Difference between revisions of "2018-05-11"

From Administrative
Jump to: navigation, search
Line 66: Line 66:
 
----
 
----
 
'''Computational Physicss - Y. Roblin'''<br>
 
'''Computational Physicss - Y. Roblin'''<br>
 +
Kirsten Deitrick
 +
*Last two weeks:
 +
**Straight Merger at CBeta:
 +
*:-finish figures
 +
*:-make presentation for CBeta
 +
*:-data taking during experiment at CBeta
 +
**CSR:
 +
*:-continue working on KEK CSR tech note
 +
**eCloud:
 +
*:-tracking ion beam parameters around ring
 +
*:-have received NERSC allocation/account - working on getting script from Jean-Luc Vay for WARP+POSINST
 +
*:-got in contact Jim Crittenden at Cornell
 +
*:-contact M. Furman about POSINST
 +
**Impedance:
 +
*:-send out request for meeting items
 +
**Meetings:
 +
*:-CBETA weekly meeting
 +
**Misc:
 +
*:-LINAC18 travel authorization
 +
*Next two weeks:
 +
**Magnetized beam:
 +
*:-zeroing after FEL booster
 +
*:-start evaluating merger options?
 +
**Straight Merger at CBeta:
 +
*:-redo simulations using machine settings (with/without slits)
 +
*:-analyze measurement data
 +
*:-work on write-up
 +
**CSR:
 +
*:-continue working on tech note for KEK CSR
 +
**eCloud:
 +
*:-continue following up with Furman, J.L. Vay, Crittenden
 +
*:-ion ring tracking for input into PyECLOUD
 +
**Impedance:
 +
*:-meeting
 +
**Meetings:
 +
*:-impedance
 +
*:-CASA
 +
*:-JLEIC R&D (x2)
 +
*:-Cooler (x2)
 +
<br>
 +
 +
Yves Roblin
 +
**CEBAF:
 +
*:-model audit.
 +
*:-audit of the 5 pass extraction setup., make a model for  predicting the setup from correctors settings in the machine
 +
*:-re-setup of the extraction at 5 pass
 +
*:-beam studies
 +
*:-BTEAM meetings
 +
*:-raytrace/autosteer development.
 +
*:-preparing plan for summer shutdown (model development, etc..)
 +
**OTHER:
 +
*:-NSF proposal review  due may 18
 +
*:-IPAC paper review
 +
*:-Nim Journal Review  .
 +
*:-Computer Physics Communications journal Review  due may 22
 +
 +
He Zhang
 +
 +
JLEIC Cooling Simulation
 +
Continue cooling simulations for JLEIC following Steve’s suggestion. First, compare the cooling with (a) round beam, (b) flat beam with reduce drift emittance, and (c) flat beam with unchanged drift emittance. Simulation suggests (b) provides the best cooling. (c) is slightly worse than (b), but obviously better than (a). (b) also avoids the difficulty to further reduce the drift emittance of the electron beam, so it may be preferred for JLEIC cooling.  Second, simulate the cooling for high density proton beam for 64 GeV CM energy with 6.4 nC/bunch electron beam. Unfortunately, even with 6.4 nC/bunch electron beam, we still do not have enough cooling for the high density proton beam. We will need to reduce the proton beam current for an equilibrium between the IBS and cooling. Third, Simulate the cooling for 200 GeV proton beam with 3.2 nC/bunch electron beam. As expected, cooling is not strong enough. We will need to either reduce the proton current or increase the electron beam current. Some of the above results were presented in the cooling ring design meeting. I mentioned to Steve there are two things we should do to optimize the cooling. One is to use an optimizer, e.g. NSGA, to explore the proton beam parameter domain with a fixed electron beam. The other is to figure out which part of the ion ring contributes the most to the horizontal IBS and investigate whether we can optimize the optics to reduce the horizontal IBS. Probably we can start this after we finish the pCDR draft.
 +
 +
Fast Multipole Method
 +
 +
Generate new data to compare our traceless FMM code for Coulomb kernel with the pyfmmlib. Pyfmmlib is a Python wrapper for a well optimized Fortran FMM lib for Coulomb kernel and Helmholtz kernel. Our code is developed for illustration of the idea. We have been careful to avoid redundant calculations, but the code has not been fully optimized. The numerical results show that with a lower tensor rank for relative error about 1e-4 or larger, our code is faster. But as the tensor rank goes up, the pyfmmlib outperformed our code. This is as expected since the Cartesian tensor FMM has a higher order dependence on the rank than the solid harmonic FMM. Although our work reduce the order dependence from O(n6) to O(n5), it does not change the conclusion in general. But The Cartesian tensor FMM can be used for simulations with moderate accuracy requirement. The Cartesian tensor FMM also has higher flop rate than other FMMs. As the bandwidth becomes more expansive. The Cartesian tensor FMM may be preferred in future. I add the new data and some discussion of the Cartesian tensor FMM to the FMM manuscript and submitted the revised one to JCOMP.
 +
 +
Investigate a few math parsers and found one called metl, which allows to parse the expression into a function with user defined variables. I have tested this parser with my tpsa and da code. It supports the DAVector data type without problem. This means our kernel-independent FMM code will be able to treat different kernels without recompiling. I started to write a paper on the kernel-independent FMM. Now I am writing on the analytical part. Will need more numerical result to complete the paper. 
 +
 +
Work plan in the following two weeks
 +
 +
Write for pCDR.
 +
Write some code (probably in Python) to solve the dynamic equations for dust particle transfer.
 +
Generate some data for kernel-independent FMM and write a paper on it.
 +
Benchmark and add the turn-by-turn tracking model into JSPEC.
 +
  
