3D codes: TraFiC4 and CSRtrack Torsten Limberg DESY Zeuthen 2003 As an introduction a few slides from: CSR code benchmarking 3-D vs. 1-D models (M. Dohlus) Simulation Models projected model (Borland, Emma,...) phase space (eg. particle tracking) 1d charge density, 1d self-field neglect transverse forces neglect transverse beam dimensions neglect deformation of retarded distribution ( (s,t) (s-c 0 t ) local rigid bunch approximation) extract 1d singularity -bunch instability (Borland, Emma, Saldin, Schneidmiller, Yurkov, Kim, Stupakov...) initial shape and energy distortions are amplified; transverse emittance & uncorrelated enegy spread are counteracting Simulation Models sub-bunch models (trafic4, R.Li, tredi...) field generating sub-bunches with fixed shape (so far) sampling particles (field monitors) selfconsistent calculation of generating sub-bunches and sampling particles reasons to mistrust the projected model: transverse bunch shape neglected transverse cross-section of bunch neglected transverse dependence of longitudinal forces neglected no space charge fields no compression work no transverse forces charge distribution at end of magnet 3: retarded charge distribution at end of magnet 3: (observer is center particle) s/m x/m driftmagnet 3 retarded 2d charge distribution retarded projected distribution: s/m s x head is neglected ! 500 MeV case: longitudinal profile head q/nC conditions for non-selfconsistent model are not fulfilled for q > 0.2nC projected modeltilted: 4x10001 head q/nC 0 1 P.Emma Conclusions projected model: trafic4-like & tilted models: fast & efficient misses important physical effects some results seem to be correct (eg. -bunch instab., benchmark bc: transv. phase space) it is not clear under what conditions complete field calculation for generating distribution transverse dimensions & forces are not negligible modeling of slice dimensions not quite satisfying First Version of 3D CSR Tracking Code: Wake Simple Input Deck Fortran Tracking Code (no quads...) Plot Beam Parameters with GNUPLOT 3D-CSR Solver from first principles With 1-3 D distributed sub-bunches M. Dohlus T. Limberg Output file TraFiC 4 (A. Kabel) Readable Input Deck (Bison Parser) Variable Bunch Set Up Self-consistent C++ Tracking Code (parallel processing with MPI) Plot Beam Parameters with GNUPLOT T4P2 (P. Piot) 3D-CSR Solver from first principles With 2-3 D distributed sub-bunches M. DohlusBig structuredOutput file containing phase space, fields, RMS values etc. CSRtrack Simple Input Files for: Calculation Parameters,Lattice, Mesh, Initial Particle Distribution Self-consistent Fortran Tracking Code (parallel processing with MPI) Separate File for Fields and Phase Space at each Mesh Point 1D projected method 3D-CSR Solver With variable sub-bunch length Greens-function method CSRtrack Interfaces Provide Matlab (or Octave) routines to interface the simple CSRtrack files and to other codes like elegant and ASTRA F. Stulle Status All necessary Interface routines exist in a basic form First calculations for S2E done. Performance on a linux cluster with 20 cpus (900 MHz): XFEL bc-3 with 50 mesh points, 200 field generating particles, 200 test particles distributed transversely to calculate slice emittance growth and optics mismatch: ca.5min 200 field generating particles, 5000 test particles: ca.2.5h To be done Introduce Quadrupole Magnets Track vertical plane Improve Interface routines for comfortable S2E simulations...