Re: [Wien] Question about large RKmax

2017-09-14 Thread 안교훈 
Thank you for your helpful comments.
I will try to check them. (i, ii, iii, and PPS)
If there is something special, I will briefly report here.

Thank you again :)
- Kyohn
 ---원본 메세지---
보낸사람: "Peter Blaha"
받는사람: wien@zeus.theochem.tuwien.ac.at
보낸시간: 2017-09-15 00:04:59 GMT +0900 (Asia/Seoul)
제목: Re: [Wien] Question about large RKmax

Hi,

Yes, your graphs make sense to me. On the scale of a tenth of a mRy
convergence can become very difficult, in particular for heavy elements.

I'd believe the RKM=12 results, but beware my final note.
I'm not sure where the oszillation in Etot comes from.

Important for this kind of precision: RKMAX is not the only parameter.
You mentioned BIG spheres: Did you make sure that
i) lvns has been increased and results are converged
ii) there's no problem with linearization energies (maybe add a HDLO,..)
iii) eventually you even need to increase the LM list in case.in2, in
particular when :FIT (in the scf file) is not good.

PPS: Since you have 5d elements, there are 4f semi-core states. For
these spheres I'd put them into the core, as they are so confined an
never leak out. You have to use charge-cutoff instead of E-cutoff in lstart.

Regards
Peter Blaha

On 09/14/2017 03:34 PM, Kyohn Ahn wrote:
> Dear WIEN2k users
>
> I'm sorry.
> In the last mail, the link was broken,,
>
> I have a problem about large RKmax.
> Please see the attached file:
> There are some parts of bandstructure of my system
> with various RKmax values.
>
> In my case only a large RKmax (11~12) shows good results.
> Is there anyone who had similar experiences?
>
> # Details:
> # The smallest Rmt = 2.5 (atomic distances are pretty long) /
> # Heavy elements from Os to Bi line /
>
> My biggest concern, really, is that
> whether the results are in the "numerical linear dependency" problem
> or not. Is there any other method to check the issue?
>
> Thank you for reading this email.
> Any response in this regard will be very helpful for me.
>
> With best regards
>
> - Kyohn
>
>
> ___
> Wien mailing list
> Wien@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
>

--

P.Blaha
--
Peter BLAHA, Inst.f. Materials Chemistry, TU Vienna, A-1060 Vienna
Phone: +43-1-58801-165300 FAX: +43-1-58801-165982
Email: bl...@theochem.tuwien.ac.at WIEN2k: http://www.wien2k.at
WWW: http://www.imc.tuwien.ac.at/TC_Blaha
--
___
Wien mailing list
Wien@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

___
Wien mailing list
Wien@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

Re: [Wien] Question about large RKmax

2017-09-14 Thread Peter Blaha 

Hi,

Yes, your graphs make sense to me. On the scale of a tenth of a mRy 
convergence can become very difficult, in particular for heavy elements.


I'd believe the RKM=12 results, but beware my final note.
I'm not sure where the oszillation in Etot comes from.

Important for this kind of precision: RKMAX is not the only parameter.
You mentioned BIG spheres: Did you make sure that
i) lvns has been increased and results are converged
ii) there's no problem with linearization energies (maybe add a HDLO,..)
iii) eventually you even need to increase the LM list in case.in2, in 
particular when :FIT (in the scf file) is not good.


PPS: Since you have 5d elements, there are 4f semi-core states. For 
these spheres I'd put them into the core, as they are so confined an 
never leak out. You have to use charge-cutoff instead of E-cutoff in lstart.


Regards
Peter Blaha

On 09/14/2017 03:34 PM, Kyohn Ahn wrote:

Dear WIEN2k users

I'm sorry.
In the last mail, the link was broken,,

I have a problem about large RKmax.
Please see the attached file:
There are some parts of bandstructure of my system
with various RKmax values.

In my case only a large RKmax (11~12) shows good results.
Is there anyone who had similar experiences?

# Details:
# The smallest Rmt = 2.5 (atomic distances are pretty long) /
# Heavy elements from Os to Bi line /

My biggest concern, really, is that
whether the results are in the "numerical linear dependency" problem
or not. Is there any other method to check the issue?

Thank you for reading this email.
Any response in this regard will be very helpful for me.

With best regards

- Kyohn


___
Wien mailing list
Wien@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



--

  P.Blaha
--
Peter BLAHA, Inst.f. Materials Chemistry, TU Vienna, A-1060 Vienna
Phone: +43-1-58801-165300 FAX: +43-1-58801-165982
Email: bl...@theochem.tuwien.ac.atWIEN2k: http://www.wien2k.at
WWW:   http://www.imc.tuwien.ac.at/TC_Blaha
--
___
Wien mailing list
Wien@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

[Wien] Question about large RKmax

2017-09-14 Thread Kyohn Ahn 
Dear WIEN2k users

I'm sorry.
In the last mail, the link was broken,,

I have a problem about large RKmax.
Please see the attached file:
There are some parts of bandstructure of my system
with various RKmax values.

In my case only a large RKmax (11~12) shows good results.
Is there anyone who had similar experiences?

# Details:
# The smallest Rmt = 2.5 (atomic distances are pretty long) /
# Heavy elements from Os to Bi line /

My biggest concern, really, is that
whether the results are in the "numerical linear dependency" problem
or not. Is there any other method to check the issue?

Thank you for reading this email.
Any response in this regard will be very helpful for me.

With best regards

- Kyohn
___
Wien mailing list
Wien@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