Convergence problem

Dear Developers,

I have encountered an issue while performing basic optimizations of molecules with constrained degrees of freedom, such as bonds, angles, or dihedrals. As a newcomer to Psi4, I have previously been using Gaussian09/16 for these optimizations.

I have noticed a significant disparity in the convergence speed between Psi4 and G16. It appears that Psi4 requires a considerably higher number of cycles to converge compared to G16, sometimes even an order of magnitude more.

To illustrate this issue, I attach a test on a benzene molecule as an example. With Psi4, optimization took 66 cycles, while G16 accomplished the same task in 5 cycles. I can provide additional examples of molecules and/or degrees of freedom if necessary.

I would greatly appreciate any assistance or insights you could provide to help me address this issue. Thank you in advance for your time and consideration.


Convergence problems require @AlexanderH.