Cmad simulation code status • Tracking the beam in a MAD lattice, parallel code, interaction with cloud at each element in the ring and with different cloud distribution, single-bunch instability studies, threshold for SEY, dynamic aperture tune shift … • Single-bunch instability preliminary studies ongoing. • Electron cloud build-up (SEY, vacuum chamber, etc.) to be added. ILC DR ILC DR bunch at injection as input Electron cloud distribution in bends and straights (so far from POSINST) ILC DR lattice MAD to track beam with an electron cloud at each element location.
Cmad simulation code status Tracking the beam in a MAD lattice, parallel code, interaction with cloud at each element in the ring and with different cloud distribution, single-bunch instability studies, threshold for SEY, dynamic aperture tune shift … - PowerPoint PPT Presentation
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Cmad simulation code status • Tracking the beam in a MAD lattice, parallel code, interaction with cloud at each
element in the ring and with different cloud distribution, single-bunch instability studies, threshold for SEY, dynamic aperture tune shift …
• Single-bunch instability preliminary studies ongoing.• Electron cloud build-up (SEY, vacuum chamber, etc.) to be added.
ILC DR
ILC DR bunch at injection as input
Electron cloud distribution in bends and straights (so far from POSINST)
ILC DR lattice MAD to track beam with an electron cloud at each element location.
Goals– simulate electron cloud instability threshold in the
LC DRs, LHC, SPS and storage rings
Status– Benchmarked against codes at CERN web page.
Good agreement with existing codes (HEAD-TAIL “new simulation results 2006”) for 1 interaction point/turn. Ongoing SPS and ILC DR simulations and code benchmarking.
– Beta version, still good room to gain in speed: Electric fields calculation to be upgraded, parallel features to be optimized
Dynamics
Beam and electron cloud Dynamics:• MAD sectormap and optics functions files as input• Tracking 1 bunch in the lattice by 1st (switch for 2nd)
order transport maps R (and T)• Particle in cell PIC code• Tracking 6D beam phase space, 3D beam dynamics• 3D electron cloud dynamics• Apply beam-cloud interaction at each element of
MAD lattice• 2D forces beam-cloud, cloud-cloud computed at
interaction point• Electron dynamics: cloud pinching and magnetic
fields included
Code benchmarking CMAD with Head-Tail code “new simulation results 2006” on web page. 1 interaction point/turn. CMAD with 32 processors in this simulation on seaborg/nersc IBM6000.
SPS simplified model, R. Thomas
Emittance growth simulated with TAILHEAD considering a single electron-beam interaction point at the center of each half cell (left) and with two interaction points, close to the QD and QFF quadrupoles (right). The electron cloud responds either dynamically, or it is frozen. The electron density is 2.75e11 m-3 for the left picture and either 1e11 m-3 or 2e11 m-3 on the right. SPS simplified model.
SPS simplified model
Emittance growth simulated with CMAD considering a single electron-beam interaction point at the center of each half cell (left) frozen potential after first interaction . The electron density is 2.5e11 m-3 to 7.5e11 m-3. SPS simplified model.
SPS simplified model
Improvements:• Electric field computation is actually a direct node to
node (slow) computation and interpolation – Room for improvement here…!
• Actually looking into Multigrid parallel like PHAML
Improvements:• Electric field computation is actually a direct node to
node (slow) computation and interpolation – Room for improvement here…!
• Actually looking into Multigrid Poisson parallel solver like PHAML
• Suggestions for improvement?!
• Parallel improvement:– Actually computing beam-cloud interaction with all processors
for each ring element– Next: assign a ring section of N elements to N processors to
each compute beam-cloud interaction, then track the beam through the N elements (assume weak beam changes)
Approximations to be added:• Optional: If beta functions are identical at two
elements in phase (2*pi*n), apply earlier computed cloud-potential
Approximations [Longer term!] to be added:• Build-up of the electron cloud initially computed until
saturation for each representative of magnet class and density distributed over the ring; Then updated at each turn (assuming weak changes of the beam turn by turn)..
phase space conserved with R and T tracking, ILCdr 1000 turns)
Discussion, ideas, suggestions
• !****************************************• ! SET UP PARALLEL COMPUTATION• ! ***************************************• CALL MPI_INIT(ierror) ; ! Initialize MPI• CALL MPI_COMM_SIZE(MPI_COMM_WORLD, numTasks, ierror); ! Find number of Tasks (processors)• CALL MPI_COMM_RANK(MPI_COMM_WORLD, me , ierror); ! Find the ID of this task