ABINIT Discussion Forums

No... Sorry.
In the future, the KSS file will not be supported anymore anyway.

Fabien

Ok, this was not obvious to me.
But anyway, thanks for your answer.

Fabien

Dear all,

In the latest version of abinit, MPI can be activated only if both mpif90 and mpicc are provided and are operational.
mpicc used to be optional in the previous versions.
Is there a reason for this change?
Thanks for your tips.

Fabien

Thanks.
It will be fixed in the forthcoming version.

Fabien

No no, it is the real screened exchange.
It uses the calculated ab initio screening epsilon^-1.
You can take a look at my old PRB (Bruneval, Vast & Reining) to get some extra explanations.
Cheers,

Fabien

It seems to be related to the convergence vs k-points. The converge of the exact exchange vs k-points is noticeably slower than the convergence of LDA. A way to improve the convergence is to use a truncated Coulomb interaction as proposed by Spencer and Alavi in their PRB. Could you make a try using...

Thank you all for your help.
Cheers,
Fabien

It's difficult to follow two discussions in the same thread...

Alain, you were right. I erased the netcdf tarball and it worked again. Thank you very much!

Now comes the question: how do we disable the plugins nowadays? This option has simply disappeared from the './configure --help'
Bye.

Fabien

Sorry Alain, I attached the wrong file.
Here it is!

Fabien

Hello guys, Here is the config.log I obtained. However, as I told you, I can see a lot of changes in the build system from 6.7.0 to 6.7.1. All the options have changed. For instance, my favorite option doesn't work... ./configure --disable-all-plugins configure: WARNING: unrecognized options: --disa...

Hello everybody, I'm trying to build the newest development branches and I badly fail. I just downloaded my 6.7.1-private. makemake works fine and then ./configure gives configure: error: cannot build Abinit - aborting I did the same operations with my 6.7.0-private and it works. Note that I did not...

The shifts of your kpoint grids are weird.
Try to turn to, let's say, a unique Gamma centered grid.

Fabien

What system do you calculate?
Is it a weird phase of silicon? So is it a metal?
If so, the dielectric constant is very large and does not fit the format in Fortran. This can explain the "****".

Cheers,

Fabien

Input file please?

Fabien

You may be able to calculate the GW corrections for your non convential grid by switching off the symmetries. Set something like nsym=1 This comes with no warranty at all and the calculations are going to be much heavier. I would rather try to find the specific k-point you want from a conventional s...

1) David asked you for your log file...
2) Your k-points are over-over-converged.

Fabien

Hi,

I think the kshift is not correct for the GW part. Try to use either 0. 0. 0.5 or 0. 0. 0. and let us know if it's working better.
Good luck.

Fabien

Check your acell values!

Hi,

I strongly advise you to turn to "kssform 3" that does not perform a full diago of the KS hamiltonian but rather saves the iterative solution of the previous calculation.
Good luck.

Fabien

One of your systems is 8 times larger than the other one... With a tiny bit of imagination, this may lead you to some kind of answer.

Furthermore, the output does not correspond to the input you provide.

Fabien

By the way, the number of empty bands you have is extremely low. Please perform a convergence study with respect to nband.
bye.

Fabien

In this type of pseudo, you have several projectors per angular momentum. This is not allowed so far in the GW implementation of abinit.
Try to use instead traditional troullier-martins pseudos.

Fabien

Which element? which semicore? which pseudo?

Please give the complete log file.

Fabien

Your k-point set appears a bit too small. Try to increase it.

Fabien