[CP2K:7856] Optimize RI Basis sets

[hut... at chem.uzh.ch]

Hi

MOLOPT basis sets should not be used for RPA (or MP2) 
calculations. They are optimized for GGA functionals.
For correlation methods you need more flexibility and higher
L quantum numbers.
You should also carefully test the setup for the RPA in order
to be able to handle systems without a band gap.

I can't say anything about the RI basis optimization.

You are trying a very challenging system!

regards

Juerg
--------------------------------------------------------------
Juerg Hutter                         Phone : ++41 44 635 4491
Institut für Chemie C                FAX   : ++41 44 635 6838
Universität Zürich                   E-mail: hut... at chem.uzh.ch
Winterthurerstrasse 190
CH-8057 Zürich, Switzerland
---------------------------------------------------------------

-----cp... at googlegroups.com wrote: -----To: cp2k <cp... at googlegroups.com>
From: Michael Deible 
Sent by: cp... at googlegroups.com
Date: 06/22/2016 09:25PM
Subject: [CP2K:7856] Optimize RI Basis sets

Hi,
I would like to do an RPA calculation to determine the binding energy of an organic molecule to an iron surface.  To do this, I need RI basis sets for iron, carbon and oxygen.  I was planning to use the MOLOPT basis sets, and use the OPTIMIZE_RI_BASIS method to generate the auxiliary basis sets. To get started, I ran the regtest: cp2k-3.0/tests/QS/regtest-ri-mp2/opt_basis_O_auto_gen.inp.  This completes as expected.  However, If I only change the basis set in this file to SZV-MOLOPT-GTH-q6, I get a segmentation fault.  
I compiled CP2K-3.0 using the install_cp2k_toolchain.sh script on a Cray XC40 (Excalibur) (arch file attached).  To make sure that it wasn't my compilation that was the problem, I downloaded several of the precompiled binaries.  With the cp2k-3.0.ssmp binary,  I now get the following error: Note: The following floating-point exceptions are signaling: IEEE_UNDERFLOW_FLAG.  For the DZVP-MOLOPT basis, I get the error:  Error in `./cp2k-3.0-Linux-x86_64.ssmp': malloc(): memory corruption: 
I've noticed that most of the literature concerning the RI_RPA method uses the correlation consistent basis sets.  Is there some reason not to use the MOLOPT basis sets with RI_RPA, and is that why I am having issues generating an auxiliary basis set for them?  Or (more likely) is my compilation bad, and the precompiled versions aren't optimized enough to fix this?
Thanks,Mike



-- 

You received this message because you are subscribed to the Google Groups "cp2k" group.

To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+uns... at googlegroups.com.

To post to this group, send email to cp... at googlegroups.com.

Visit this group at https://groups.google.com/group/cp2k.

For more options, visit https://groups.google.com/d/optout.



[attachment "local.psmp.rtf" removed by Jürg Hutter/at/UZH]