After 31 + 28 iterations the minimization steps are further away from equilibrium than near the start:
@ Step Energy Delta E Gmax Grms Xrms Xmax Walltime
@ ---- ---------------- -------- -------- -------- -------- -------- --------
@ 0 -368.88619110 0.0D+00 0.20456 0.08812 0.00000 0.00000 14219.7
@ 1 -368.89189741 -5.7D-03 0.04448 0.00724 0.00193 0.01113 18205.2
@ 2 -368.88937811 2.5D-03 0.06379 0.02765 0.00889 0.05178 23054.9
@ 3 -368.89179122 -2.4D-03 0.03417 0.00845 0.00417 0.02792 27977.7
@ 4 -368.88757045 4.2D-03 0.03445 0.00904 0.00034 0.00199 32570.5
@ 24 -368.85420735 -1.8D-05 0.01990 0.00918 0.00033 0.00320 74232.9
@ 25 -368.85402214 1.9D-04 0.01896 0.00880 0.00037 0.00363 78316.2
@ 26 -368.85126484 2.8D-03 0.00764 0.00362 0.01301 0.12692 81890.3
@ 27 -368.85320622 -1.9D-03 0.01442 0.00698 0.00573 0.04692 85842.4
@ 28 -368.85003684 3.2D-03 0.00987 0.00364 0.01486 0.14546 89606.5
Note that -368.85 > -368.89.
Could there be an incompatibility between the lmbfgs algorithm and the R-3c space group? How do I report a possible bug? Or am I doing something wrong.
My input deck follows:
title "NaNO3 testing - using R-3c symmetry"
echo
memory 4000 mb
permanent_dir ./perm
scratch_dir ./scratch
geometry nocenter noautosym noautoz print
system crystal
lat_a 5.071
lat_b 5.071
lat_c 16.825
alpha 90.0
beta 90.0
gamma 120.0
end
symmetry R-3c
Na 0.00000 0.00000 0.00000
N 0.00000 0.00000 0.25000
O 0.2448 0.00000 0.25000
end
set nwpw:cif_filename NaNO3
|