Hello,
I have some questions regarding the convergence criteria of ionic steps:
1. Force and energy are both options of convergence criteria, but are there any fundamental differences between them? When should I use force and when should I use energy convergence?
2. How should I determine if my structure is relaxed well enough?
3. Are there some general settings that helps increases the efficiency of reaching convergence in ionic steps?
General considerations on convergence of ionic steps?
Moderators: Global Moderator, Moderator
-
Leon1999526
- Newbie
- Posts: 10
- Joined: Tue Apr 18, 2023 2:19 am
-
marie-therese.huebsch
- Full Member
- Posts: 212
- Joined: Tue Jan 19, 2021 12:01 am
Re: General considerations on convergence of ionic steps?
Hi,
The first two questions are rather scientific.
Ad 1) It depends a lot on the purpose of the relaxation and the other quantities of interest. I would say relaxing until the forces are zero is the simplest and most naive approach, but good enough in many considerations. Relaxing the total energy goes more toward relaxing the equation of state. Using an equation of state (there are some possible choices again depending on the pressure regime etc.) has proper reasoning from the standpoint of thermodynamics. Finally, you may need to analyze the phonon spectrum to ensure you are not considering a metastable structure.
Ad 2) It is necessary to perform a convergence study on the quantity of interest.
Ad 3) Check the description of IBRION and the related tags, e.g. POTIM. After you have chosen the algorithm, use some small system relatively close to its structural minimum and play with it. This way, you will understand how to set up your calculation when you start the computationally more demanding calculation of the material you are interested in.
I hope these answers are not too vague. It will help if you share a specific calculation you want to perform or what quantities you aim to compute.
Marie-Therese
The first two questions are rather scientific.
Ad 1) It depends a lot on the purpose of the relaxation and the other quantities of interest. I would say relaxing until the forces are zero is the simplest and most naive approach, but good enough in many considerations. Relaxing the total energy goes more toward relaxing the equation of state. Using an equation of state (there are some possible choices again depending on the pressure regime etc.) has proper reasoning from the standpoint of thermodynamics. Finally, you may need to analyze the phonon spectrum to ensure you are not considering a metastable structure.
Ad 2) It is necessary to perform a convergence study on the quantity of interest.
Ad 3) Check the description of IBRION and the related tags, e.g. POTIM. After you have chosen the algorithm, use some small system relatively close to its structural minimum and play with it. This way, you will understand how to set up your calculation when you start the computationally more demanding calculation of the material you are interested in.
I hope these answers are not too vague. It will help if you share a specific calculation you want to perform or what quantities you aim to compute.
Marie-Therese