1 ucla plans 04-06 warren b. mori john tonge michail tzoufras university of california at los...
Post on 21-Dec-2015
228 views
TRANSCRIPT
1
UCLA Plans 04-06
Warren B. Mori
John Tonge
Michail Tzoufras
University of California at Los Angeles
Chuang Ren
University of Rochester
Particle models are needed (Rochester MHD Simulation Data)
PIC can be used from nc (1021cm-3) to solid density (1023cm-3)
3
UCLA plan
1. Develop OSIRIS, UPIC, and Dawson Cluster (4.5 Tflops).
2. Use OSIRIS and UPIC to study:
a.Absorption of light in both “global” 2D simulations and reduced scale 3D simulations
b. Transport of energetic electrons and ions in both “global” 2D simulations and reduced scale 3D simulations
3. Use OSIRIS and UPIC to study collisions:
a. Reduce cell size to a fraction of a Debye length to study collisions from first principles (Work with MIT).
b. Study finite size particle collisions for in OSIRIS an UPIC(Work with Rochester). Set the stage for asking the validity of fluid theory.
4. Compare full PIC results against LSP (Work with Rochester and Reno)
5. Continue to develop theory for the “Weibel” instabilityrelevant to fast ignition. We are concentrating on collisionless Weibel. (Work with Rochester).
4
UCLA plan
UPIC:Task I. Add dynamic load balancing into UPIC
Task II. Add open boundary conditions into UPIC.
Task III. Add arbitrary initial density function n(x,y,z) profiles into UPIC
5
UCLA plan
OSIRIS:
Task I. Add open boundary conditions for all directions into OSIRIS.
Task II. Add static load balancing into OSIRIS.
Task III. Add a “core” region in which energetic particles are absorbed properly.
Task IV. Add particle tracking diagnostics.
6
UCLA plan
Code comparison:
Task I. Working jointly with other Center members, come up with a set of runs for comparing output from the various codes being used across the Center.
Task II. Compare results from OSIRIS, UPIC, Reno codes, and LSP on relevant runs.
7
UCLA plan
Physics:Task I. Use 2D and 3D simulations to understand the role
of return current and the ions on the filamentation of the hot electrons.
Task II. Work with experimentalists to prioritize and identify key physics questions that can be studied using PIC.
8
Sample results:Mocking up the core
Possible methods:1.Radial boundary which thermalizes or thermalizes and reflects particles crossing the boundary while maintaining continuity equation. (Implemented -causes large charge build up at boundary) 2.Add position(radial) and velocity dependent drag to particles in core.3.Add position dependent 2 particle collisions to core (perhaps best solution).
Core diagnostics particle energy and direction at the core (in progress)
Parameters for 2D SimulationRen reference
• Vacuum region between target and boundary to reduce boundary effects
• 1203212032 grids (x=0.33 c/p), with current smoothing, 4 particle/cell
• 2.4 108 particles and 6104 steps
• Initial Te=7.4keV and Ti=1 keV e/p0.035
• 1m-laser from left wall antenna, I=1020 W/cm2, spot size (FWHM) 7.5 m, 1 ps long, s-&p-polarized.
100m
40nc
16m
7.5mlaser
25.5mcoronal plasma
10
Results:half the radius of Ren et al.
No core
11
Results:half the radius of Ren et al.
Thermalizing core
12
Dawson Cluster
256 node dual processor G5 x-serve cluster
4.5 Tflops-------2.3 Tflops on Linpack
With static load balancing, each Ren type run takes 3-4 days on 64 nodes on Dawson
M.Tzoufras, F.S.Tsung, J.W.Tonge, W.B.Mori
UCLA
C.Ren
University of Rochester
M.Fiore, L.O. Silva
IST (Portugal)
Emergence of space charge effects in the linear stage of the Weibel like current filamentary instability