For a quick overview, see https://en.wikipedia.org/wiki/Force_field_(chemistry). For a wikipedia article, it is a very detailed discussion with many references to the appropriate publications,

For more details, follow the references quoted in the article.

In adition to what Annemarie Honegger suggested:

Development of each force field is accompanied with the paper describing improvements of a given ff. Also, the choice of a ff is closely related to the system of interest – some ffs simply work better for some systems. Therefore, doing literature research and checking what other people reported for similar systems is a good starting point.

Dear Bijaya Pathak

The first main difference is the level of description, which is basically the resolution. Some of them (e.g. CHARMM) are all-atom force fields, that is they describe every single atom in a molecule.

Others are united-atom force fields, that is they describe most of the atoms, but neglect some of them in exchange for a speed up in computation. GROMOS is an example of a united-atom force field, where CHn aliphatic groups are represented by only an atom, usually a 'special' carbon atom which is parametrised to take into account also the missing hydrogen atoms.

Lastly, there are coarse-grained force fields, like MARTINI, which apply the same principle of united-atoms force fields, but to all the atoms of the system. A common approach is to map four heavy atoms into a single one, named bead. This leads to a drastically reduced number of particles in the simulation box, which concurrently allows for larger systems and faster simulations. The price is clearly a decrease in resolution, the systems is indeed more granulous.

The principle behind all of them is more or less the same. You want to integrate the Newton eqn of motion, which in molecular dynamics terms means that you need the potential for your system so that you can take its gradient and compute the force. The potentials have various forms, but usually are split into the bonded (~ intra-molecular) and non-bonded (~ inter-molecular) terms.

Different force fields can differ both in the form of these functions and in the set of parameters that you plug in to actually compute numerically the forces. These in turn depend also on how the force field was parametrised, such as the cut-off length of the interactions, the integration step, the treatment of h-bonds etc.

The choice of force fields, strictly speaking, depends mostly on what you want to do and the computational power available. If you want full atomistic detail and huge systems simulated for long times, be prepared for the corresponding huge computational power demand.

Many books and papers have been written on these topics. A good general start are always

[Frenkel D. et al.] Understanding molecular simulation

[Leach] Molecular Modelling

[Allen M.P. et al.] Computer Simulation of Liquids

Then if you want to focus on specific force fields, you will have to read their associated papers to understand how they were modeled and what are the sims parameters that should be used.

Hope this helps,

Nicola

How about the nature of proteins. What of proteins with metals inside of them (like containing co-factors)? How will those metals be parametized and what force field is suitable for such?