  

Revision as of 08:50, 14 May 2018

Director - M. Spata
No report. Acting Accelerator Division Head.



Accelerator Physics - T. Satogata
Todd Satogata

  Previous two weeks (Apr 31-May 11)
     - IPAC18: Scientific secretariat work (student program, light peer review)
     - JLEIC: pCDR writing, template, EIC lunch, visitors (Peggs,
  Next two weeks (May 14-May 25)
     - JLEIC: pCDR writing, template, EIC lunch, impedance and R&D meetings
     - JLEIC: P&C Meeting (May 21)
     - ODU: Continue Randy advising, presentation to committee (May 24)
     - ADMIN: Leadership/group meetings, staffing input and planning
     - ADMIN: Prepare for summer student arrivals
     - IPAC18: Complete proceedings, light peer review, delegate survey
     - OPS: BTeam meetings, Hall D optics analysis

Alex Bogacz

   Previous two weeks (Apr 31-May 11)
     - IPAC18: Apr 31-May 4 (light peer review)
     - VACATION: May 7-11
   Next two weeks (May 14-May 25)
     - IPAC18: Light peer review development
     - PERLE process followup

Rui Li

   Previous two weeks (Apr 31-May 11)
     - IPAC18: Apr 31-May 4 (delegate)
     - IPAC18: CSR meeting with experts in field, path forward in field
       for theory, simulation, and experiment
     - JLEIC: Clean up IPAC papers, start pCDR documentation
     - JLEIC: Prepare for impedance meeting
   Next two weeks (May 14-May 25)
     - JLEIC: Impedance meeting (May 14)
     - JLEIC: Work on pCDR writeup
     - VACATION: (May 17-18)

Ed Nissen

   Previous two weeks (Apr 31-May 11)
     - JLEIC: Finalize IPAC paper
     - JLEIC: Reading up on octupole detuning, set up madx optimization
     - JLEIC: initial draft writing for pCDR
   Next two weeks (May 14-May 25)
     - JLEIC: Finish draft writing for pCDR
     - JLEIC: Continue booster nonlinear optimization work

Chris Tennant

   Previous two weeks (Apr 31-May 11)
     - IPAC18: Apr 31-May 4 (delegate)
     - IPAC18: Finish IPAC papers and documentw rok
   Next two weeks (May 14-May 25)
     - ADMIN: Migrate laptops (to Dave's old one)
     - JLEIC: translate DIMAD ERL-to-CCR transport to elegant
       to build up full S2E of cooler
     - PERLE: Continued CSR analysis



Accelerator R & D - Y. Zhang



Computational Physicss - Y. Roblin
Kirsten Deitrick

  • Last two weeks:
    • Straight Merger at CBeta:
    -finish figures
    -make presentation for CBeta
    -data taking during experiment at CBeta
    • CSR:
    -continue working on KEK CSR tech note
    • eCloud:
    -tracking ion beam parameters around ring
    -have received NERSC allocation/account - working on getting script from Jean-Luc Vay for WARP+POSINST
    -got in contact Jim Crittenden at Cornell
    -contact M. Furman about POSINST
    • Impedance:
    -send out request for meeting items
    • Meetings:
    -CBETA weekly meeting
    • Misc:
    -LINAC18 travel authorization
  • Next two weeks:
    • Magnetized beam:
    -zeroing after FEL booster
    -start evaluating merger options?
    • Straight Merger at CBeta:
    -redo simulations using machine settings (with/without slits)
    -analyze measurement data
    -work on write-up
    • CSR:
    -continue working on tech note for KEK CSR
    • eCloud:
    -continue following up with Furman, J.L. Vay, Crittenden
    -ion ring tracking for input into PyECLOUD
    • Impedance:
    -meeting
    • Meetings:
    -impedance
    -CASA
    -JLEIC R&D (x2)
    -Cooler (x2)


