choreography in gravity
play

Choreography in Gravity Simon Portegies Zwart, Jeroen Bedorf, - PowerPoint PPT Presentation

Choreography in Gravity Simon Portegies Zwart, Jeroen Bedorf, Evghenii Gaburov, Tjarda Boekholt, Michiko Fujii, Tomoaki Ishiyama, Keigo Nitadori An ape on the shoulders of a giant, still is an ape. Larger comuters require more complex


  1. Choreography in Gravity Simon Portegies Zwart, Jeroen Bedorf, Evghenii Gaburov, Tjarda Boekholt, Michiko Fujii, Tomoaki Ishiyama, Keigo Nitadori

  2. An ape on the shoulders of a giant, still is an ape.

  3. Larger comuters require more complex software. But does this lead to a better understanding? ● More scales leads to more complexity ● More physics leads to more complexity ● More complexity leads to less understanding

  4. Therefore Bonsai Small & resilient

  5. Leiden LGM 400GPUs=0.5PFflops Tsukuba 4GPUs = 0.005PFlops 40 GPUs=0.05PFlops CSCS Piz Daint 4000GPUs=5PFflops ~20000GPUs= 25PFflops ORNL Titan

  6. HPC on Titan's GPU-farm

  7. 2 ×  8 years  × 100 steps × 60 operations 10 years 10 11 Stars   10 2.5 × 10

  8. Jeroen Bédorf etal: simulation of Andromeda/Milky Way encounter on Titan

  9. Bonsai simulation Hoag's object (HST)

  10. Being able to perform large calculations is not the same as being able to perform accurate calculations ● “ Errors in calculations of n-body systems grow exponentially … and may therefore invalidate the results ... ” (Miller 1964)

  11. BRUTUS a brute force arbitrary-precision N-body code ● Two ingredients: ● Gragg-Bulirsch-Stoer method – Modified midpoint method – Richardson extrapolation – Tolerance parameter ● Arbitrary-Precision arithmetic – Number of significant digits Tjarda Boekholt 12

  12. Red: dE/E <10 -74 Black: dE/E <10 -11

  13. Exponential divergence δ = 0.5 log 10 1/(6N) ∑ (x 2 -x 1 ) 2 + (v 2 -v 1 ) 2 14

  14. 10,000 realizations of N=3 give no systematic bias 15

  15. Conclusions ● While computers get bigger software gets more complex ● Continuing trend in more demanding simulations ● By keeping codes small and dedicated one gains flexibility, strength and tunable performance 16

Recommend


More recommend