xtb
xtb copied to clipboard
$FIX: not respecting fixation in xcontrol file anymore (v.6.7.1)
Describe the bug When trying to optimize only H-Atoms, xtb does not respect the fixation of atoms in place anymore. I tested also 6.7.0, where this issue is not occurring.
To Reproduce
xtb glycineAcid.xyz --input glycineAcid.xcontrol --opt
The xyz file:
10
C 0.0834925009735182 0.0000000000000000 -1.6668515099899917
O 0.0834925009735182 -1.1445070789971263 -2.0630581700150108
O 0.0834925009735182 1.0691091199790088 -2.4983246289908534
C 0.0834925009735182 0.4230545059847334 -0.2020795040210518
H 0.9565438779960600 1.0664427810136214 -0.0395754850004047
N 0.0834925009735182 -0.6677380050243458 0.7545830369902874
H 0.8900621800066862 -1.2628578009803400 0.5783254700023934
H -0.7230771790121688 -1.2628578009803400 0.5783254700023934
H -0.7895588770015425 1.0664427810136214 -0.0395754850004047
H 0.0834928001703133 -0.2969331876126342 1.6184980029345206
The xcontrol file:
$samerand
$fix
elements: C, N, O
$end
Expected behaviour No movement in the atoms except for H-Atoms.
Output shows:
Lowest eigenvalues of input Hessian
0.000000 0.000000 0.000000 0.000000 0.000000 0.000000
0.012712 0.016253 0.017576 0.047909 0.064565 0.074817
0.090051 0.112446 0.112874 0.127281 0.155430 0.226076
Expected (and in older versions):
Lowest eigenvalues of input Hessian
-0.000000 -0.000000 -0.000000 -0.000000 -0.000000 -0.000000
0.000000 0.000000 0.000000 0.000000 0.000000 0.000000
0.000000 0.000000 0.000000 0.000000 0.000000 0.000000
This bug seems to be related to the default Approximate Normal Coordinate Rational Function optimizer (ANCopt) in xtb version 6.7.1. A way to still do the exact fixing with this xtb version is to use another optimization engine like L-ANCopt or FIRE.