Maximum value of NGXF & Bader analysis
Posted: Thu Mar 09, 2017 10:36 am
Hello.
I have been studying hydration of caesium ion by VASP simulation (v. 5.3). While the supplied PAW PBE pseudopotentials appear to yield reasonable results for MD simulations, I have trouble with electron density analysis based on the Bader scheme. For that purpose I've been using the code developed by the Henkelman group (University of Texas). However, the Bader charges of the total density appear to be invalid, because the analysis cannot reproduce the total number of electrons correctly, not to mention that charge of certain atoms (including caesium) is evidently wrong (e.g., +5 electrons or so).
When treating the total (core+valence) density by Bader analysis, it is advisable to use sufficiently dense grid, and indeed I observed some promising trend of the results with the increasing values of NGXF, NGYF and NGZF; however, it appears that these values have an upper limit in the code. With NGXF = NGYF = NGZF = 700, VASP crashes, possibly due to hitting the limit. My results suggest that I should try with considerably larger number of grid points, e.g. NGXF = 2000 or more. I speculate that this can be achieved by accordingly modifying the VASP source code.
I would appreciate information in which part of the code (which source file) the maximum value of NGXF is defined, and if it is possible to increase that limit fairly easily. My coding skills are quite limited, and I have no idea if the possible upper limit of NGXF is related to other varibles such that the increase in the maximum for NGXF triggers several other modifications that need to be done in order the program to run properly.
Thanks in advance!
I have been studying hydration of caesium ion by VASP simulation (v. 5.3). While the supplied PAW PBE pseudopotentials appear to yield reasonable results for MD simulations, I have trouble with electron density analysis based on the Bader scheme. For that purpose I've been using the code developed by the Henkelman group (University of Texas). However, the Bader charges of the total density appear to be invalid, because the analysis cannot reproduce the total number of electrons correctly, not to mention that charge of certain atoms (including caesium) is evidently wrong (e.g., +5 electrons or so).
When treating the total (core+valence) density by Bader analysis, it is advisable to use sufficiently dense grid, and indeed I observed some promising trend of the results with the increasing values of NGXF, NGYF and NGZF; however, it appears that these values have an upper limit in the code. With NGXF = NGYF = NGZF = 700, VASP crashes, possibly due to hitting the limit. My results suggest that I should try with considerably larger number of grid points, e.g. NGXF = 2000 or more. I speculate that this can be achieved by accordingly modifying the VASP source code.
I would appreciate information in which part of the code (which source file) the maximum value of NGXF is defined, and if it is possible to increase that limit fairly easily. My coding skills are quite limited, and I have no idea if the possible upper limit of NGXF is related to other varibles such that the increase in the maximum for NGXF triggers several other modifications that need to be done in order the program to run properly.
Thanks in advance!