From: Brian Radak (brian.radak_at_gmail.com)
Date: Fri Aug 24 2018 - 10:16:30 CDT
I would probably not recommend using the constant-pH MD code in NAMD unless
you actually want that functionality. The extra file system calls can be
rather expensive, especially for large systems and even when the switch
times are short (in fact it might actually be worse then).
The final PSF/COOR files from a constant-pH MD simulation will represent
the latest protonation states from the trajectory as well as include dummy
atoms for those sites which are not interacting. This formally changes the
dynamics of the system, but most thermodynamic properties are rigorously
unperturbed (similar to hydrogen mass repartitioning).
It is fairly straightforward to "clean up" the dummy atoms using psfgen
with the "readpsf" and "delatom" commands, but this currently must be done
by hand. Unless you are really attached to the structures that you get from
constant-pH MD, it might just be easier to build the system anew and assign
the protonation states that you want.
On Fri, Aug 24, 2018 at 3:36 AM Yasser Bruno Ruiz Blanco <
> Hi all,
> Here is a probably odd question:
> What difference there might be, according computational performance and
> physical description of the system, between a standard MD simulation and a
> constant pH simulation with 0 (or a very very low number) of switching
> steps. (i.e. the switching will always give an energy that most probably
> lead to a non acceptable change)
> I am looking for the easiest way of continuing -directly- a standard MD
> from the final state of a constant pH MD. But this can only makes sense if
> the computational performance will not suffer too much and in no systematic
> perturbation is introduce to the velocities or energy of the system after
> the "unsuccessful" switching attempts.
This archive was generated by hypermail 2.1.6 : Fri Oct 18 2019 - 23:19:51 CDT