There's a big difference between calculating the stress and prescribing it. LAMMPS certainly has global and local (atomic) measure of stress and pressure that can be calculated during a simulation. However, the only methods to make the system converge on a target value of stress are use of the npX fixes (using periodic bc's) and the method I suggest.
If you're concerned about initial stress present in the system, I'd suggest first performing an energy minimization using the minimize and related commands to relieve these stresses. Then, apply the boundary conditions as you wish.
As Jon Z suggests, to relax the stresses existing in a non-periodic system one can use the minimize command.
I remember that we had a discussion on NPT using Berendsen pressure control on a system with free boundaries. Steve did discuss the implementation details but I do not know if it has found its way into LAMMPS version that has been subsequently released. The latest LAMMPS documentation suggests that fix press/berendsen still requires PBCs.
Nano-sized solids do exist in vacuum. Therefore I believe it is OK to allow fix press/berendsen for non PBCs in solids for interatomic potentials that are not purely repulsive and have a minima (and therefore a attractive part). One must however keep in mind that most of these interatomic potentials are developed for atoms in the bulk and may not be appropriate for simulating surface atoms. However that has not stopped many of us from simulating surfaces using these interatomic potentials.
2009/5/3 Zimmerman, Jonathan A <[email protected]…3…>