I need to use the “fix ACKS2/ reaxff” command in my simulation. But, I could not find the potential file supporting ACKS2 for my system.
As my background is not computer science, I need to know whether defining two ACKS2-specific parameters (global bond softness parameter Λ and element-specific thresholding parameters σi) is enough or I need to optimize all other related parameters so that I could use the “fix ACKS2/ reaxff” command.
Typically you need to reparametrize all related parameters for charge equilibration, if not the entire ReaxFF potential since everything is intertwined. Just adding the two ACKS2 parameters to an existing model is not likely to be accurate.
Thank you so much Stamoor. It was really helpful.
I apologize I have another question. I compared water potential files (qeq_ff.water and acks2_ff.water).
In these files, only two parameters have been changed.
and the other parameters in acks2 are the same as the qeq potential file.
Honestly, my question is if there is not potential file for other elements except water supporting this command, why it has been defining in lammps. Because based on your comment it is not easy for all users to provide a potential file for other elements to support acks2/reaxff command. I will be grateful of you could guide me.
See note here: https://doi.org/10.1021/acs.jctc.1c00975
“The ReaxFF forcefields used for this study are provided in the Supporting Information and based on the forcefield used in Achtyl et al.47 In general, the QEq water forcefield is trained for density and diffusion. For the ACKS2 forcefield, two ACKS2-specific parameters were tuned against the same training data via a relatively simple single-parameter-at-a-time search while keeping the same QEq water forcefield parameters for the rest. A more rigorous ACKS2 forcefield would require a full reparameterization against the training data that includes the atomic parameters, bonded parameters, angle parameters, etc. As such, the fitting of the ACKS2 parameters will cause deviations from the QEq water behavior, where the full QEq parameterization was optimized against the DFT training data.”
Only you can determine what is needed for your research. We don’t have ACKS2 files in the LAMMPS repo other than water, but you could reach out to Adri van Duin and see if he has others.
Thak you for your explanation. I really appreciate it.