[Wien] :WARN : WARNING: RKmax reduced due to NMATMAX

Tue Nov 7 17:19:31 CET 2017

Thank you Prof. Alay for a nice hint to solve the problem.
Let me arrange computational facility if I can then I will get back here.
Thank you very much all!

    On Tuesday, 7 November 2017 9:25 PM, M abbas <alayabbas at gmail.com> wrote:

  Gavin and Professor Marks are right insaying that you will need more computing resources (MPI a must) if youfinally decide to use a small RMT for O (in my experience only an RMT =1.1 or smaller works well for O2 dimer's relaxation). That being said, one can usesome workarounds for this problem. The following procedures come handy foroxides, especially when you are interested in computing cohesive energies and thermodynamicstability but don't want to use small RMT of O for subsequent calculations involving oxides and their supercells.

(1) Compute your own cohesive energy(will need MPI and small cluster) (a) first relax an O2dimer and compute energy of a single O atom using RMT = 1.1 andRKMAX = 5.5. (b) work out the cohesive energyE[cohesive_O_rmt1.1] = E[O2_rmt1.1] – 2 * E [O_atom_rmt1.1] (c) compute energy of an O atomwith RMT and RKMAX of your choice (e.g. RMT = 1.6 and RKMAX = 8.0);E [O_atom_rmt1.6] (d) To get the energy of O2 dimer witha (imaginary) RMT of 1.6 add the above computed cohesive energy into E [O_atom_rmt1.6]: i.e. E[O2_rmt1.6] = 2* E [O_atom_rmt1.6] +E[cohesive_O_rmt1.1]
(2) Use an already reported value ofcohesive energy (works fine for small computers)
 (a) Since the cohesive energies of O2for frequently used XC functionals within WIEN2k are alreadyavailable in literature (e.g.https://doi.org/10.1103/PhysRevB.75.115131), the energy of O2 dimer with a (imaginary) RMT of 1.6 can be easily had from  E[O2_rmt1.6] = 2* E [O_atom_rmt1.6] +E[cohesive_O_computed_using_wien]
The above procedures are logical butdo not allow one to play much with the RMT values of O2 dimerwithin the muffin-tin model here. I did however perform some testswith a P2 dimer some time ago and found both the methods to be correct.
Best,Alay
On Tue, Nov 7, 2017 at 7:00 PM, Laurence Marks <L-marks at northwestern.edu> wrote:

You do not need want RKMAX 6.5 for O2, because you will need RMTs of about 1.2. 6.0 or 5.75 will be fine. BUT you have to use the same O RMT & RKMAX for the oxide.
You need more computing resources.

On Nov 7, 2017 13:46, "chin Sabsu" <chinsabsu at yahoo.in> wrote:

Thank you Sir,
runsp_lapw -it gave me an error:  foreach. So I am running with runsp_lapw ....  script. 

If I run it with Rkmax=6 then it does not show me any warning. At the FAQ of Wienk page standard RMT for O is 6.5 but it also gave me RKmax reduced warning.

If someone advises me that I can take RKMAX=6 for O2 system then only I can proceed.
I do not have any good facility.

On Tuesday, 7 November 2017 6:59 PM, Gavin Abo <gsabo at crimson.ua.edu> wrote:

To make that calculation feasible, it looks like you need a better computing system like a small cluster and mpi.If ~8 GB is your total RAM, keep in mind that the Linux operating might use around 1 GB.  Using top [https://superuser.com/question s/282867/why-does-the-memory- usage-in-top-not-add-up ] or a system monitor [https://askubuntu.com/question s/29757/what-can-replace- system-monitoring-in-the-top- gnome-panel-in-unity ], you can check for any unnecessary processes and kill them to try to gain back more "free" memory.  Though, while this might help a little with memory, it likely won't help with the speed issue.For speed, you likely need mpi or iterative diagonalization.  Are using iterative diagonalization (runsp -it) as was mentioned before:
http://zeus.theochem.tuwien.ac .at/pipermail/wien/2017-Novemb er/027303.htmlIf none of that helps, it looks like your stuck with doing less accurate calculations using small RKmax values.
On 11/6/2017 9:24 AM, chin Sabsu wrote:

Thank you Sir,

As per memory test, I got the figure: 46743  it means ~4.5 GB RAM is enough to run the case. But my system is having ~8 GB RAM:
7.6 GiB
Intel® Core™ i5-3570K CPU @ 3.40GHz × 4
Intel® Ivybridge Desktop

DearLyudmila Dobysheva
I increaseNMATMAX  upto45000 and NUME at 10000. At this value, it gave me lapw1 error while at NMATMAX (NUME)=35000(8000) it taking too much time and even after three hrs lapw1 is running.

I want to keep consistent RKMAX with the literature. Otherwise, my job is to calculate the thermodynamic stability and for the same, I need the cohessive energy of O also.

Can I use RKmax 6.0 for O2 and for others I am tacking 7. If not then what should I do to run this O2 system?

Chin

______________________________ _________________
Wien mailing list
Wien at zeus.theochem.tuwien.ac.a t
http://zeus.theochem.tuwien.ac .at/mailman/listinfo/wien
SEARCH the MAILING-LIST at:  http://www.mail-archive.com/wi en at zeus.theochem.tuwien.ac.at/ index.html

______________________________ _________________
Wien mailing list
Wien at zeus.theochem.tuwien.ac. at
http://zeus.theochem.tuwien. ac.at/mailman/listinfo/wien
SEARCH the MAILING-LIST at:  http://www.mail-archive.com/ wien at zeus.theochem.tuwien.ac. at/index.html

_______________________________________________
Wien mailing list
Wien at zeus.theochem.tuwien.ac.at
http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
SEARCH the MAILING-LIST at:  http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://zeus.theochem.tuwien.ac.at/pipermail/wien/attachments/20171107/91e2d1bb/attachment.html>