thank you very much for your answer Steven. We have been compiling siesta since
2008 with intel compilers and never saw this problem.
The problem occurs for the serial version, so intel´s mpi is not the problem,
maybe mkl...
If somebody has been able to run calculations for hydrogen using the trunk458
version, would you please share the pseudo?
Many thanks.
Best regards,
pablo
From: Bailey, Steven
Sent: Sunday, March 16, 2014 6:28 AM
To: mailto:[email protected]
Subject: Re: [SIESTA-L] siesta-trunk458 Hydrogen Ghost states (only this
version and trun 454)
If you compile with intel this might well be the problem. I will see how one of
our PhDs students found a way round tomorrow (Monday)
Steve Bailey from iPad
On 16 Mar 2014, at 05:30, "Dr. Pablo A. Denis" <[email protected]> wrote:
Dear all,
we have compiled the trunk version 458 employing mpiifort
4.1.0.024 and MKL 10.3 update 6. When VDW-DF calculations are performed for
hydrogen containing systems, we always got the error message:
GHOST: WARNING: Ghost state for L = 1
KBgen: WARNING: Some parameter should be changed in the
KBgen: WARNING: pseudopotential generation procedure.
Stopping Program from Node: 0
Many pseudos for hydrogen were tested and we always receive the same error
message. The thing which puzzles us more is that the pseudos that we have work
well with previous versions of siesta (like trunk 320 and trunk339).
Notes: a) The problem only occurs when the VDW-DF functional is called, for
PBE everything goes well.
b) For C,N,O,F,S,Si the VDW-DF pseudos work well
any clue?
Many thanks
Best regards,
pablo
