I am having issues using external CPI files in nwchem 6.1.1 on amd64.
My target at the moment is simply asserting I am able to run my own PAW Cpi files.
I have the PAW in /usr/local/NWChem/data/libraryps/CPI, in extensionless files in the fashion of:
>ls
C O Si U
permissions should be fine.
Using native CPI files for Si and C and it seems I am running successfully a modified example from
"Optimizing a unit cell and geometry for Silicon-Carbide" [1]
To avoid cluttering the post, here's just the the diff with my custom one.
> charge 0
> memory stack 1000 heap 1000 global 2000 mb
27a30,33
> pseudopotentials
> C library CPI
> Si library CPI
> end
This simulation runs just fine, but how can I be sure that nwchem is using PAW from my CPI and not from the stock db?
Now I move to U, missing in the PAW database.
CPI for U and O in this case were in .fhi format [08-O.GGA.fhi and 92-U.GGA.fhi[2] and converted using psp_converter [3]
title "paw steepest descent test"
start paw_test
charge 0
memory stack 1000 heap 1000 global 2000 mb
geometry units au nocenter noautoz noautosym
U -0.43798 1.73025 0.00000
O 0.84978 3.23711 -0.00000
O -1.94488 3.01797 -0.00000
end
nwpw
time_step 15.8
ewald_rcut 1.50
tolerances 1.0d-8 1.0d-8
pseudopotentials
U library CPI
O library CPI
end
end
set nwpw:lcao_iterations 1
set nwpw:minimizer 2
task pspw energy
task paw energy
in this example, nwchem bails out after the successful pspw energy calculation it bails out with the following:
library name resolved from: .nwchemrc
NWCHEM_NWPW_LIBRARY set to: </usr/local/NWChem/data/libraryps/>
Generating 1d pseudopotential for U
Error: <atom> section not found
Is it an issue with the converted CPI files?
Is the successful SiC example really using my external CPI PAW file?How to tell?
Any suggestion on converting successfully the .fhi to .cpi so to avoid generating the from scratch using Atompaw?
Thank you,
|