# Re: Re: vmd-l: Interactive Simulations, ligand size limits....

Date: Fri Apr 29 2005 - 04:01:27 CDT

Hi John (and others),

>>> John Stone said:
>> Brian,
>> By its very nature, adding forces through IMD will add energy to
>> the simulation, so you won't have conserved energy regardless of
>> of energy conservation that I'm missing, you're not going to get that
>> from an SMD or IMD type of simulation.

This is perfectly true. But I think there is indeed another aspect. If you
do a molecular simulation, you have a Hamiltonian. This Hamiltonian needs
to be consistent between its energies and the corresponding gradients
(re:forces). So if you add a force - from IMD/SMD - to your Hamiltonian, you
also need to add a corresponding energy term. In our hands this is particular
true for doing energy minimization. As far as I can tell this is currently not
implemented in NAMD (see here ).

We have implemented an IMD protocol in our own simulation codes, and added an
energy term depending on the IMD force to the Hamiltonian. From what I saw in
the VMD source code - so far for the PickModeForceResidue::pick_molecule_move
routine (but I guess the forces in other tools are consistent with this ??) the
force in VMD is calculated by

f_imd = (10/14) * (toolPos - atomPos)

which - assuming that the energy would be zero in case the atom reached the
actual position of the tool and linearly dependent on the position wrt to

E_imd (x) = 1.4 * f_imd^2 + f_imd * atomPos - f_imd * x

where x is the position of the atom onto which the force is applied for a
given integration or minimization step and atomPos is the original position
at which the force was applied - so to say a reference position. We haven't
yet extended this to applying several forces or forces to several atoms, but

It can be argued whether you want to rather add positive penalty energy at the
point where the atom initially is (the above solution), or whether you want
to lower the energy starting from 0 penalty when moving along the force
direction towards the tool position.

Any comments and suggestions are very welcome,
```--