...inadequate

Also the computation of van der Waals forces required in MD simulations scales with 4#4; However, as these forces decay much more rapidly with increasing interaction distance than the Coulomb forces, they can safely be truncated at about 10Å and, therefore, are computationally inexpensive .

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...charges

In a heteronuclear molecule each atom carries a partial charge; however, usual parameterizations of protein force fields are chosen such that small partial charges are neglected while the sum of partial charges is locally kept at integer values.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...monopoles

In contrast to the original SAMM version, we now also combine monopoles of opposite sign into dipolar objects at higher levels, if the total charge of these objects vanishes; this procedure allows to represent the electrostatics of neighboring ion pairs at large distances as dipoles.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...size

In MD simulations usually the integration time step is set to 1 fs, which is short enough to smoothly describe the fastest degrees of freedom in protein simulations.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...derive

Note, that this strategy also saves computer memory, since the local Taylor expansion of a selected object is used to calculate the forces on all particles contained in that object; storage of all these forces would inevitably consume much more memory.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...suppressed

To assure this observation, we have performed a second cutoff simulation and have observed similar effects.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...96 MB RAM

Due to limitations of memory we were unable to extend our tests to larger systems.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.