Atomic gradients?

[ghutchis]

Is there already a specification for forces/gradients on atoms?

I ask, because this would be really useful for the forcefield interface and I wanted to know if I should write my own syntax (e.g., adding an attribute to atoms) or adopting existing practice.

[cryos]

Would these be vectors, or what exact form is the term expected to take? I would guess a three component vector at each atom, but wanted to clarify.

[ghutchis]

These would be vectors - much like vibrational displacements. I can imagine either stored as a vector of 3-component vectors or a vector of floats like atomic coordinates.