hflip not working for COOH ?? does not flip

[jouteurestaque]

hey guys,

I just notice the default launch of crest 3.0 does not seem to do it right for flipping the XH of of COOH function.
here two of an initially xyz file of the same molecule (just flipping the OH of the acid). Clearly crest is not doing the hflip somehow, while it is the default option.

hflip_crest_COOH.zip

[hiSandog]

Could you include the exact CREST version/command, charge and multiplicity, starting XYZ, and the conformer/output that demonstrates the missed flip? The assertion should compare a measurable COOH dihedral before and after hflip rather than visual orientation alone. Testing the same input with only hflip enabled and several random seeds would show whether the move is rejected, never generated, or later deduplicated.