search

etmc / tmLQCD

tmLQCD is a freely available software suite providing a set of tools to be used in lattice QCD simulations. This is mainly a HMC implementation (including PHMC and RHMC) for Wilson, Wilson Clover and Wilson twisted mass fermions and inverter for different versions of the Dirac operator. The code is fully parallelised and ships with optimisations for various modern architectures, such as commodity PC clusters and the Blue Gene family.
http://www.itkp.uni-bonn.de/~urbach/software.html
GNU General Public License v3.0
32 stars 47 forks source link

QUDA interface / QUDA-MG: study how the MG behaves when theta is changed #441

Open kostrzewa opened 5 years ago

kostrzewa commented 5 years ago

Currently, the QUDA interface does not track how the gauge field and MG setup are updated when the theta angles are changed in TBC calculations. This was never an issue before because different angles were done in different runs, but we now need to add facilities to the code to track changes of boundary conditions and update the fields accordingly. In the worst case scenario, the MG setup will have to be discarded and redone...

sunpho84 commented 5 years ago

In the nissa-MG intergace I store a checksum of the configuration used for setup

https://github.com/sunpho84/nissa/blob/aef9815de954e01030aed42bcac784b23b0d285d/src/base/DDalphaAMG_bridge.cpp#L142

and then I compare each time the used configuration. If they differ, a new setup is issued. Maybe it's an overkill because with small angles the setup might work, but in general it's needed

kostrzewa commented 5 years ago

Yes, for inversions I also thought of doing it this way but it doesn't generalise well to the HMC where updates are done less frequently and one has to have a notion of "distance between configurations". Ideally, in order not to complicate matters too much, I would of course like to use the same mechanism in the HMC as I do for simple inversions. I have to give it some thought.