Re: [Wien] Basic question regarding mixing parameter & force optimization

2021-10-06 Thread shamik chakrabarti 
Dear Prof. Marks,

   Thank you for your explanations Sir. Comments
are very informative.

with regards,

On Wed, 6 Oct 2021 at 18:30, Laurence Marks 
wrote:

> Some mixer and optimization comments, to clarify.
>
> 99.999% of cases where there are convergence problems are due to:
>
> 1) Inappropriate physics. A bad model does not have to have a fixed-point
> solution, that is there may be no converged solution. (Nothing says there
> has to be.)
>
> 2) Inappropriate numerics. For instance, a functional that jumps around
> very noisily may not be smooth. I know this can happen for 4f electrons
> using -eece.
>
> 3) Discontinuity at the solution. There is such a discontinuity
> (electronic phase transition) for systems such as isolated Fe atoms without
> spin, T=0.
>
> 4) User error. I recently ran a case with 16 symmetry elements which only
> should have had 8. While it came close, it would not converge.
>
> Sometimes the pseudo-forces are quite bad with large density residues, so
> MSR1a can be slow. There can also be issues if the electron-phonon coupling
> is large.
>
> PORT is sometimes better, particularly with repetitive calculations such
> as lattice parameter optimizations. However, it is very sensitive to
> inconsistent forces and energies.
>
> In my opinion NEWT & NEWT1 should be removed. My opinion.
>
> Always look first at how the model is setup.
>
> --
> Professor Laurence Marks
> Department of Materials Science and Engineering
> Northwestern University
> www.numis.northwestern.edu
> "Research is to see what everybody else has seen, and to think what
> nobody else has thought" Albert Szent-Györgyi
> ___
> 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
>


-- 
Dr. Shamik Chakrabarti
Research Fellow
Department of Physics
Indian Institute of Technology Patna
Bihta-801103
Patna
Bihar, India
___
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] Basic question regarding mixing parameter & force optimization

2021-10-06 Thread Laurence Marks 
Some mixer and optimization comments, to clarify.

99.999% of cases where there are convergence problems are due to:

1) Inappropriate physics. A bad model does not have to have a fixed-point
solution, that is there may be no converged solution. (Nothing says there
has to be.)

2) Inappropriate numerics. For instance, a functional that jumps around
very noisily may not be smooth. I know this can happen for 4f electrons
using -eece.

3) Discontinuity at the solution. There is such a discontinuity (electronic
phase transition) for systems such as isolated Fe atoms without spin, T=0.

4) User error. I recently ran a case with 16 symmetry elements which only
should have had 8. While it came close, it would not converge.

Sometimes the pseudo-forces are quite bad with large density residues, so
MSR1a can be slow. There can also be issues if the electron-phonon coupling
is large.

PORT is sometimes better, particularly with repetitive calculations such as
lattice parameter optimizations. However, it is very sensitive to
inconsistent forces and energies.

In my opinion NEWT & NEWT1 should be removed. My opinion.

Always look first at how the model is setup.

--
Professor Laurence Marks
Department of Materials Science and Engineering
Northwestern University
www.numis.northwestern.edu
"Research is to see what everybody else has seen, and to think what
nobody else has thought" Albert Szent-Györgyi
___
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] Basic question regarding mixing parameter & force optimization

2021-10-06 Thread Tran, Fabien 
I think that the original question of Shamik concerns PORT, and not MSR1. In 
some cases, in particular with non-standard potentials, I could converge only 
with PORT and with a very small mixing (typically 0.02). Then, I was checking 
what happens if the calculation is restarted (after a save_lapw) with normal 
mixing scheme (MSR1 with 0.20), and in most cases (but not all) it was ok (no 
pseudo-convergence).


From: Wien  on behalf of Laurence 
Marks 
Sent: Wednesday, October 6, 2021 9:48 AM
To: A Mailing list for WIEN2k users
Subject: Re: [Wien] Basic question regarding mixing parameter & force 
optimization

On Wed, Oct 6, 2021, 2:20 AM Tran, Fabien 
mailto:fabien.t...@tuwien.ac.at>> wrote:
0.01 should in principle leads to slower convergence, except of course if 0.2 
is too large and does not allow for convergence at all.

Sorry, but this is not how the mixing code works. The input parameter mainly 
controls the maximum unpredicted step, and small values trigger a few internal 
parameters to be more conservative. If you do "grep :MIX *.scf" you can see how 
the GREED is dynamically changing.

The commonly seen statement in dft folklore "reduce the mixing factor if it 
does not converge" is a misunderstanding of the math.

If the convergence can be achieved properly with 0.2 and 0.01, then the results 
should be identical.


