COMPASS all
-
hands meeting 9/17
-
18/2007
Robert Ryne
Beam Dynamics Overview
Robert D. Ryne
COMPASS all
-
hands meeting
Sept 17
-
18, 2007
Fermilab
COMPASS all
-
hands meeting 9/17
-
18/2007
Robert Ryne
Overall Goals
•
Develop advanced beam dynamics capability
to meet the mission needs of DOE/SC HEP,
NP, and BES accelerator projects
•
Develop reusable software components to
produce a comprehensive, scalable (to
petascale), lasting accelerator modeling
capability for present and future
accelerator projects
COMPASS all
-
hands meeting 9/17
-
18/2007
Robert Ryne
COMPASS Beam Dynamics physics areas fall mainly in
7 categories
1.
Space
-
charge
2.
Beam
-
beam
3.
Multi
-
species
4.
Beam
-
environment
5.
Optics, errors, feedback
6.
High brightness e
-
beams, radiation
7.
IBS
COMPASS all
-
hands meeting 9/17
-
18/2007
Robert Ryne
Space Charge
•
Maintain SciDAC1 solvers, port/optimize
for SciDAC2 platforms
•
Develop/incorporate new solvers, working
math math/cs partners, to meet new
requirements (boundary conditions, etc)
—
See math/cs talks Tuesday
COMPASS all
-
hands meeting 9/17
-
18/2007
Robert Ryne
Beam
-
beam effects
•
Codes used
—
BeamBeam3D
—
Lifetime,Nimzovitch
COMPASS all
-
hands meeting 9/17
-
18/2007
Robert Ryne
BeamBeam3D
•
Developed by Ji Qiang
•
Multiple models (strong
-
strong, weak
-
strong)
•
Multi
-
slice (finite bunch length effects)
•
New algorithm
--
Shifted Green function
--
efficiently
treats long
-
range parasitic collisions
•
Particle
-
based decomp (perfect load balance)
•
Lorentz boost handles crossing angle collisions
•
Multi
-
IP collisions, varying phase advance,…
•
Arbitrary closed orbit sep (static or time
-
dep)
•
Applied to Tevatron, LHC, PEP
-
II, KEK
-
B, RHIC, RHIC/LARP
RHIC B
-
B
-
growth vs
x
,
y
•
Strong collaboration, code development by Stern et al at FNAL
—
Fourier 3D solver validated with observed synchro
-
betatron modes
—
Resistive
-
wall impedance model growth rate matches predictions
—
Chromaticity with coupled
-
motion maps and impedance matches predictions
—
Arbitrary bunch collision patterns w/ measured Tevatron optics & helix incorporated
COMPASS all
-
hands meeting 9/17
-
18/2007
Robert Ryne
BeamBeam3D code validation comparing with VEPP
-
II
data (E. Stern, A. Valisev, FNAL; J. Qiang, LBNL)
Beam
-
beam code validation comparing
with VEPP
-
II data
COMPASS all
-
hands meeting 9/17
-
18/2007
Robert Ryne
Sequence of frames from a BeamBeam3D simulation of a collision
at the Tevatron @ 200x nominal intensity (E. Stern, FNAL)
COMPASS all
-
hands meeting 9/17
-
18/2007
Robert Ryne
Lifetime, Nimzovitch
•
Developed by Andreas Kabel
•
LIFETIME application
—
Uses PLIBB to calculate lifetimes in storage
rings; applied to Tevatron, RHIC current wire
experiment, LHC
•
NIMZOVITCH
—
Strong
-
strong beam
-
beam code optimized for large
number of bunches/IP’s/parasitic crossings
COMPASS all
-
hands meeting 9/17
-
18/2007
Robert Ryne
Beam
-
beam plans
•
BeamBeam3D:
—
Implement wire compensation model
—
Implement rotating beam colliding w/crossing angle, test
on LHC luminosity monitor
—
Implement full nonlinear symplectic tracking
—
Implement quantum effects, test and perform high
-
resolution simulation of ILC beam
-
beam interactions
—
Incorporate solver into multi
-
physics framework
—
Use
to investigate antiproton intensity limits in the
Tevatron and the growth of multi
-
bunch modes and
electron cooling beam
-
beam compensation operation
COMPASS all
-
hands meeting 9/17
-
18/2007
Robert Ryne
Beam
-
beam plans, cont.
