Session IV: Innovative Algorithms

1
Session IV Innovative Algorithms

• The Sum
• Statistics
• Clusters by Content
• Clusters by Algorithm
• Examples from the session
• Interpretation

2
Summary, but How?

• Very diverse topics
• Almost impossible to summarize You should have
  been there.
• Solution Different approaches!

3
Summary The Sum

• Adding it all up
• (the transparencies, that is)

4
Summary Statistics

• 3 posters
• 15 talks
• From about 13 experiments
• 4h 20min of oral interaction in room 02
• One overhead projector broken
• Doesnt help either.

5
Summary For Real

• Clusters of posters and talks
• Handling of large amounts of data Compression,
  many-parameter-problem
• Parallel and localized computing algorithms
• Interpreting detector responses (particle
  identification, energy reconstruction,
  percolation processes, triggers, digital
  filtering)
• New tracking and vertexing algorithms
• Code development guidelines in HEP

6
Algorithms

• Algorithmic clusters
• Adaptive algorithms, Riemann space tracking
• Neural Nets Improved, applied and studied
• Genetic Algorithms Applied and adopted for
  parallel computing
• Filtering, classical optimizations, compression
• ISO-15504 for software management
• and many others

7
Example Tracking

• New development by CMS tracking group
  Improvement over Kalman fitter seen fast, robust
  and precise algorithms.

Combined development of new tracking and
vertexing algorithms with application for CMS and
performance studies See talks by R. Fruehwirth,
A. Strandlie, W. Waltenberger
8
Example NN, Genetic Algorithms

• Applications in online triggers
• J.M. de Seixas on removing Atlas TileCal
  outsiders
• S. Abdullin on genetic algorithm for optimizing
  CMSs SUSY triggers
• Further applications
• V. Vinogradov on LHC online HadCal E
  reconstruction
• M. Bogolyubsky on direct photon ID with ALICEs
  photon spectrometer

9
Example Links Outside

• Some had a look outside and returned with news
• Some are outside EHEP and told us about it
• H.-P. Wellisch on ISO software management
  standards and their application and feasibility
  for HEP. It works. As long as one has the
  support of Senior Personnel.
• M. Kosov on an chiral invariant phase space event
  generator for hadrons which can predict e.g.
  hadron masses, structure functions and cross
  sections

10
Interpretation

• Disclaimer Personal view, not a result of an
  algorithm
• Future and running experiments highly active in
  extensions of algorithms, but also development of
  new ones
• Judging from session, highest activity at CERN
  and in Russia (and their overlap, which seems to
  be Vienna)

11
Interpretation

• Interesting algorithms ? nice to see once
• They should be shared and spread!
• common languages (OO helps, too), common
  software libs, links across experiments, regions,
  languages and (software) cultures
• Sometimes algorithm not recognized to be
  fundamental, or just the last extra step not
  taken for the algorithm to be re-usable.

-
12
Interpretation

• ACAT plays an important role here
• To discuss and share new algorithms
• To remind us that often we dont write a specific
  solution, but a more fundamental algorithm for
  which other users exist
• To create links across experiments, countries and
  even fields of application
• To feed authors of algorithms with great shashlik
  and smooth vodka

3