Posts

Showing posts from June, 2015

Gromacs games

Additional energy terms in custom Hamiltonians, exchanges The standard energy function in looks something like this:

E = E_bonds + E_angles + E_dihedrals + E_LJ + E_Culombic

This energy term, at some temperature T then becomes:

U = exp(-E/kb*T)

Let's say one has a n such Hamiltonians, U1...Un. These can then exchange, if the energies overlap. The details of the exchange scheme are skipped here, for gromacs implementation if the exchange see src/kernel/repl_ex.c

In temperature replica exchange one uses a ladder of T values, hoping that U1...Un overlap. In solute tampering replica exchange, one modifies the energy function E, keeping the T constant, hoping again to see exchange.

In deciding if a pair of replicas on a ladder will exchange, one compares their potential energy (Epot).

When including extra energy terms into the hamiltonian (custom biases, for example) one has to make sure they will be included in the Epot.

For instance, let's see if the gromacs pull code adds its e…