| Issue |
EPJ Web Conf.
Volume 175, 2018
35th International Symposium on Lattice Field Theory (Lattice 2017)
|
|
|---|---|---|
| Article Number | 02002 | |
| Number of page(s) | 8 | |
| Section | 2 Algorithms and Machines | |
| DOI | https://doi.org/10.1051/epjconf/201817502002 | |
| Published online | 26 March 2018 | |
https://doi.org/10.1051/epjconf/201817502002
Multigrid accelerated simulations for Twisted Mass fermions
1
Department of Physics, University of Cyprus, PO Box 20537, 1678 Nicosia, Cyprus
2
Fakultät für Mathematik und Naturwissenschaften, Bergische Universität Wuppertal, Wuppertal, Germany
3
Computation-based Science and Technology Research Center, The Cyprus Institute, Nicosia, Cyprus
* Speaker, e-mail: s.bacchio@hpc-leap.eu
Published online: 26 March 2018
Simulations at physical quark masses are affected by the critical slowing down of the solvers. Multigrid preconditioning has proved to deal effectively with this problem. Multigrid accelerated simulations at the physical value of the pion mass are being performed to generate Nf = 2 and Nf = 2 + 1 + 1 gauge ensembles using twisted mass fermions. The adaptive aggregation-based domain decomposition multigrid solver, referred to as DD-αAMG method, is employed for these simulations. Our simulation strategy consists of an hybrid approach of different solvers, involving the Conjugate Gradient (CG), multi-mass-shift CG and DD-αAMG solvers. We present an analysis of the multigrid performance during the simulations discussing the stability of the method. This significant speeds up the Hybrid Monte Carlo simulation by more than a factor 4 at physical pion mass compared to the usage of the CG solver.
© The Authors, published by EDP Sciences, 2018
This is an Open Access article distributed under the terms of the Creative Commons Attribution License 4.0, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. (http://creativecommons.org/licenses/by/4.0/).
Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.