[Wien] the difference of the total energy is about 10 eV using wien16.1 and wien17.1

[Gavin Abo]

Just a thought, maybe it doesn't help, but have you checked the :log 
file for which programs are involved in your calculation. Then, used 
that to check if there are differences in the code for those programs 
that could affect your calculation using the updates list on WIEN2k website:

http://susi.theochem.tuwien.ac.at/reg_user/updates/

For example, if you see lapw1 in your :log file, do you see any change 
to SRC_lapw1 in the update list that could effect your calculation?

Similarly, are there any fixes from the mailing list archive that you 
have applied to the WIEN2k 17.1 code [ 
https://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/msg17267.html 
] and/or WIEN2k 16.1 code that look like they could cause differences in 
your calculation?  Or are you using the original code packages without 
any modifications?

What compilers did you use to build WIEN2k?  For the 2016/2017/2018 
Intel compilers, there seems to be a lot of reports in the mailing list 
indicating that many calculations run fine but some calculations that 
would normally run fine consistently fail but somewhat randomly such 
that corruption of file data might be caused by Intel compiler bugs or 
breaking of older file I/O code to improve the compliance with more 
recent Fortran specifications [ 
https://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/msg17253.html 
].  If there are differences in code files, the compiler might optimize 
differently producing different executable instructions. If your not 
already using -O0 [ https://software.intel.com/en-us/node/677989 ] for 
both the WIEN2k 16.1 and WIEN2k 17.1 compiler settings, you could try 
that to see if it makes a difference or not.

On 4/26/2018 8:08 AM, Kefeng wang wrote:
> Dear Dr. Tran,
>
> Thanks.
>
> I checked the input files carefully and repeated both of them twice. I 
> still obtained the same results. For all of them, the SCF convergence 
> has been reached and there is no error observed.
> :DIS at the last iteration for wien17.1 is  0.000038 and for wien16.1 
> is 0.000190.
>
> Best,
> K. Wang

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://zeus.theochem.tuwien.ac.at/pipermail/wien/attachments/20180426/7098f04f/attachment.html>