We have finished our extensive studies of plateaus for various source
separations. Our current running plans are to use (thanks to a useful
suggestion by Norman Christ) periodic (P) and antiperiodic (A) boundary
conditions in the following way. We will create propagators with P and A BCs
at t=0 on our lattices. We then will generate the two combinations
Prop1 = 0.5*(P+A)
Prop2 = 0.5*(P-A)
so that Prop1 propagates in the positive t direction while Prop2
propagates in the negative t direction. This effectively doubles the
time direction of the lattice while not requiring a doubling of our
computation need (as would be needed if we used a second source not
at t=0). Tests have shown that this technique is free from any large
wraparound effects (since Teffective is twice the actual T
direction), and also gives a rather nice plateau.
Currently we have begun production running on the MILC 0.007 (a=0.125) ensemble
with valence masses = 0.007 , 0.01 , 0.02 , 0.03 , 0.05.