added some more detail to the bond special doc page

Description

"""""""""""

The *special* bond style can be used to impose weighted Lennard Jones and/or

Coulombic interactions on any two particles in the system. It can be used for

cases that cannot be handled in :doc:`special_bonds <special_bonds>`, such as

1-5 interactions. It is a potential of the form:

The *special* bond style can be used to create conceptual bonds which

effectively impose weightings on the pairwise Lennard Jones and/or

Coulombic interactions between selected pairs of particles in the

system.  The form of the pairwise interaction will be whatever is

computed by the :doc:`pair_style <pair_style>` command defined for the

system; this command defines the weightings for its two terms.

This command can thus be useful to apply weightings that cannot be

handled by the :doc:`special_bonds <special_bonds>` command, such as

on 1-5 or 1-6 interactions.  Or it can be used to add pairwise forces

between one or more pairs of atoms that otherwise would not be include

in the :doc:`pair_style <pair_style>` computation.

The potential for this bond style has the form

.. math::

----------

This style has strict requirements on the :doc:`special_bonds <special_bonds>`

setting. 1-2 interactions must have weights of zero. 1-3 interactions must

either have weights of one or the *angle* setting must be turned on. 1-4

interactions must have weights of one or the *dihedral* setting must be turned

on. These requirements ensure that the new bonds created by this style do not

create spurious 1-2, 1-3 or 1-4 interactions.

This style should be used in conjunction with a regular bond style via

:doc:`bond_style hybrid <bond_hybrid>`. Since it can be used to create

bonded interactions between particles that are further away than usual

(e.g. 1-5 or 1-6 interactions), this style may require an increase in the

communication cutoff via the :doc:`neigh_modify <neigh_modify>` command.

Normally this bond style should be used in conjunction with one (or

more) other bond styles which compute forces between atoms directly

bonded to each other in a molecule.  This means the :doc:`bond_style

hybrid <bond_hybrid>` command should be used with bond_style special

as one of its sub-styles.

Note that the same as for any other bond style, pairs of bonded atoms

must be enumerated in the data file read by the :doc:`read_data

<read_data>` command.  Thus if this command is used to weight all 1-5

interactions in the system, all the 1-5 pairs of atoms must be listed

in the "Bonds" section of the data file.

This bond style imposes strict requirements on settings made with the

:doc:`special_bonds <special_bonds>` command.  These requirements

ensure that the new bonds created by this style do not create spurious

1-2, 1-3, or 1-4 interactions within the molecular topology.

Specifically 1-2 interactions must have weights of zero, 1-3

interactions must either have weights of unity or :doc:`special_bonds

angle yes <special_bonds>` must be used, and 1-4 interactions must

have weights of unity or the *dihedral* setting must be turned on or

:doc:`special_bonds dihedral yes <special_bonds>` must be used.

If this command is used to create bonded interactions between

particles that are further apart than usual (e.g. 1-5 or 1-6

interactions), this style may require an increase in the communication

cutoff via the :doc:`comm_modify cutoff <comm_modify>` command.  If

LAMMPS cannot find a partner atom in a bond, an error will be issued.

Restrictions

""""""""""""

USER-MISC package.  See the :doc:`Build package <Build_package>` doc

page for more info.

This bond style requires use of a :doc:`pair_style <pair_style>` which

computes a pairwise interaction.  Many-body potentials do not.

Q: Does this command work with long-range Coulombics?  E.g. if used to

weight 1-5 interactions between charged particles and also used with

PPPM, does it give the right answer?  The special bond weight settings

are treated explicity in pair styles like pair lj/cut/coul/long.

Either way, the answer to this Q should be explained on this page.

And if the answer is no, then I think an error check should be made in

the code.

Related commands

""""""""""""""""