Online Data Analysis and Simulation

1
Online Data Analysis and Simulation

• Sven Reiche
• UCLA - 09/22/04

2
Goals

• Help operators in commissioning the machine.
• Essential for semi-automated procedures (e.g.
  BBA).
• Confirm (or reject) certain models for beam
  propagation.
• Complete understanding of the machine, required
  for further modification/upgrades.

3
Requirements

• Database of machine settings and diagnostic
• Integrated analysis software (e.g. emittance
  measurement)
• Core simulation suite (Parmela/Astra, Elegant,
  Genesis/Ginger)
• Data conversions modules
• Scripting environment
• Hardware resources

4
Basic Set-up
Machine/Experiment
Input for Simulation
Change Settings
Diagnostic/ Measurement
Derived Beam Model
Comparison (Operator)
Derived Measurement
Complete Beam Model
Update Model
Start-end Simulation
5
Online Data Analysis

• Goal to model the electron beam or to confirm an
  existing model.
• Direct and calibrated measurements (Charge, RF
  Phases etc).
• Programs to extract information (e.g. emittance
  from quad scan or tomography).
• Stores information in database.
• Requires eventually to automate control of the
  machine.

6
Online Simulation

• Simulation, based on machine set-values and/or
  experimental beam model.
• Reproduction of existing measurement (e.g. beam
  profiles at OTR screens).
• Comparison with experimentally derived beam
  information.
• Interface with machine set-value and/or
  experimental results.

7
Injector

• Fast calculation of beam envelope (HOMDYN,
  Trace3D) or more time-consuming Particle Tracking
  (PARMELA, Astra).
• Depends critical of the underlying model (e.g.
  thermal emittance) and injector settings (e.g.
  solenoid field, rf phase).
• Weak/non-existing interface to drive laser
  (import profile of virtual cathode into the
  codes).

8
Linac

• ELEGANT as solely choice for Linac simulation.
• Execution of 3D CSR calculation (Traffic4,
  CSRTrack) too time consuming for
  online-simulation.
• Because simulation depends on various machine
  set-values, automated interface with database is
  essential.

9
Undulator/FEL

• No dynamic machine parameters (undulator lattice
  only).
• Requires a detailed model of the electron beam.
• Calculation varies from minutes (FEL amplifier
  model) to days (full bunch SASE simulation).
• Background signal from spontaneous radiation not
  negligible.

10
Realization I

• Extension of established start-end simulation by
  automated interface between codes and/or machine
  database.
• Support by the code authors is essential.
• Codes and code-interface programs should be
  scriptable or callable by a programming
  environment (e.g. Matlab). Simplest form of
  system calls (e.g. system in C/C or spawn
  in IDL) should be sufficient.

11
Realization II

• Interface to database for machine set-values
  (dynamic parameters) and machine parameters
  (static parameters).
• Postprocessing of simulation output is written
  back to database or handed over to next code
  (agreement on format essential).
• For commissioning the postprocessor should mimic
  diagnostic for better comparison.

12
Simplification for Operation

• No particle tracking for Injector
• No CSR codes for Linac tracking
• FEL Amplifier model only for time-dependent runs
• Far-field approximation for calculation of the
  spontaneous background
• Piecewise calculation (no single-button start-end
  simulation)

13
Conclusion

• Diagnostic has to be defined so that codes can
  model the equivalent virtual LCLS.
• Automation between codes is straight forward if
  support by authors exists.
• Embedded in an easy to program environment
  (Matlab).
• Online simulation have to be fast to support the
  operation of the machine and not the data
  analysis.