Precision of EIGENVAL output

Message

darshw · #1 Post by **darshw** » Tue Sep 10, 2013 2:09 am

Hi all,

I am currently running VASP simulations on fairly large supercell structures. I've noticed as I increase the size of my supercell the precision of the output in the EIGENVAL file decreases from the default output of 6 decimal places to 4 decimal points.

I am curious if there is a way to increase this degree of precision since I need to compare energy differences on the the order of a few hundreths of a meV at times. Would this require a re-compilation?

I have seen similar posts in the forum (eg. http://cms.mpi.univie.ac.at/vasp-forum/ ... php?4.5893) alas with no response.

Any help would be appreciated.

Thanks,

Darshana
UC Riverside
Dept. of Electrical Engineering

#2 Post by **admin** » Tue Sep 10, 2013 9:40 am

Yes, this is easily treated by changing the output format in the source code and recompilation.

darshw · #3 Post by **darshw** » Wed Sep 11, 2013 8:31 am

Thanks, this is good to know. Doing a simple grep of EIGENVAL in the source files shows there are multiple files in which EIGENVAL is referenced. There is also a section in the main.F file which seems to make the necessary function calls to output the data contained in EIGENVAL.

I'd like to know which parts of the source code needs to be altered and recompiled to allow for this higher precision to be included in the EIGENVAL file output. Our lab currently has a valid license for VASP 4.6 and VASP 5.x

Thanks again for your help.

Best,
Darshana
<span class='smallblacktext'>[ Edited Wed Sep 11 2013, 08:32AM ]</span>