AMBER Archive (2005)Subject: AMBER: MMGBSA and dielectric constant with AMBER8  dielc and intdiel
From: Fabien Cailliez (Fabien.Cailliez_at_ibpc.fr)
Date: Tue Feb 08 2005  11:16:51 CST
Dear all,
I have a question about dielectric constant in MMGBSA calculation with
AMBER8.
When I tried to put intdiel=2, I had a message that said me thta intdiel
must be equal to dielc.
Thus I put dielc=2 too, but I am confused with the results I obtained.
Here is an example of one calculation with a dielectric constant of 1
and 2 :
***** dielc and intdiel=1
1
BOND = 674.6114 ANGLE = 1662.9186 DIHED =
2067.4398
VDWAALS = 1487.8876 EEL = 15883.4589 EGB =
3734.9502
14 VDW = 754.7371 14 EEL = 9784.9053 RESTRAINT =
0.0000
surface area = 12257.6574
***** dielc and intdiel=2
1
BOND = 674.6114 ANGLE = 1662.9186 DIHED =
2067.4398
VDWAALS = 1487.8876 EEL = 3970.8647 EGB =
924.1456
14 VDW = 754.7371 14 EEL = 2446.2263 RESTRAINT =
0.0000
surface area = 12257.6574
As you can see, when I put dielc and intdiel=2, I have all the
electrostatic contributions (EEL, 14 EEL, and GB) divided by
approximately 4.
I thought that the results shoud be divided by approximately 2 (and not
4). I guess that dielc should not be set to 2 if
intdiel=2.
Am I right or wrong ? Should the energies be divided by 2 or 4 when
dielectric constant is multiplied by 2 ?
Thanks for your help,
Fabien
PS : I went through the code and I understood that when dielc is set to
2, the charges are divided by sqrt(2), so this is one
division by 2. For GB, there is another division by intdiel in egb.f, so
that makes the whole thing divided by 4. This is the same
for 14EEL in subroutine ephi. But I could not find where EEL is
calculated (somewhere in ewald_force I guess)

__________________________________________________________________
Fabien Cailliez Tel : 01 58 41 51 63
Laboratoire de Biochimie Théorique email : cailliez_at_ibpc.fr
IBPC 13, rue Pierre et Marie Curie
75005 Paris
__________________________________________________________________

The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu
