[Towhee-users] Towhee chemical potential changes again

[Marcus M. <mar...@us...>]

Hello Towhee Users,

After productive discussions with Michael Greenfield and Ilja Siepmann
I finally understand how to properly normalize the bond length
distribution.  There were some subtle points relating to a molecule
having a single translational degree of freedom, and one way to
enforce that is to impose a maximum allowable bond length.  This has
been added to version 6.1.0 as an optional variable in towhee_input
(with a default value of 3.0) and that value is now used to normalize
the bond length distribution, instead of the previous unnormalized
version.

Therefore, the measured chemical potential of any molecule that has
one or more bonds will now be different.  Grand Canonical folks will
need to recompute.  Please note that changing the value of the maximum
bond length WILL change the measured chemical potential.  Pick a value
and stick with it.

On the bright side, all of the terms are now normalized properly so I
have hopefully run out of ways to modify the chemical potential in the
future.  Therefore, this should be the last time a recomputation of
the chemical potential is required.  Please note that the new maximum
bond length value is enforced in all of the bond potentials so make
sure to set it to a sufficiently large value not to interfere with
your simulations.  Hopefully the default value of 3.0 is large enough
for almost all "realistic" simulation potentials, but folks working
with "toy" models have the option to reset this value.

I am continuing to quantify the extent of the previously disclosed bug
that affected fixed-endpoint and cyclic CBMC.  It appears the
acceptance rate for those moves is now around half of what it was
previously.  This likely means the versions prior to 6.0.5 were
producing some magnitude of errors in the bond, angle, or dihedral
distributions for those molecule.  Once some of my longer test
simulations have completed I will comment with some specific data.

Many of the variable options in towhee_input have now been made
optional with default values that are either the suggested ones, or
ones that disable the feature.  The main goal was to free the users
from including variables whose only purpose was to turn off the
feature related to the variable.

Marcus

-- 
Marcus G. Martin
Director, Useful Bias Incorporated
88 Martinez Road
Edgewood NM 87015-8222
ph. (505) 286-4457
www.usefulbias.com
www.photobirder.com