Yves Roblin

    • CEBAF:
    -model audit.
    -audit of the 5 pass extraction setup., make a model for predicting the setup from correctors settings in the machine
    -re-setup of the extraction at 5 pass
    -beam studies
    -BTEAM meetings
    -raytrace/autosteer development.
    -preparing plan for summer shutdown (model development, etc..)
    • OTHER:
    -NSF proposal review due may 18
    -IPAC paper review
    -Nim Journal Review .
    -Computer Physics Communications journal Review due may 22

He Zhang

JLEIC Cooling Simulation Continue cooling simulations for JLEIC following Steve’s suggestion. First, compare the cooling with (a) round beam, (b) flat beam with reduce drift emittance, and (c) flat beam with unchanged drift emittance. Simulation suggests (b) provides the best cooling. (c) is slightly worse than (b), but obviously better than (a). (b) also avoids the difficulty to further reduce the drift emittance of the electron beam, so it may be preferred for JLEIC cooling. Second, simulate the cooling for high density proton beam for 64 GeV CM energy with 6.4 nC/bunch electron beam. Unfortunately, even with 6.4 nC/bunch electron beam, we still do not have enough cooling for the high density proton beam. We will need to reduce the proton beam current for an equilibrium between the IBS and cooling. Third, Simulate the cooling for 200 GeV proton beam with 3.2 nC/bunch electron beam. As expected, cooling is not strong enough. We will need to either reduce the proton current or increase the electron beam current. Some of the above results were presented in the cooling ring design meeting. I mentioned to Steve there are two things we should do to optimize the cooling. One is to use an optimizer, e.g. NSGA, to explore the proton beam parameter domain with a fixed electron beam. The other is to figure out which part of the ion ring contributes the most to the horizontal IBS and investigate whether we can optimize the optics to reduce the horizontal IBS. Probably we can start this after we finish the pCDR draft.

Fast Multipole Method

Generate new data to compare our traceless FMM code for Coulomb kernel with the pyfmmlib. Pyfmmlib is a Python wrapper for a well optimized Fortran FMM lib for Coulomb kernel and Helmholtz kernel. Our code is developed for illustration of the idea. We have been careful to avoid redundant calculations, but the code has not been fully optimized. The numerical results show that with a lower tensor rank for relative error about 1e-4 or larger, our code is faster. But as the tensor rank goes up, the pyfmmlib outperformed our code. This is as expected since the Cartesian tensor FMM has a higher order dependence on the rank than the solid harmonic FMM. Although our work reduce the order dependence from O(n6) to O(n5), it does not change the conclusion in general. But The Cartesian tensor FMM can be used for simulations with moderate accuracy requirement. The Cartesian tensor FMM also has higher flop rate than other FMMs. As the bandwidth becomes more expansive. The Cartesian tensor FMM may be preferred in future. I add the new data and some discussion of the Cartesian tensor FMM to the FMM manuscript and submitted the revised one to JCOMP.

Investigate a few math parsers and found one called metl, which allows to parse the expression into a function with user defined variables. I have tested this parser with my tpsa and da code. It supports the DAVector data type without problem. This means our kernel-independent FMM code will be able to treat different kernels without recompiling. I started to write a paper on the kernel-independent FMM. Now I am writing on the analytical part. Will need more numerical result to complete the paper.

Work plan in the following two weeks

Write for pCDR. Write some code (probably in Python) to solve the dynamic equations for dust particle transfer. Generate some data for kernel-independent FMM and write a paper on it. Benchmark and add the turn-by-turn tracking model into JSPEC.




Diagnostic Development - K. Jordan



LERF - S. Benson
April 30-May 11

  • Worked on preparing the LERF for the isotope visit
  • Hosted Cooler ring meeting May 9th (no quorum for the meeting on the 2nd) and wrote up minutes
  • Prepared and presented talk for IOTA/FAST workshop and held discussions with NIU grad students about the FOA work.
  • Started putting together AWP for FOA proposal
  • Hosted visit from Steve Peggs for CBETA discussions

May 14-25

  • Arrange for collaboration with HZDR for Elbe II facility.
  • Complete AWP for FOA proposal
  • Derive draft budget for isotope production
  • Revise LOD to match new ASE document.
  • Continue cleaning up the LERF for isotope visit.
  • Attend teleconference with DOE for isotope project.