Loading branch information
@mbarbetti
mbarbetti committed Mar 8, 2024
1 parent ea199b8 commit 2d66307
Show file tree
Hide file tree
Showing 4 changed files with 98 additions and 64 deletions.
Binary file modified docs/poster-A3-h.pdf
View file
Binary file not shown.
Binary file modified docs/poster-A3-v.pdf
View file
Binary file not shown.
82 changes: 49 additions & 33 deletions docs/poster.html
View file
Original file line number Diff line number Diff line change
Expand Up @@ -27,11 +27,25 @@ <h1 property="headline">Lamarr: implementing a flash-simulation paradigm at LHCb
<p>in <strong><em>22nd International Workshop on Advanced Computing and Analysis Techniques in Physics Research</em></strong> (ACAT 2024)</p>
<address>
<span class="medium">
<a property="author"><strong>M. Mazurek</strong><sup>a</sup> on behalf of the LHCb Simulation Project</a>
</span>
<a property="author">L. Anderlini<sup>1</sup></a>,
<a property="author">M. Barbetti<sup>2</sup></a>,
<a property="author">S. Capelli<sup>3,4</sup></a>,
<a property="author">G. Corti<sup>5</sup></a>,
<a property="author">A. Davis<sup>6</sup></a>,
<a property="author">D. Derkach<sup>7</sup></a>,
<a property="author">M. Martinelli<sup>3,4</sup></a>,
<a property="author"><strong>M. Mazurek</strong><sup>8</sup></a>
</span>
<br/> <span class="normal">
<sup>a</sup><a property="sourceOrganization">National Centre for Nuclear Research (NCBJ), Poland</a>
</span>
<sup>1</sup><a property="sourceOrganization">INFN-Firenze</a>,
<sup>2</sup><a property="sourceOrganization">INFN-CNAF</a>,
<sup>3</sup><a property="sourceOrganization">INFN-MiB</a>,
<sup>4</sup><a property="sourceOrganization">University of Milano-Bicocca</a>,
<sup>5</sup><a property="sourceOrganization">CERN</a>,
<sup>6</sup><a property="sourceOrganization">University of Manchester</a>,
<sup>7</sup><a property="sourceOrganization">HSE University</a>,
<sup>8</sup><a property="sourceOrganization">NCBJ</a>
</span>
</address>
<span class="publication-info">
<span property="publisher">ACAT 2024</span>,
Expand All @@ -47,11 +61,11 @@ <h1 property="headline">Lamarr: implementing a flash-simulation paradigm at LHCb
<article>
<header><h3>1. Motivation</h3></header>
<p>
The <strong>detailed simulation</strong> of the interaction between the traversing particles
<strong>Detailed simulation</strong> of the interaction between the traversing particles
and the LHCb active volumes is the major consumer of CPU resources. During the LHC Run2, the
LHCb experiment has spent <strong>more than 90% of the pledged CPU time</strong> to produce
simulations. Matching the upcoming and future demand for simulated samples means that the
development of <strong>faster simulation options</strong> is critical.
LHCb experiment has spent <strong>more than 90% of the pledged CPU time</strong> to simulate
events of interest. Matching the upcoming and future demand for simulated samples means that
the development of <strong>faster simulation options</strong> is critical.
</p>
</article>