From: Wien 
mailto:wien-boun...@zeus.theochem.tuwien.ac.at>>
 on behalf of shamik chakrabarti 
mailto:shamik15041...@gmail.com>>
Sent: Wednesday, October 6, 2021 8:29 AM
To: A Mailing list for WIEN2k users
Subject: [Wien] Basic question regarding mixing parameter & force optimization

Dear Wien2k users,

 I have few questions regarding mixing parameters & 
force optimization
(1) If we reduce the mixing parameter from 0.2 to say 0.01, whether the 
convergence will come slowly?
(2) Whether it will be accurate?
(3) If we do not get force convergence with PORT should we go for NEWT/NEW1?

Any response will be eagerly awaited.

with regards,

--
Dr. Shamik Chakrabarti
Research Fellow
Department of Physics
Indian Institute of Technology Patna
Bihta-801103
Patna
Bihar, India
___
Wien mailing list
Wien@zeus.theochem.tuwien.ac.at<mailto:Wien@zeus.theochem.tuwien.ac.at>
https://urldefense.com/v3/__http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien__;!!Dq0X2DkFhyF93HkjWTBQKhk!Hu2SdWCMGmWk7-YRYOy96bo0-6RCuS84ucxA-PrBMmDS4R9YUM23AeAocqPdSVzhyvAgUg$
SEARCH the MAILING-LIST at:  
https://urldefense.com/v3/__http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html__;!!Dq0X2DkFhyF93HkjWTBQKhk!Hu2SdWCMGmWk7-YRYOy96bo0-6RCuS84ucxA-PrBMmDS4R9YUM23AeAocqPdSVwsCn6ceg$
 small
___
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] Basic question regarding mixing parameter & force optimization

2021-10-06 Thread Laurence Marks 
NEWT is obsolete.

Also, as I said before, unfortunately Fabien's answer iwas not really
right. 🤓

--
Professor Laurence Marks
Department of Materials Science and Engineering
Northwestern University
www.numis.northwestern.edu
"Research is to see what everybody else has seen, and to think what
nobody else has thought" Albert Szent-Györgyi

On Wed, Oct 6, 2021, 2:32 AM shamik chakrabarti 
wrote:

> Dear Prof. Tran,
>
>  Sorry to bother you again. What about changing PORT
> to NEWT/NEW1 for force convergence?
>
> with regards,
>
> On Wed, 6 Oct 2021 at 12:57, shamik chakrabarti 
> wrote:
>
>> Dear Prof. Tran,
>>
>> Thank you for your reply. It removes my confusion :)
>>
>> On Wed, 6 Oct 2021 at 12:50, Tran, Fabien 
>> wrote:
>>
>>> 0.01 should in principle leads to slower convergence, except of course
>>> if 0.2 is too large and does not allow for convergence at all.
>>>
>>> If the convergence can be achieved properly with 0.2 and 0.01, then the
>>> results should be identical.
>>>
>>> 
>>> From: Wien  on behalf of
>>> shamik chakrabarti 
>>> Sent: Wednesday, October 6, 2021 8:29 AM
>>> To: A Mailing list for WIEN2k users
>>> Subject: [Wien] Basic question regarding mixing parameter & force
>>> optimization
>>>
>>> Dear Wien2k users,
>>>
>>>  I have few questions regarding mixing
>>> parameters & force optimization
>>> (1) If we reduce the mixing parameter from 0.2 to say 0.01, whether the
>>> convergence will come slowly?
>>> (2) Whether it will be accurate?
>>> (3) If we do not get force convergence with PORT should we go for
>>> NEWT/NEW1?
>>>
>>> Any response will be eagerly awaited.
>>>
>>> with regards,
>>>
>>> --
>>> Dr. Shamik Chakrabarti
>>> Research Fellow
>>> Department of Physics
>>> Indian Institute of Technology Patna
>>> Bihta-801103
>>> Patna
>>> Bihar, India
>>> ___
>>> 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
>>> 
>>>
>>
>>
>> --
>> Dr. Shamik Chakrabarti
>> Research Fellow
>> Department of Physics
>> Indian Institute of Technology Patna
>> Bihta-801103
>> Patna
>> Bihar, India
>>
>
>
> --
> Dr. Shamik Chakrabarti
> Research Fellow
> Department of Physics
> Indian Institute of Technology Patna
> Bihta-801103
> Patna
> Bihar, India
> ___
> Wien mailing list
> Wien@zeus.theochem.tuwien.ac.at
>
> https://urldefense.com/v3/__http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien__;!!Dq0X2DkFhyF93HkjWTBQKhk!AeH8ocuBlj6QnBeXz1yjLTXMMMVGavZsN8yjUCEhdnkdHVA_SY5PHr52gUxZHAItMn3_6w$
> SEARCH the MAILING-LIST at:
> https://urldefense.com/v3/__http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html__;!!Dq0X2DkFhyF93HkjWTBQKhk!AeH8ocuBlj6QnBeXz1yjLTXMMMVGavZsN8yjUCEhdnkdHVA_SY5PHr52gUxZHAJ5ciM8zw$
>
___
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] Basic question regarding mixing parameter & force optimization