•
Nimzovitch:
—
Make Nimzovitch go away by reformulating under
enhanced PLIBB:
•
more physics (IBS, noise, imperfections)
•
low noise PIC
•
enforce symplectic correctness in 3D
•
beamline parallelization for multi
-
bunch calculations
•
Lifetime:
—
Apply PLIBB w/ IBS module to RHIC, experimental
validation
—
Strong
-
strong module: apply to LHC multi
-
bunch effects
COMPASS all
-
hands meeting 9/17
-
18/2007
Robert Ryne
Multi
-
species effects
•
Main emphasis on electron
-
cloud
•
2 approaches:
—
Full 3D using WARP/POSINST (A. Friedman, D. Grote, J.L.
-
Vay, M.
Furman, et al)
—
Quasi
-
static using QuickPIC (W. Mori, V. Decyk, T. Katsouleas et al)
e
-
i
+
halo
e
-
i
+
= ion
e
-
= electron
g
= gas
= photon
= instability
Positive
Ion Beam
Pipe
e
-
i
+
g
g
e
-
e
-
e
-
e
-
e
-
•
Ionization of
-
background gas
-
desorbed gas
•
ion induced emission from
-
expelled ions hitting vacuum wall
-
beam halo scraping
•
photo
-
emission from synchrotron radiation (HEP)
•
secondary emission
from electron
-
wall
collisions
COMPASS all
-
hands meeting 9/17
-
18/2007
Robert Ryne
Calculating the e
-
cloud effects in the ILC DR
wiggler is an
immense
numerical challenge
•
3D
-
fields and dynamics
•
Self
-
consistent (beam
electrons)
•
Large range of spatial scales (sets resolution,
memory req’t)
transverse:
longitudinal:
Must resolve beam
, but 3000 x 3000 x 6000 ~ 10
10
-
cell mesh!
•
Huge number of timesteps required
t =
(e
-
traverse <1 cell in 1 timestep)
COMPASS all
-
hands meeting 9/17
-
18/2007
Robert Ryne
E
-
cloud modeling using WARP
-
POSINST
•
3D field dynamics and dynamics
•
Fully self
-
consistent
•
Realistic boundary conditions
•
Detailed electron generation models
(POSINST, including energy
spectrum)
•
Drift Lorentz electron “mover”
(correct space charge w/out
resolving
cyclic orbit)
•
Mesh refinement
(spatial resolution only where needed
--
essential!)
•
Velocity sub
-
cycling
(small
t only for particles that need it)
•
Parallelized
QuickPIC uses a quasi
-
static model; under certain circumstances,
agrees well with self
-
consistent but is orders of magnitude faster
A 2
-
D slab of electrons
(macroparticles) is stepped backward
(with small time steps) through the
beam field and 2
-
D electron fields are
stacked in a 3
-
D array, that is used to
push the 3
-
D beam ions (with large
time steps) using maps (as in
HEADTAIL
-
CERN) or Leap
-
Frog (as in
QUICKPIC
-
UCLA/USC).
2
-
D slab of electrons
3
-
D beam
bend
drift
drift
quad
s
s
0
lattice
100x improvement with “no” loss in accuracy
QuickPIC: Pipelining
COMPASS all
-
hands meeting 9/17
-
18/2007
Robert Ryne
Calculation in boosted frame provides x10
n
speedup*
-
proton bunch through a given e
–
cloud
-
hose instability of a
proton bunch
Proton energy:
=
500 in Lab
•
L= 5 km, continuous
focusing
Code: WARP (Particle
-
In
-
Cell)
electron
streamlines
beam
proton bunch radius vs. z
CPU time:
•
lab frame:
>2 weeks
•
frame with
2
=512:
<30 min
Speedup x1000
*J.
