From: Giacomo Fiorin (giacomo.fiorin_at_gmail.com)
Date: Mon Dec 10 2012 - 16:29:52 CST
Hello Jian, with k = 1 kcal/mol/Å^2 the harmonic potential 1/2 k (d -
d_low)^2 gives you only 2 kcal/mol of energy cost if |d - d_low| = 2 Å.
With only 2 kcal/mol of energy cost, it is absolutely not a surprise that
the colvar goes outside the boundary.
I see that you set the width to 0.1: in case you're in doubt, that will
only affect the force constant of the dedicated (multidimensional) harmonic
bias that you're not using.
Yes, increase the force constant of the boundary potential.
Giacomo
On Mon, Dec 10, 2012 at 5:15 PM, DAI, JIAN <jdai2_at_fsu.edu> wrote:
> Dear NAMD fellows:
> We are calculating the free energy landscape of a protein using two
> distance restraints that describe the distance between two domains of the
> protein.
> The problem is: I noticed that the boundary values of the colvars have
> constantly exceeded the restraint values. Although the force constants that
> are put on these distances are small, i.e., 1 kcal/mol A^2, the distances
> should not exceed the upperboundary and lowerboundary too much, shouldn't
> they? In the simulation, for example, when the boundary is set at
> lowerboundary 9.0 and upperboundary 10.0,
> the actual value of the second colvar drops down to even 7.3, which I
> think is way too low.
> The configuration and colvars definition files are attached. Should we
> expect to see that these distance values exceed the boundary values or a
> stronger force constants will solve this problem? However, I've seen in
> another work that also uses a force constant of 1 (Wereszczynski and
> McCammon, 2012, PNAS).
> Thanks in advance.
> Jian
This archive was generated by hypermail 2.1.6 : Mon Dec 31 2012 - 23:22:21 CST