2021-10-06 Thread Laurence Marks 
On Wed, Oct 6, 2021, 2:20 AM Tran, Fabien  wrote:

> 0.01 should in principle leads to slower convergence, except of course if
> 0.2 is too large and does not allow for convergence at all.
>

Sorry, but this is not how the mixing code works. The input parameter
mainly controls the maximum unpredicted step, and small values trigger a
few internal parameters to be more conservative. If you do "grep :MIX
*.scf" you can see how the GREED is dynamically changing.

The commonly seen statement in dft folklore "reduce the mixing factor if it
does not converge" is a misunderstanding of the math.

>
> If the convergence can be achieved properly with 0.2 and 0.01, then the
> results should be identical.
>
> 
> From: Wien  on behalf of shamik
> chakrabarti 
> Sent: Wednesday, October 6, 2021 8:29 AM
> To: A Mailing list for WIEN2k users
> Subject: [Wien] Basic question regarding mixing parameter & force
> optimization
>
> Dear Wien2k users,
>
>  I have few questions regarding mixing parameters
> & force optimization
> (1) If we reduce the mixing parameter from 0.2 to say 0.01, whether the
> convergence will come slowly?
> (2) Whether it will be accurate?
> (3) If we do not get force convergence with PORT should we go for
> NEWT/NEW1?
>
> Any response will be eagerly awaited.
>
> with regards,
>
> --
> Dr. Shamik Chakrabarti
> Research Fellow
> Department of Physics
> Indian Institute of Technology Patna
> Bihta-801103
> Patna
> Bihar, India
> ___
> Wien mailing list
> Wien@zeus.theochem.tuwien.ac.at
>
> https://urldefense.com/v3/__http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien__;!!Dq0X2DkFhyF93HkjWTBQKhk!Hu2SdWCMGmWk7-YRYOy96bo0-6RCuS84ucxA-PrBMmDS4R9YUM23AeAocqPdSVzhyvAgUg$
> SEARCH the MAILING-LIST at:
> https://urldefense.com/v3/__http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html__;!!Dq0X2DkFhyF93HkjWTBQKhk!Hu2SdWCMGmWk7-YRYOy96bo0-6RCuS84ucxA-PrBMmDS4R9YUM23AeAocqPdSVwsCn6ceg$
>  small
>
___
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] Basic question regarding mixing parameter & force optimization

2021-10-06 Thread shamik chakrabarti 
Dear Prof. Tran,

 Sorry to bother you again. What about changing PORT to
NEWT/NEW1 for force convergence?

with regards,

On Wed, 6 Oct 2021 at 12:57, shamik chakrabarti 
wrote:

> Dear Prof. Tran,
>
> Thank you for your reply. It removes my confusion :)
>
> On Wed, 6 Oct 2021 at 12:50, Tran, Fabien 
> wrote:
>
>> 0.01 should in principle leads to slower convergence, except of course if
>> 0.2 is too large and does not allow for convergence at all.
>>
>> If the convergence can be achieved properly with 0.2 and 0.01, then the
>> results should be identical.
>>
>> 
>> From: Wien  on behalf of shamik
>> chakrabarti 
>> Sent: Wednesday, October 6, 2021 8:29 AM
>> To: A Mailing list for WIEN2k users
>> Subject: [Wien] Basic question regarding mixing parameter & force
>> optimization
>>
>> Dear Wien2k users,
>>
>>  I have few questions regarding mixing parameters
>> & force optimization
>> (1) If we reduce the mixing parameter from 0.2 to say 0.01, whether the
>> convergence will come slowly?
>> (2) Whether it will be accurate?
>> (3) If we do not get force convergence with PORT should we go for
>> NEWT/NEW1?
>>
>> Any response will be eagerly awaited.
>>
>> with regards,
>>
>> --
>> Dr. Shamik Chakrabarti
>> Research Fellow
>> Department of Physics
>> Indian Institute of Technology Patna
>> Bihta-801103
>> Patna
>> Bihar, India
>> ___
>> 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
>>
>
>
> --
> Dr. Shamik Chakrabarti
> Research Fellow
> Department of Physics
> Indian Institute of Technology Patna
> Bihta-801103
> Patna
> Bihar, India
>


