AMBER Archive (2004)

Subject: AMBER: one question about dielectric dependent model

From: Xiaowei (David) Li (
Date: Sun Dec 05 2004 - 11:42:56 CST

Dear AMBER users:
It is widely used technique to mimic the the influence of solvent using
implicit solvent(Generally, water). In AMBER, my understanding is that
there are two methods to use implicit solvent. One is Generalized-Born
(GB) model , and the other is to use distance-dependent dielectric
model. Beause there is no GB model implemented in NMODE, so I use
distantce-dependent dielectic model when I perform normal mode analysis
of nucleic acids. My question is concering about the distance-dependent
model in AMBER.
Based on the Debye-Hückel model, the screened Coulomb potential (phi) is
phi=(Z*q)/(epsilon *r) exp(-kr), here the 1/k is the D-H length, epsilon
is the dielectric constant in vacumm.
Therefore, the effective dielectric constant is (epsilon * exp(kr)). If
|kr|<1, then exp(kr) is approximated as (1+kr). So the effective
dielectric constant is (epsilon*(1+kr)).
In NMODE, we can set dielc=1-4 and idiel=0 to get the effective
dielectric constant as (1-4)*r, which means kr>>1 and the dielectric
constant is distant dependent. However, if the amount of water is very
small, then k is very close to 0. Thus, we should set the dielectric
constant as (epsilon*(1+kr)), instead of (epsilon*kr). But it seems to
me that there is no options in AMBER to set the dielectric constant as
(epsilon*(1+kr)). In other words, we can set constant dielectric or
distant-dependent dielectric constant in AMBER. But there is
discontinuity between these two cases. This is my understanding about
this question. If I am wrong, please feel free to correct me. Thank you
very much.


The AMBER Mail Reflector
To post, send mail to
To unsubscribe, send "unsubscribe amber" to