Since ReaxFF combines the advantages of DFT and Classical MD, would there be any need for periodic DFT calculations if we can have all the forcefields of all the elements of all the possible systems?
$\begingroup$
$\endgroup$
2
-
$\begingroup$ Some references for ReaxFF would be useful $\endgroup$– Ian BushApr 16, 2020 at 16:01
-
2$\begingroup$ I would argue there could also be a use for DFT no matter if we have good force fields or otherwise machine learned values. For one, many of these ad hoc approaches require extensive re-parameterization based on the very same values they are replacing. For two, DFT is a formally exact theory, while force-fields are not. Therefore there is always value in re-affirming the model of your model by checking the exact theory. $\endgroup$– Cody AldazApr 16, 2020 at 16:03
Add a comment
|
1 Answer
$\begingroup$
$\endgroup$
Quoting from the article The ReaxFF reactive force-field: development, applications and future directions
"Such empirical methods, including reactive force-field (ReaxFF),1 trade accuracy for lower computational expense, making it possible to reach simulation scales that are orders of magnitude beyond what is tractable for QM."
So yes, if you want accuracy, especially for something which the approximations made in generating a ReaxFF force field are not good, you are going to have to get out the Quantum Mechanics.