-
L. Vay, PRL 98, 130405 (2007)
COMPASS all
-
hands meeting 9/17
-
18/2007
Robert Ryne
WARP
-
POSINST plans
•
Code needs development to perform run in boosted frame
and complete set of e
-
cloud related physics
—
Implementation of magnetoinductive (“Darwin”) model, or
reduce version of it if sufficient,
—
implement an interface linking zones of 3
-
D PIC
simulations to zones of MAPS transport in beween,
—
upgrade diagnostics to allow for results given in frame
different from the one of calculation,
—
implement self
-
consistent generation and tracking of
photo
-
electrons, based on Monte
-
Carlo methods,
—
implement adaptive macro
-
particle management
(reduction/coalescence),
—
upgrade parallel decomposition from 1
-
D to 2
-
D/3
-
D.
COMPASS all
-
hands meeting 9/17
-
18/2007
Robert Ryne
Beam
-
environment interactions
•
Maintain SciDAC1 wakefield modules,
port/optimize for SciDAC2 platforms
•
Implement circuit model for time
-
dep
beam loading effects
•
Fully self
-
consistent calculation using
VORPAL
COMPASS all
-
hands meeting 9/17
-
18/2007
Robert Ryne
Optics, errors, feedback
•
Maintain existing optics libraries that are
used in the BD framework, port/optimize
for SciDAC2 platforms
•
Extend multi
-
bunch capabilities
•
Implement models for dynamically changing
quantities (e.g. jitter), machine errors, and
feedback systems
COMPASS all
-
hands meeting 9/17
-
18/2007
Robert Ryne
High brightness electron beam dynamics
•
Codes used:
—
Elegant
—
IMPACT (collaboration with synergistic high
brightness e
-
beam activities at LBNL funded by
other sources)
•
Essential goal of this work is to support LCLS
commissioning, operation, and optimization with
fast, high
-
fidelity modeling tools
•
Large scale computing essential for detailed study
of the microbunching instability
COMPASS all
-
hands meeting 9/17
-
18/2007
Robert Ryne
Elegant: status and limitations
•
CSR and longitudinal space charge parallelized in elegant
•
Limited to ~60 million particles presently
•
With ~1.5 billion particles we could look at modulations on the 1 mm
level relevant to proposed laser/undulator beam heaters
•
Addressing I/O and memory management issues related to this
•
Present fast CSR algorithms are 1
-
d simplifications
•
Existing 3
-
d algorithms are coarse
-
grained, time
-
consuming
•
No standardized, accepted tools exist for transferring information
between various accelerator codes (elegant, IMPACT) and radiation
modeling codes (GINGER, GENESIS, SPUR)
•
No way to take a snapshot of an existing FEL, simulate it, then
compare simulated and real diagnostics
•
LCLS is already reporting [Frisch, PAC07] unexplained effects with
very short bunches in the first compressor
•
Must be able to optimize to match a selection of diagnostics, then
extrapolate to other diagnostics
FERMI FEL Microbunching Instability Simulated with elegant
BLS
BC1
BC2
Tiny initial density modulations
build up in bunch compression
systems due to CSR and space
charge. Gain increases
to ~2000
-
fold down to 25
m
modulation. Can't presently
go shorter than this!
COMPASS all
-
hands meeting 9/17
-
18/2007
Robert Ryne
IMPACT: status and limitations
•
Successfully used to perform 1B macroparticle
simulations of Fermi FEL linac
•
Limitations: 1D CSR model, difficult to use for
design optimization, simple matrix description of
RF elements, not fully integrated with FEL codes
COMPASS all
-
hands meeting 9/17
-
18/2007
Robert Ryne
1B particle simulation of microbunching in FERMI FEL
linac using IMPACT
Final Longitudinal Phase Space Distribution Using
10M
and
1B
particles (init. 15 keV energy spread, 2BCs)
COMPASS all
-
hands meeting 9/17
-
18/2007
Robert Ryne
Summary of ANL’s tasks/plans
•
Finish parallelization of
elegant
•
Develop accepted, robust interfaces among suite of
codes involved in FEL modeling
—
IMPACT (gun and linac modeling)
—
elegant (accelerator modeling and optimization)
—
GENESIS and GINGER (FEL modeling)
•
Develop integrated graphical user interface to provide
on
-
demand, high
-
fidelity modeling of data and experients
—
Selection of codes, algorithms, detail level
—
Utilizes data drawn from the control system
—
Utilizes high
-
performance computing resources
•
Develop optimizer based on genetic algorithm to provide
guidance on FEL performance improvement.
