Re: [AMBER] Doubt about EVB/LES-PIMD

From: David A Case <>
Date: Thu, 18 May 2017 09:30:40 -0400

On Wed, May 17, 2017, Stregone wrote:
> I would like to perform a EVB/LES-PIMD simulation in order to estimate
> kinetic isotope effects.
> The reaction is SN2 and 2 atoms + 2 molecules are involved in the reaction

Just some background here: the EVB/PIMD code was originally designed to do
exactly things like you describe. But it's no longer under active
development, and I'm not sure if many (any?) people are still using this; or
if they are, that they follow this email list.

> Do I need to freeze the the atoms involved in the reaction in order to
> prevent them to spread along the periodic box or it does not matter ?

Generally (for any type of simulation) on generally uses restraints on the
solute during equilibration of the solvent. I recommend using ntr=1, not
ibelly=1, for this. I'd recommend this generally, but particularly for
something like PIMD, whose interaction with ibelly is probably untested.

> Before calculating kinetic isotope effects, I need to run a PIMD
> simulation (based on the manual AMBER 14). Then I can perform 2
> separated thermal integrations. 1st the EVB/LES-PIMD in the reactant
> region, where clambda change from 0 to 1 (I have experience in this
> part). However the 2nd calculation, along the dividing surface (TS)
> is not quite clear for me. Based on what I understood (pag. 468), in
> the second TI , I already should have found the dividing surface (TS).
> Which means that before performing such simulation, I would need to run
> Path integral quantum transition state theory, and only then, I could
> run the 2nd TI, along the dividing surface (TS).

Finding the dividing surface in a rigorous fashion can be really tough. I'd
recommend starting with umbrella restraints that get you at a reasonable
(albeit approximate) transition state, then do the second TI calculation
there. See if the results make sense. You might use trial and error to
see how sensitive the computed KIE's are to your choice of TS.

If you need a more rigorous way to define the dividing surface, you are
probably going beyond the capabilities of the current Amber codes.

....hope this helps....dac

p.s.: it would be great if someone updated this aspect of things. If you get
something to work, and are willing to share details, it would be great to
make a new tutorial.

p.p.s: cc-ing to Darrin York, whose group knows a lot more about this than
I do, and to Jiri Vanicek, whose was the driving force behind the original
work, and is still active in this area.

> Or , during the 2nd TI, the Path integral quantum transition state theory will be performed along with
> isotope calculation. Quote " Sampling of dVeff /dλ along the dividing surface is invoked in a similar fashion,
> but with ground-state dynamics replaced by biased sampling constrained to the dividing surface"
> Because it gives the sense of performing all together. In that case the commands would look like this ???
> evb_dyn = “dbonds_umb”, dbonds_umb(1)%iatom = 8, !TS
> dbonds_umb(1)%jatom= 9, dbonds_umb(1)%katom = 7, !TS
> dbonds_umb(1)%k = 400.000, dbonds_umb(1)%ezero = 0.0 !TS
> ievb=1, ipimd=2, ntt=4, nchain=4, itimass=1, clambda=0.2, !IE
> Thanks
> Best Regards,
> _______________________________________________
> AMBER mailing list

David A. Case                         |
Dept. of Chemistry & Chemical Biology |
Rutgers University                    |    office:   +1-848-445-5885
174 Frelinghuysen Road, Rm. 208b      |    cell:     +1-609-751-8668
Piscataway, NJ 08854        USA       |
AMBER mailing list
Received on Thu May 18 2017 - 07:00:03 PDT
Custom Search