# [gmx-users] pull=constraint gives zero forces

Thomas Schlesier schlesi at uni-mainz.de
Thu Oct 11 10:21:55 CEST 2012

See the SHAKE algorithm in the manual.
Especially equation (3.98)
-> G_i = \sum_k \lambda_k * (\partial \sigma_k)/(\partial r_i)
where G_i is the constraint force
\sigma_k are the equations for the constraints
and \lambda_k is the lagrange multiplier

'Understanding molecular simulation' (D. Frenkel, B. Smit)
gives these equations for bond-constraints as
\sigma = r_ij^2 - d_ij^2
where r_ij is the real distance, and d_ij to one to which one wants to
constrain the bond length

from this \lambda_k would correspond to a force constant.

Greetings
Thomas

Am 10.10.2012 18:41, schrieb gmx-users-request at gromacs.org:
> How can there be forces for holonomic constraints? Is this described by an equation in the manual?
> Just because there are values in the pullf.xvg file does not mean that these values are forces.
> If they are forces, what is the force constant and what is the equation that defines this force?
>
> Thank you,
> Chris.
>
> -- original message --
>
> But for GMX 4.0.7 there are forces in the pullf.xvg. The forces which
> arise rom the contraint the hold the two groups fixed. I use them for
> thermodynamic integration...
>
> I use the following mdp-parameters, probably this gives you an idea what
> you might make different:
> ; AFM OPTIONS
> pull                =  constraint
> pull_geometry       =  distance
> pull_dim            =  Y Y Y
> pull_start          =  yes
> pull_nstfout        =  10
> pull_nstxout        =  500000
> pull_ngroups        =  1
> pull_group0         =  REF
> pull_group1         =  ZUG
> pull_rate1          = 0.000
> pull_init1          = 0.000
> pull_constr_tol          = 1e-06
>
> But as Chris said, better post the complete mdp-file (probably for both
> GMX 4.0.7 and 4.5.x) so we can comment on them.
>
> Grettings
> Thomas