I am using fix npt to apply pressure with triclinic box. I am wondering the direction of x, y,z for triclinic box? Is the force applied along x y and z still along the orthogonal direction or the force is applied along the three boundaries of the triclinic box?

I am using fix npt to apply pressure with triclinic box. I am
wondering the direction of x, y,z for triclinic box? Is the force applied
along x y and z still along the orthogonal direction or the force is applied
along the three boundaries of the triclinic box?

fix npt doesn't apply a force but changes the dimensions (and tilt) of
the box and rescales the atom positions with it (or not, depending on
the settings).

Thanks very much! So for xy, yz and xz, the stress is still applied to
orthogonal plane
? Thanks very much for your help?

let me repeat for the n+1st time. there is no "stress" (or force)
applied to any plane anywhere. perhaps you should explain why you
insist on asking this question rather than repeating it over and over
again.

atom positions are converted to fractional coordinates. then the box
is deformed (along the axes and tilted, depending on the fix npt
settings, all at the same time) and after the box is updated, the atom
positions are converted back to cartesian coordinates. full stop. this
is what the documentation explains and that is what happens (unless
you tell fix npt to not modify the atom positions at all). where is
the problem in understanding it?

Thanks very much! Then how about the direction of the stress output by the dump command? Is the stress pxy on the xy plane which is in the orthogonal coordinate system?

I am combining the simulation results with continuum mechanics. In continuum mechanics, the direction of the stress matters. So I care about the direction of the stress output from lammps. Sorry for bothering. Thanks very much!

I am combining the simulation results with continuum mechanics. In continuum
mechanics, the direction of the stress matters. So I care about the
direction of the stress output from lammps. Sorry for bothering. Thanks very
much!

in LAMMPS you have no choice of direction, since the box geometry has
to follow certain rules.
the cross terms in the tensor represent shearing and thus result in
changing of the tilt factors.