Smearing

GLU has support for a wide range of smearings.

This is all kicked of from this being set in the input file

MODE = SMEARING

Vanilla smearings

Of course there is support for the common (less so with LOG) smearings:

   SMTYPE = {APE,STOUT,LOG}

APE https://inspirehep.net/literature/21836

STOUT https://arxiv.org/abs/hep-lat/0311018

LOG https://arxiv.org/abs/0709.4110

And the strength is determined by the single parameter

   ALPHA_1 = 0.75

To bring all the definitions together (they are really very similar) I always have this conversion:

   rho = ALPHA_1/(2*(ND-1))

Now, I also support smearing in the full ND-directions or the Spatial (ND-1) directions via the input file parameter

   DIRECTION = {ALL,SPATIAL}

And we have the following max number of iterations

   SMITERS = 200

Hypercubic blocking

Of course this is just a schema so we support the above smearings in hypercubic form:

HYP https://arxiv.org/abs/hep-lat/0103029

HEX

HYL

These have different smearing parameters in the input file

   ALPHA_1 = 0.75
   ALPHA_2 = 0.50
   ALPHA_3 = 0.25

And their respective logical normalizations (2(ND-1),2(ND-2),2(ND-3)... etc)

One thing that is interesting is that we actually support generic ND lattices via recursion here. It is slow but it works!

Gradient flow

As we support Stout smearing it is not complicated to support the gradient flow as well and we have 4 flavors

   SMTYPE = {WFLOW_STOUT,WFLOW_LOG,ADAPTWFLOW_STOUT,ADAPTWFLOW_LOG}

The first two are fixed iteration methods based on the strict value of the parameter

   ALPHA_1 = 0.02

And the other two are two-step adaptive routines of my own design. I would recommend using only the ADAPTWFLOW_STOUT routine. This will chug along and finally (hopefully) spit out the t_0 and w_0 scales of the clover and plaquette definitions of the field.