Speaker
Roman Gruber
(ETH Zürich)
Description
We present a new take on low-mode averaging, where the dimension of the low-mode subspace is multiplied by exploiting local coherence of low modes. The fraction of the quark propagator captured by this subspace can easily be volume averaged or sampled excessively and reaches gauge variance with lower computational cost than the traditional methods. The remainder piece can be sampled stochastically by solving the Dirac equation to low precision without compromising the correctness of the solution.
| Topical area | Algorithms and Artificial Intelligence |
|---|
Primary authors
Marina Marinkovic
(Trinity College Dublin)
Roman Gruber
(ETH Zürich)