How to Maintain Structural Consistency When Switching from Diagonalization to OT Algorithm in CP2K?

[yis...@163.com]

Dear CP2K developers,

Hello everyone. I have a rather basic question to ask. I am currently using CP2K to study a metal oxide material. This system is an insulator and exhibits magnetism. When using the diagonalization method with k-point sampling, convergence is extremely difficult to achieve. Therefore, I intend to switch to the OT algorithm. However, I understand that the OT algorithm is generally not suitable for small unit cells and requires supercell expansion for calculations.

My question is: when optimizing the expanded supercell using the OT algorithm, the symmetry of the cell may be reduced. How should I set up the calculation to ensure that the structure optimized with the OT algorithm (using a supercell) remains consistent with the one obtained from diagonalization with k-point sampling (using the primitive cell)?

In the CP2K manual, I noticed the keywords KEEP_SPACE_GROUP and KEEP_SYMMETRY. Would setting both of these to .TRUE. guarantee that the symmetry of the expanded supercell is preserved, thereby ensuring structural consistency between the two approaches?

Thank you for your time and assistance!