[gmx-users] Re: Computing the total electrostatic energy (short range, 1-4, and Ewald) of an atom

[Andrew DeYoung]

Greetings,

Thank you so much, Mark!

Do you know if it is possible to extract just the real-space part (not the
reciprocal space part) of the Ewald summation?  (By "Ewald summation," I
mean either the standard Ewald summation or PME.)

Thank you very much! 

Andrew DeYoung
Carnegie Mellon University

> Hi,
> 
> Is it possible to compute the total electrostatic energy of an atom over
> time?  By "total," I mean including all of the terms (short range Coulomb,
> 1-4 Coulomb, and Ewald), not just the short range Coulomb and 1-4 Coulomb.
> 
> This question is somewhat related to a question I asked last week (
> http://lists.gromacs.org/pipermail/gmx-users/2012-March/069745.html ), and
> to which I received a great answer.  But my specific question now is: is
it
> possible to calculate the Ewald term due to a particular atom?  Or is it
> impossible to decouple the Ewald calculation among individual atoms or or
> sets of pairs of atoms? 
> 
> For example, I would like to determine the electrostatic energy between
atom
> 455 and the "rest" of the system (i.e., all other atoms in the system).
To
> try this, I added an entry "a" to my .ndx file:
> 
> [ a ]
> 455
> 
> I then added a directive to my .mdp file:
> 
> energygrps = a
> 
> I then run grompp and mdrun, passing my .ndx file to grompp.  (Note: in my
> .mdp file, I specify that energies should be saved every timestep:
nstenergy
> = 1.)  Then I run g_energy on the .edr file that resulted from mdrun.  I
am
> given the following energy terms to extract: 
> 
> ---
>   1  Bond             2  Angle            3  Ryckaert-Bell.   4  LJ-14
> 
>   5  Coulomb-14       6  LJ-(SR)          7  Coulomb-(SR)     8
> Coul.-recip.  
>   9  Potential       10  Kinetic-En.     11  Total-Energy    12
> Conserved-En. 
>  13  Temperature     14  Pressure        15  Constr.-rmsd    16  Vir-XX
> 
>  17  Vir-XY          18  Vir-XZ          19  Vir-YX          20  Vir-YY
> 
>  21  Vir-YZ          22  Vir-ZX          23  Vir-ZY          24  Vir-ZZ
> 
>  25  Pres-XX         26  Pres-XY         27  Pres-XZ         28  Pres-YX
> 
>  29  Pres-YY         30  Pres-YZ         31  Pres-ZX         32  Pres-ZY
> 
>  33  Pres-ZZ         34  #Surf*SurfTen   35  Mu-X            36  Mu-Y
> 
>  37  Mu-Z            38  Coul-SR:a-a     39  LJ-SR:a-a       40
Coul-14:a-a
> 
>  41  LJ-14:a-a       42  Coul-SR:a-rest  43  LJ-SR:a-rest    44
> Coul-14:a-rest
>  45  LJ-14:a-rest                        46  Coul-SR:rest-rest
> 
>  47  LJ-SR:rest-rest                     48  Coul-14:rest-rest
> 
>  49  LJ-14:rest-rest                     50  T-System
> ---
> 
> Choices 42 and 44 are two parts of what I would like: short range Coulomb
> and 1-4 Coulomb energy terms for a-rest.  But is it possible to somehow
also
> calculate the Ewald term for a-rest?

 
Not easily, and the value of it would be debatable even if you could compute
it. Someone did post a several-step method a few years ago using mdrun
-rerun and perhaps tpbconv -zeroq that achieved something like the
decomposition of the reciprocal-space term you seem to be after.
 
Mark