-- 
Dr. Shamik Chakrabarti
Research Fellow
Department of Physics
Indian Institute of Technology Patna
Bihta-801103
Patna
Bihar, India
___
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] Basic question regarding mixing parameter & force optimization

2021-10-06 Thread shamik chakrabarti 
Dear Prof. Tran,

Thank you for your reply. It removes my confusion :)

On Wed, 6 Oct 2021 at 12:50, Tran, Fabien  wrote:

> 0.01 should in principle leads to slower convergence, except of course if
> 0.2 is too large and does not allow for convergence at all.
>
> If the convergence can be achieved properly with 0.2 and 0.01, then the
> results should be identical.
>
> 
> From: Wien  on behalf of shamik
> chakrabarti 
> Sent: Wednesday, October 6, 2021 8:29 AM
> To: A Mailing list for WIEN2k users
> Subject: [Wien] Basic question regarding mixing parameter & force
> optimization
>
> Dear Wien2k users,
>
>  I have few questions regarding mixing parameters
> & force optimization
> (1) If we reduce the mixing parameter from 0.2 to say 0.01, whether the
> convergence will come slowly?
> (2) Whether it will be accurate?
> (3) If we do not get force convergence with PORT should we go for
> NEWT/NEW1?
>
> Any response will be eagerly awaited.
>
> with regards,
>
> --
> Dr. Shamik Chakrabarti
> Research Fellow
> Department of Physics
> Indian Institute of Technology Patna
> Bihta-801103
> Patna
> Bihar, India
> ___
> 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
>


-- 
Dr. Shamik Chakrabarti
Research Fellow
Department of Physics
Indian Institute of Technology Patna
Bihta-801103
Patna
Bihar, India
___
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] Basic question regarding mixing parameter & force optimization

2021-10-06 Thread Tran, Fabien 
0.01 should in principle leads to slower convergence, except of course if 0.2 
is too large and does not allow for convergence at all.

If the convergence can be achieved properly with 0.2 and 0.01, then the results 
should be identical.


From: Wien  on behalf of shamik 
chakrabarti 
Sent: Wednesday, October 6, 2021 8:29 AM
To: A Mailing list for WIEN2k users
Subject: [Wien] Basic question regarding mixing parameter & force optimization

Dear Wien2k users,

 I have few questions regarding mixing parameters & 
force optimization
(1) If we reduce the mixing parameter from 0.2 to say 0.01, whether the 
convergence will come slowly?
(2) Whether it will be accurate?
(3) If we do not get force convergence with PORT should we go for NEWT/NEW1?

Any response will be eagerly awaited.

with regards,

--
Dr. Shamik Chakrabarti
Research Fellow
Department of Physics
Indian Institute of Technology Patna
Bihta-801103
Patna
Bihar, India
___
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] Basic question regarding mixing parameter & force optimization

2021-10-05 Thread Laurence Marks 
I am 200% certain that these questions arise because Wien2k is not giving
you the result you think it should. I am 2000% certain it is not converging
because you have made mistakes. You must check that you have a valid
calculation.

On Wed, Oct 6  wrote:

> Dear Wien2k users,
>
>  I have few questions regarding mixing parameters
> & force optimization
> (1) If we reduce the mixing parameter from 0.2 to say 0.01, whether the
> convergence will come slowly?
>

Do not do this.

(2) Whether it will be accurate?
>

Not a relevant question.

(3) If we do not get force convergence with PORT should we go for NEWT/NEW1?
>

No. If you do not get force convergence with PORT it means that you have
invalid patameter

>
> Any response will be eagerly awaited.
>
> with regards,
>
> --
> Dr. Shamik Chakrabarti
> Research Fellow
> Department of Physics
> Indian Institute of Technology Patna
> Bihta-801103
> Patna
> Bihar, India
> ___
> Wien mailing list
> Wien@zeus.theochem.tuwien.ac.at
>
> https://urldefense.com/v3/__http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien__;!!Dq0X2DkFhyF93HkjWTBQKhk!Gix5BplczH55Vcufg1h1aDLA_D_xJJMnNXTrFKrB7we1nsdA1G33qfAzQddGqIrl3aHQpg$
> SEARCH the MAILING-LIST at:
> https://urldefense.com/v3/__http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html__;!!Dq0X2DkFhyF93HkjWTBQKhk!Gix5BplczH55Vcufg1h1aDLA_D_xJJMnNXTrFKrB7we1nsdA1G33qfAzQddGqIptYc0C0w$
>
___
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