|
Changes from version 3.1 to 3.2 |
|
Wednesday, 12 October 2005 |
New features
- Exact restarts are now also possible with
temperature and pressure coupling. To achieve this coupling data is
written to the energy file, which can be passed to tpbconv and grompp.
A parameter init_step has been added to the mdp file so also the step
numbers can be continued.
- Conjugate gradient minimization uses a new, more efficient algorithm: Polak-Ribiere.
- A new random number generator has been implemented which uses a state of 624 numbers. It should never cycle in practice.
- The
random Gaussian number generator now uses a table lookup which is at
least an order of magnitude faster than the old one. This speeds up BD
and SD simulations of simple systems significantly.
- Complete
rewrite of the pull code. The forces are now corrected for
mass-weighting when dynamic pull-groups are used. There are some new
features with constraint pulling.
- With shift and switch potentials rlist is now allowed to be larger than rcoulomb and rvdw.
- Dispersion
correction can now also be applied to switched potentials (Note that in
3.2.0 and 3.2.1 the virial correction is incorrect, but the pressure
correction is correct).
Bug fixes
- A bug
has been fixed with PME and triclinic boxes that have angles that are
far from 90 degrees, such as a rhombic dodecahedron and a trucated
octahedron. When a molecule larger than half the box-size would be at
the edge of the box the periodicity was not taken into account
correctly.
- A bug has been fixed which could cause some LJ
interactions at distances close to rvdw to be neglected when rvdw >
rcoulomb (in practice this would only occur with GROMOS forcefields
combined with PME).
|