CCL: Periodic boundary conditions

 Sent to CCL by: Lubos Vrbka []
 David Santos Carballal spectrum_dav*_* wrote:
 Sent to CCL by: "David Santos Carballal"
 Dear CCl members:
 I am doing some molecular dynamics in TINKER.
 I made a box of solvent (water) with the program xyzedit.
 In the keyword file evoke keywords: rattle (the responsible of shake method in
 order to restrain O-H distances and angles in water); a-axis (to define lattice
 lenght); octahedron (to simulate a sistem closer to a sphere) and cutoff (to
 evaluate the integrals in a region smaller than the half of dimension of the box
 of solvent).
 My system have 100 water molecules.
 First of all I made a minimization with a conjugate gradient of 0.01. After it I
 run dynamic with the Beeman integration method and constant temperature (298K)
 and pressure (1 atm.). When the dynamic begins temperature stabilizes so fast
 but pressure is very high, it wuold be around 4000 atm. The pressure fall down
 at 1 atm. after 12000 steps, but the lattice length grow up so much, the initial
 lattice length was 14 angstroms and the final is 200 angstroms, with one
 consecuence: density fall down  until 0.0010 grams/cc, that is desagree with
 water density. I use MM3 force field.
 My question is I am using correctly periodic boundary conditions?
 Why this results?
if your initial system is ok and looks like it should look like then first thing i could think of: after minimization, do constant volume run first before switching to constant pressure to get somehow relaxed system. this might help a bit...