TREDI test for CSR

1
TREDI test forCSR

• L. Giannessi M. Quattromini

Presented at
2
TREDI SELF FIELDS Retarded Potentials

• Trajectories are stored and the fields evaluation
  requires bracketing of
• the retarded condition.
• Cpu memory consuming
• Non trivial field reglarization by grid
  assignement (Parmela mode)
• extended particles requires careful treatment
  of ret. condition

3
TREDI Features

• 15000 lines in C language
• Scalar Parallel (MPI 2.0)
• Unix Windows versions
• Tcl/Tk Gui (pre-processing)
• Mathematica MathCad frontends
  (post-processing)
• Output format in NCSA HDF5 format (solve
  endian-ness/alignement problems)

4
Conclusions of Zehuten workshop

• The noise suppression method has reduced the
  effects of SF on longitudinal phase space,
  without being completely effective in the
  transverse phase space
• A rigorous model of fields regularization,
  relying on a realistic momentum dispersion of
  macroparticles will be soon implemented
• The low number of macroparticles in severely
  limiting the reliability of the results
• Diagnostic on fields will be implemented to
  improve insight on the smoothing procedure
• The reason of the slow down of the code must be
  understood
• Before the end of the workshop the 1000 particles
  case will be finished - we will see.

5
Effect of Noise (1st bend - no screening)
6
six months after Zehuten, Chia Laguna

• Introduced radiative energy loss (ISR)
• Now use HDF5 data format support to fix
  endianess/alignments problems (output
  portability to different platforms)
• ?? Improved acceleration fields smothness (more
  work required, no manifestly covariant, CPU
  consuming)
• Fields blow up when we have collinear
  divergencies
• Solution target macro particles are given a
  finite extension in space
• Big speed up (improved retarded time condition
  routine)
• six months ago very few particles (300 particles
  ? 4h on an IBM SP3/16 nodes - 400 MHz each)
• Now 1000 particles ? 35m on an equivalent
  platform
• 10000 particles in 27h on a 32CPUs platform
• improvements and bug fixes recently introduced
  a Parmela-like mode (instantaneous
  interactions,MUCH faster still experimental)

7
In Zehuten Phase space at exit ? still noisy !
8
After changes to field regularization
Ex2.3 mm-mrad
Ex1.8 mm-mrad
9
5 GeV 1 nC Gaussian
10
5 GeV 1 nC Gaussian
11
5 GeV - 1 nC Gaussian
12
5 GeV 0.5 nC Gaussian
13
5 GeV 0.5 nC Gaussian
14
5 GeV 0.5 nC Gaussian
15
5 GeV 1.0 nC Uniform
16
5 GeV 1.0 nC Uniform
17
5 GeV 1.0 nC Uniform
18
500 MeV 1.0nC Gaussian
19
500 MeV 1.0nC Gaussian
20
500 MeV 1.0nC Gaussian
21
ICFA Beam Dynamics mini workshopCoherent
Synchrotron Radiation and its impact onthe
dynamics of high brightness electron
beamsJanuary 14-18, 2002 at DESY-Zeuthen
(Berlin, GERMANY)http//www.desy.de/csr
15 cut of charge to reduce noise
22
Conclusions

• The agreement with other codes is improved with
  the new macroparticles model
• The cpu time is greatly reduced, this allows to
  run a larger number of macroparticles, but not
  yet sufficient to simulate microbunching.
• Fields regularization requires more work (still
  time consuming, not covariant)
• The code provides reasonable results in a wide
  range of conditions start to FEL simulations ?