Point Multipole Embedding for Gradients

[aaa16797]

Hi all,

I am using TURBOMOLE 7.7. The manual states that it does not permit for calculation of gradients with point multipole embedding, yet when I provide one and compute the gradient through grad, it gives me a different gradient than if I were to not supply them (I am using the same converged density input for both cases and only running grad, so a rerun of scf is not the issue here)? I am noticing some of my optimizations behaving odd that is why I started looking into this issue, because its unclear if the program is computing gradients properly here. Thank you for any help you all can provide on this.