Expand All @@ -72,7 +86,8 @@ <h1 property="headline">Lamarr: implementing a flash-simulation paradigm at LHCb
<figcaption>
<small>
<strong><em>Fast simulation</em></strong> techniques aim to speed up the Geant4-based
simulation production by parameterizing the energy deposits instead of relying on physics.
simulation production by parameterizing the energy deposits instead of relying on
physics models.
</small>
</figcaption>
</figure>
Expand All @@ -93,8 +108,8 @@ <h1 property="headline">Lamarr: implementing a flash-simulation paradigm at LHCb
<header><h3>3. What is Lamarr?</h3></header>
<p>
<strong><em>Lamarr</em></strong> is the novel flash-simulation framework of LHCb, able to offer the
fastest option for simulation. Lamarr consists of a <strong>pipeline of</strong> (ML-based) <strong>modular
parameterizations</strong> designed to replace both the simulation and reconstruction steps.
fastest option to produce simulated samples. Lamarr consists of a <strong>pipeline of</strong> (ML-based)
<strong>modular parameterizations</strong> designed to replace both the simulation and reconstruction steps.
</p>
<figure>
<img style="width: 98%" src="img/schemes/lamarr-pipeline.png" alt="Lamarr modular layout">
Expand Down Expand Up @@ -134,15 +149,15 @@ <h1 property="headline">Lamarr: implementing a flash-simulation paradigm at LHCb
following models:
<ul>
<li><strong><u>propagation:</u></strong> approximates the trajectory of charged particles through the
dipole magnetic field (parametric model);</li>
dipole magnetic field &#8594; <em>parametric model</em>;</li>
<li><strong><u>geometrical acceptance:</u></strong> predicts which of the generated tracks lay within a sensitive
area of the detector (DNN model);</li>
<li><strong><u>tracking efficiency:</u></strong> predicts which of the generated tracks in acceptance are properly
reconstructed by the detector (DNN model);</li>
area of the detector &#8594; <em>DNN model</em>;</li>
<li><strong><u>tracking efficiency:</u></strong> predicts which of the generated tracks in the acceptance are properly
reconstructed by the detector &#8594; <em>DNN model</em>;</li>
<li><strong><u>tracking resolution:</u></strong> parameterizes the errors introduced by the reconstruction algorithms
to the track parameters (GAN model);</li>
to the track parameters &#8594; <em>GAN model</em>;</li>
<li><strong><u>covariance matrix:</u></strong> parameterizes the uncertainties assessed by the Kalman filter
procedure (GAN model).</li>
procedure &#8594; <em>GAN model</em>.</li>
</ul>
</p>
<hr/>
Expand All @@ -164,12 +179,14 @@ <h1 property="headline">Lamarr: implementing a flash-simulation paradigm at LHCb
Lamarr parameterizes the high-level response of the <strong>LHCb PID system</strong> relying on the following
models:
<ul>
<li><strong><u>RICH:</u></strong> parameterizes DLLs resulting from the RICH detectors (GAN model);</li>
<li><strong><u>MUON:</u></strong> parameterizes likelihoods resulting from the MUON system (GAN model);</li>
<li><strong><u>RICH PID:</u></strong> parameterizes DLLs resulting from the RICH detectors &#8594;
<em>GAN model</em>;</li>
<li><strong><u>MUON PID:</u></strong> parameterizes likelihoods resulting from the MUON system &#8594;
<em>GAN model</em>;</li>
<li><strong><u>isMuon:</u></strong> parameterizes the response of a FPGA-based criterion for muon loose
boolean selection (DNN model);</li>
boolean selection &#8594; <em>DNN model</em>;</li>
<li><strong><u>Global PID:</u></strong> parameterizes the global high-level response of the PID system,
consisting of CombDLLs and ProbNNs (GAN model).</li>
consisting of CombDLLs and ProbNNs &#8594; <em>GAN model</em>.</li>
</ul>
</p>
<p>
Expand All @@ -192,7 +209,7 @@ <h1 property="headline">Lamarr: implementing a flash-simulation paradigm at LHCb
<article>
<header><h3>7. Neutral particles pipeline: the ECAL detector</h3></header>
<p>
The flash simulation of the LHCb ECAL detector is a non-trivial task:
The flash simulation of the LHCb ECAL detector is not trivial task:
<ul>
<li>bremsstrahlung radiation, converted photons, or merged \(\pi^0\) may lead to have \(n\) <strong>generated
particles</strong> responsible for \(m\) <strong>reconstructed objects</strong> (in general, with \(n \ne m)\);</li>
Expand Down Expand Up @@ -237,10 +254,10 @@ <h1 property="headline">Lamarr: implementing a flash-simulation paradigm at LHCb
<p>
The deployment of the ML-based models follows a <strong><em>transcompilation approach</em></strong> based on
<strong><span class="monospace">scikinC</span></strong>. The models are translated to C files, compiled as
<em>shared objects</em>, and then dynamically linked to the LHCb simulation software (Gauss).
<em>shared objects</em>, and then dynamically linked in the LHCb simulation software (Gauss).
</p>
<p>
The integration of Lamarr with Gauss unlocks:
The integration of Lamarr with Gauss enables:
<ul>
<li>interface with all the <strong>LHCb-tuned physics generators</strong> (e.g., Pythia8, EvtGen);</li>
<li>compatibility with the <strong>distributed computing middleware</strong> and production environment;</li>
Expand All @@ -256,7 +273,7 @@ <h1 property="headline">Lamarr: implementing a flash-simulation paradigm at LHCb
<figcaption>
<small>
Validation plots for the \(\Lambda_c^+\) mass obtained from Pythia8 (left) and particle-gun (right)
generators by Lamarr against detailed simulation. Reproduced from
generators by Lamarr VS. detailed simulation. Reproduced from
<a href="https://cds.cern.ch/record/2814081">LHCB-FIGURE-2022-014</a>.
</small>
</figcaption>
Expand All @@ -266,14 +283,13 @@ <h1 property="headline">Lamarr: implementing a flash-simulation paradigm at LHCb
<article>
<header><h3>9. Preliminary timing studies</h3></header>
<p>
Overall time needed for producing simulated samples has been analyzed for fully detailed simulation
(Geant4-based propagation) and Lamarr. When Lamarr is employed, the particle generation (in particular,
Pythia8) becomes the new <strong>major CPU consumer</strong>.
Overall time needed for producing simulated samples has been analyzed for detailed simulation (Geant4-based)
and Lamarr. When Lamarr is employed, the generation of particles from collisions (e.g., with Pythia8)
becomes the new <strong>major CPU consumer</strong>.
</p>
<p>
Lamarr allows to reduce the CPU cost for the simulation phase of (at least)
<strong>two-order-of-magnitude</strong>. Further timing improvements can be achieved by generating only
the signal of interest (i.e., particle-gun approach).
<strong>two-order-of-magnitude</strong>. Further timing will require speeding up the generators.
</p>
<p class="boxed center">
<strong><u>Detailed simulation:</u></strong> Pythia8 + Geant4<br/>
Expand All @@ -297,12 +313,12 @@ <h1 property="headline">Lamarr: implementing a flash-simulation paradigm at LHCb
</p>
<p>
DNN-based and GAN-based models succeed in describing the high-level response of the LHCb tracking and
PID detectors for <strong>charged particles</strong>, while work is still required to parameterize the
PID detectors for <strong>charged particles</strong>. Work is still required to parameterize the
response of the ECAL detector due to the <strong>particle-to-particle correlation problem</strong>.
</p>
<p>
The future development of Lamarr looks to design a flash-simulation framework that, although
integrated within the LHCb software stack, can also be run as <strong>standalone</strong>.
Future development Lamarr aims to support both integration within the LHCb software stack and its
use as a <strong>stand-alone</strong> package.
</p>
</article>

Expand Down
Loading

0 comments on commit 2d66307

Please sign in to comment.