COLVARS metadynamics: first timestep depends on condition

Dear users, good evening,
I faced the unobvious behavior of COLVARS metadynamics output. Could you explain it to me, please?

I use metadynamics from COLVARS package and notice different behavior depending on parameter “newHillFrequency”:
When I set it to 80, the first hill in file “.state” has the same timestep as the first line in file “.traj”. (I have 70k steps of thermalization before metadynamics, so timestep is 70000)
But when I set newHillFrequency to 160, the first hill in “.state” is 70080 instead of 70000. The next one is 70240, it means that interval is 160, as expected.
The first line in file “.traj” remains 70000. Both scripts were generated automatically from one pattern, so “newHillFrequency” is the only difference.

This feature is not crucial for work, but I’m wondering - what could be the reason for such difference?
Thanks for your kind attention!

Sincerely,
Karen Fidanyan,
Moscow Institute of Physics and Technology

Dear users, good evening,
I faced the unobvious behavior of COLVARS metadynamics output. Could you
explain it to me, please?

I use metadynamics from COLVARS package and notice different behavior
depending on parameter "newHillFrequency":
When I set it to 80, the first hill in file "<fname>.state" has the same
timestep as the first line in file "<fname>.traj". (I have 70k steps of
thermalization before metadynamics, so timestep is 70000)
But when I set newHillFrequency to 160, the first hill in "<fname>.state"
is 70080 instead of 70000. The next one is 70240, it means that interval is
160, as expected.
The first line in file "<fname>.traj" remains 70000. Both scripts were
generated automatically from one pattern, so "newHillFrequency" is the only
difference.

​when starting from 70000, the first time step divisible by 80 is 70000,
but the first step divisible by 160 is 70080. no mystery at all. if you
want to change this behavior, you have to use the reset_timestep command to
reset it, e.g. to 0, before defining the colvars fix.

axel.​