1. Evolving Architecture for Beyond the Standard Model Kihyeon CHO (KISTI) Yonsei Nuclear and Particle Workshop Yonsei University, Seoul, Korea April 29, 2015

2. Contents Beyond Standard Model Simulation Computing Results Summary 2 SimulationBeyond Standard ModelComputing

3. The SM is now complete? 3 Higgs Discovery (July 4, 2012) Beyond Standard Model

4. What is next? Unification forces? Why these masses? Origin of dark mass and dark energy? Matter and anti- matter asymmetry Evolving stars? Astro-nuclear? … 4 Beyond Standard Model

5. The Universe Today Standard Model 5

6. distance (light-year) rotation velocity (km/s) 6

7. 7 Energy Frontier Intensity Frontier Cosmic Frontier P5 Report (2014.5.22) HiggsNeutrino Mass Dark Matter Dark Energ y The Unknown SMBeyond the Standard Model After P5 Report ⇒ Before P5 Report Beyond Standard Model

8. Beyond the Standard Model 8 Beyond Standard Model

9. Simulation 9

10. Simulation for Experiment The primary application running on WLGC - Simulation LHC simulation (Run1) –Several 10 7 volumes, 10 10 events –10 12 sec CPU time using 250,000 cores –60% of WLGC (expected to 65% in LHC Run2) Challenges for High-luminosity LHC –Need at least x5 computing power at current budget ⇒ New architectures 10 S. Y. Jun Simulation

11. HEP Simulation 4 11 (physics) Simulation We focus on MadGraph & Geant4.

12. Vision for HEP Simulation To have a massively parallelized particle transportation engine To comply with different architecture (GPU, MIC and etc.) To draw community interests for collateral effort 12 Simulation

13. 13 Simulation Geant4 is the most successful model in HEP. HEP user community – BaBar(2001), LHC(2003), Belle II Other community: Medical, Space, DNA, Solid Physics

14. ⇒ Evolving Computing Architecture 14 Computing P5 report

15. Evolving Computing Architecture Servers control GPU and MIC. GPU and MIC share memory. Heterogeneous platform 15 Computing

16. S. Y. Jun 16 Computing

17. Results 17

18. 1. HTC module into MadGraph We embed HTC module into MadGraph. ⇒ Korean Economic News (2014.10.31) Then, using it we study BSM. ⇒ arXiv: 1412.1541 [hep-ph] 18 Results

19. 2. Finite Volume Effects on B K 19 ⇒ To reduce the error of B K, we have to calculate the finite volume effect on the lattice. CP violation in Kaon System Calculated using Lattice QCD Results Errors of B K Reference: Kim, Jangho et al. Phys.Rev. D83 (2011) 117501 arXiv:1101.2685 [hep-lat]Kim, Jangho

20. GPU programming using CUDA m_low m_high Results Parallel processing in GPU

21. 21 3. NPR to calculate the matching factor of B K One-Loop NPR (Non- perturbative Renormalization) Results Errors of B K B K on coarse lattice (20 3 X 64) Reference: Hwancheol Jeong, Jangho Kim et al. PoS(LATTICE2014)286 (2014) arXiv:1410.6607[hep- lat] Hwancheol

22. GPU Performance NPR measurement code is optimized for GTX 480. => Parallelization and Optimization significantly Program CPUGPU GPU vs. CPU CPU Spec. GFLOPS VGA (Peak Performance in double precision) GFLOP S Optimi- zation Xeon E5-2620 0.5 GTX 480 (168 GFLOPS) 64.338%128.6 Non- perturbative Renormalizati on(NPR) measurement Core i7- 4820K 1.13 GTX 480 (168 GFLOPS) 66.640%58.9 GTX 580 (198 GFLOPS) 76.1967.2 GTX Titan Black (1707 GFLOPS) 113.36100.3 Results

23. Summary Physics goes beyond discovery. Computing needs solutions for the evolving architecture. ⇒ To fulfill the gap between physics and computing, we need to focus on simulation R&D. 23 Summary

24. Acknowledgement Dr. Soo-hyeon Nam Dr. Junghyun Kim Dr. Jangho Kim Dr. Soon Yung Jun Prof. Weonjong Lee 24

25. Thank you. 25