COMPASS all
-
hands meeting 9/17
-
18/2007
Robert Ryne
IMPACT development plans for high brightness e
-
beam
(funded by LDRD and other non
-
SciDAC projects)
•
Develop and implement interfaces for start
-
to
-
undulator parallel simulation
•
Fully self
-
consistent CSR (difficult!)
•
Automatic beam steering
•
Integration with optimization tool
•
Incorporate nonlinear model of RF beamline
elements
COMPASS all
-
hands meeting 9/17
-
18/2007
Robert Ryne
Frameworks
•
Synergia/SciDAC1
•
IMPACT suite
•
MaryLie/IMPACT
•
UPIC
•
PLIBB
COMPASS all
-
hands meeting 9/17
-
18/2007
Robert Ryne
Synergia/SciDAC1
•
See next talk
COMPASS all
-
hands meeting 9/17
-
18/2007
Robert Ryne
IMPACT
•
A code suite (linac design, 3D rms code, 2 parallel PIC tracking
codes) developed under SciDAC1
•
Includes IMPACT
-
Z and IMPACT
-
T 3D parallel PIC codes
•
Applicable to electron and ion accelerators
•
Recent enhancements
—
Cathode emission model; cathode image effects
—
Energy binning for large
E
—
Multi
-
charge state capability (RIA)
—
SW and TW structures
—
wakefields
—
1D CSR
•
IMPACT
-
T now widely used for
photoinjector modeling
—
BNL e
-
cooling project, Cornell ERL,
FNAL/A0, LBNL/APEX, ANL, JLAB,
SLAC/LCLS, Fermi@elettra
Emission from nano
-
needle tip
COMPASS all
-
hands meeting 9/17
-
18/2007
Robert Ryne
MaryLie/IMPACT (ML/I)
•
Hybrid code combining MaryLie beam optics
with IMPACT parallel PIC + new capabilities
—
Embeds operator splitting for all thick
elements
—
Allows mixed MaryLie and MAD input
—
New software modules (wakefields, soft
-
edge magnet models, …) add functionality
—
Performance optimization (R. Gerber,
NERSC staff)
•
Multiple uses all within in the same code
—
Particle tracking, envelope tracking, map
production, map analysis, lattice
functions, fitting.
•
User manual and example suite
•
Contributions from many people from many
disciplines (follows the SciDAC model)
damping ring simulation with MLI
COMPASS all
-
hands meeting 9/17
-
18/2007
Robert Ryne
PLIBB particle dynamics framework
•
Developed at SLAC (A. Kabel)
•
a general
-
purpose C++ framework for high
-
speed,
parallel tracking studies
•
fast and easily extensible through compile
-
time
polymorphism
•
easily applied: MAD{X,8} beamline parsers &
manipulators
•
physics: magnetic elements, cavities, wakefields,
beam
-
beam
•
analysis: statistics, differential algebra, collective
quantities
COMPASS all
-
hands meeting 9/17
-
18/2007
Robert Ryne
UPIC Framework for Parallel PIC
•
Developed by V. Decyk
•
Layered, Fortran based, but could be called from C/C++
•
Goals:
—
Rapid construction of new parallel PIC codes from trused components
—
High accuracy testbed for evaluating and verifying PIC algorithms
•
Supports
—
Multiple plasma models: electrostatic, Darwin, electromagnetic
—
Multiple boundary conditions: periodic, dirichlet, neumann, open3.
Multiple levels of accuracy: linear, quadratic, gridless4. Multiple
programming paradigms: procedural, object
-
oriented5. Multiple
parallel models: threads, message
-
passing.
•
Used in QuickPIC and other applications
COMPASS all
-
hands meeting 9/17
-
18/2007
Robert Ryne
Frameworks: Plans
•
See next talk
Enter the password to open this PDF file:
File name:
-
File size:
-
Title:
-
Author:
-
Subject:
-
Keywords:
-
Creation Date:
-
Modification Date:
-
Creator:
-
PDF Producer:
-
PDF Version:
-
Page Count:
-
Preparing document for printing…
0%
Comments 0
Log in to post a comment