Fourier acceleration runs and results.
facc-070607, facc-070608 Preliminary.
facc-070610, facc-070612 Study for Lattice 07.
1. Parameters
kap=0.1249, lam=1e-4, y=0.05, nf=3*2, nx=(12,12,12,24), vev~~1.0, dt=0.5
2. AC vs m_FA, ns=05 - figure, data
3. AC vs m_FA, ns=10 - figure, data
4. Force profiles not calculated.
facc-070827 Double-check Dani's large lattices.
Results are not pretty. Screwed by the erroneous k=0 correction to mc.
facc-070906 *The* study.
1. Parameters
kap=0.122195, lam=1e-4, y=0.35, nf=3*1, nx=*, vev~~0.85, dt=*, ns=5*10,
dt adjusted for each run to give r_acc~~80%.
2. m=0.5x has k=0 "correction" to mc -- discard them.
3. AC vs L - scaling.png, scaling.dat
AC measurements for m=0.5 are pretty good; but give critical exp't of
1.8(2) -- sigh. AC measurements for m=inf are not so good; these too
yield a critical exp't of 1.8(2). The fact that for some of the m=inf
cases, there are substantially non-zero correlations at very large t
(e.g., G(1000) = 0.20(2)) can only mean that large autocorrelations have
far-reaching effects?
facc-071015 Figure out the k=0 problem.
determined that problem was only in calculation the force profile -- *not*
in the MD. It is essential to add Fpsi and Fphi before taking absolute
values. In fact, it is probably best to monitor dphi and dpc step-wise.