Hello,

I am attempting to optimize the structure of a 3d crystal whose lattice parameters are (say) a=b=c=10.0 ang and angles of alpha=beta=90 and gamma=120.0 deg using fix minimize and fix box/relax commands. In such a case the tilt factor xy = bcos(gamma) is exactly (xhi-xlo)/2.

During the minimization the tilt factor sometimes exceeds the -(xhi-xlo)/2 < xy < (xhi -xlo)/2 limit and lammps give the expected “ERROR: Triclinic box skew is too large (domain.cpp:144)” error.

Currently, I am using multiple unitcells in order circumvent this error - for the xy tilt use multiple unitcells in the x direction to get a larger xhi value. I’d really not want to continue doing this.

More importantly, I am interested in geometry optimizations of systems that are more skewed that gamma = 120 deg. For example a crystal with a=b=c and alpha=beta=gamma=60 deg.

Is there a more elegant solution to this? Or am I missing something very trivial? Is there a way to bypass the tilt limitations?

Thank You,

Regards,

Ambar