[SIESTA-L] transiesta convergence test and intel-mpi version

[Yelda Kadıoglu]

Hello Developers,
In the manual, it says that if we do not use 100 kpoints, we need to do a
convergence test, but how should we do the kpoint convergence test? It
takes a very long time to draw I-V graphs using each of 20 40 60 100 k
points with different voltages in the periodic direction. This work will
not be completed within the limited time we can use on our corporate
servers with over 60 kpoints.
Do you think it would be enough to just look at the transmission graphics
for 0V or what should we look for for transiesta convergence test ?

Another point is that the same job works on some computers but does not
work on some computers due to the Bad DM normalization error. Siesta4.1.5
compiled with intel-2018 is installed on our servers. Should we compile
this version with 2019 intel? Or does it work more smoothly with gcc?
Thanks in advance
Yelda

-- 
SIESTA is supported by the Spanish Research Agency (AEI) and by the European 
H2020 MaX Centre of Excellence (http://www.max-centre.eu/)

[SIESTA-L] Transiesta convergence

[MANOJ N MATTUR]

Hi,

PFA my fdf file for scattering region...

It's just not converging. It didn't with cut off 200Ry. Now I have reduced
it to 150Ry, still hasn't converged in 24hrs, job's still running. Any
suggestions on what could be done to speed up the convergence...

Regards,
Manoj

-- 
*Disclaimer: *This email and any files transmitted with it are confidential 
and intended solely for the use of the individual or entity to whom they 
are addressed. If you have received this email in error please notify the 
system manager. This message contains confidential information and is 
intended only for the individual named. If you are not the named addressee 
you should not disseminate, distribute or copy this e-mail. Please notify 
the sender immediately by e-mail if you have received this e-mail by 
mistake and delete this e-mail from your system. If you are not the 
intended recipient you are notified that disclosing, copying, distributing 
or taking any action in reliance on the contents of this information is 
strictly prohibited.


scattering.fdf
Description: application/vnd.fdf

-- 
SIESTA is supported by the Spanish Research Agency (AEI) and by the European 
H2020 MaX Centre of Excellence (http://www.max-centre.eu/)

[SIESTA-L] transiesta convergence Qt and Qt0 not same

[RAJAN SINGH]

Dear Siesta users

After relaxation of nanoribbon when I am running my transiesta
calculations, Qt0 and Qt values are not same. And my transiesta didnt
converge what should I do?

Efermi from SIESTA   :-3.47054
Total charge  [Qt0]  :   414.0
Charge in update region   [Qc]   :   133.70127
Charge outside update region  [Qcn]  :   280.29873
Left electrode[L]:   133.72936
Left electrode/device [L-C]  : 3.91751
Device[C]:   133.70127
Device/right electrode[C-R]  : 4.65178
Right electrode   [R]:   133.34561
Other [O]: 4.65448

 Reading GF file, with title:
   nanoribbon.TSGFL
 Title: 'Generated GF file'

 Reading GF file, with title:
   nanoribbon.TSGFR
 Title: 'Generated GF file'

ts-charge:OL  L-CC  C-RR   Qt
ts-charge:4.654  133.7293.918  329.5484.652  133.346  609.847

Thank You

With Regards
Rajan

Re: [SIESTA-L] Transiesta Convergence problem in zigzag nanoribbon of Graphene

[Younas Khan]

My Pleasure.

Younas

On Sun, Dec 13, 2015 at 1:17 PM, Riya Rogers 
wrote:

> thank you so much
>
> Regards
> Riya
>
> On Mon, Dec 7, 2015 at 10:29 AM, Younas Khan  > wrote:
>
>> Well I think You have not specified the type of run in your input file.
>> Put 'MD.TypeOfRun=CG' and try again.
>> Best of luck.
>>
>> Younas Khan
>>
>> On Wed, Dec 2, 2015 at 5:30 PM, Riya Rogers 
>> wrote:
>>
>>> Here is my fdf file.
>>>
>>> Regards
>>> Riya
>>>
>>> On Mon, Nov 23, 2015 at 9:20 PM, Younas Khan <
>>> younaskhan.phys...@gmail.com> wrote:
>>>
 Kindly share your input file so that it becomes easy to find out where
 you are making a mistake.

 Thanks


 On Mon, Nov 23, 2015 at 4:52 PM, Riya Rogers 
 wrote:

>
> Hello everyone
>
> I am trying to simulate the Zigzag Graphene Nanoribbon (hydrogen
> passivated)with boron doping for my project work. But transiesta won't
> converge.
>
> Which parameters should I vary to achieve convergence?
> Can anyone help me in converging transiesta?
>
> Thanks
>
> Regards
>
>

>>>
>>
>

Re: [SIESTA-L] Transiesta Convergence problem in zigzag nanoribbon of Graphene

[Riya Rogers]

Here is my fdf file.

Regards
Riya

On Mon, Nov 23, 2015 at 9:20 PM, Younas Khan 
wrote:

> Kindly share your input file so that it becomes easy to find out where you
> are making a mistake.
>
> Thanks
>
>
> On Mon, Nov 23, 2015 at 4:52 PM, Riya Rogers 
> wrote:
>
>>
>> Hello everyone
>>
>> I am trying to simulate the Zigzag Graphene Nanoribbon (hydrogen
>> passivated)with boron doping for my project work. But transiesta won't
>> converge.
>>
>> Which parameters should I vary to achieve convergence?
>> Can anyone help me in converging transiesta?
>>
>> Thanks
>>
>> Regards
>>
>>
>


nanoribbon.fdf
Description: Binary data

Re: [SIESTA-L] transiesta convergence tests

[Nick Papior Andersen]

You need to be below all states in your system, all states for all k.

Kind regards Nick

2013/8/28 mahmoud ali vas...@gmail.com

 Thank you Nick for your answer.
 I have found somewhere saying to start the contour bellow the lowest
 eigenvalue of the system, but as you know for each k point there are a
 bunch of energy values. So which energy value should I take?

 M


 On Wed, Aug 28, 2013 at 12:53 AM, Nick Papior Andersen 
 nickpap...@gmail.com wrote:

 Dear Mahmoud Ali

 First notice that tbtrans does not do any self-consistency loops.
 However, typically the required number of k-points for converging the
 transmission is much larger than that needed for converging the density
 matrix in transiesta. Hence you should converge your transmission as a
 function of number of k-points.

 On your question about transiesta. Yes, transiesta convergence is
 dependent on the contour parameters.

 Currently you don't have a high degree of control of how the integral in
 the complex plane is performed, but if you don't see any change in the SCF
 loop, then you most probably have chosen converged parameters.

 A general notice is that for the equilibrium contour you typically don't
 need too many contour points (as we have a smooth integral in the complex
 plane).

 But as the non-equilibrium contour has to be located close to the real
 axis you need a lot of contour points. The article of Brandbyge et al.
 emphasizes these points above, so a close read should guide you.

 Lastly, the small imaginary part is very important for placing the
 non-equilibrium contour. A large Eta means that the DOS terms in the
 non-equilibrium contour will be smoothened out, whereas a small Eta will
 not alter the DOS terms in the non-equilibrium contour greatly.
 Hence, if you have a very hard time on converging for V /= 0, and you
 feel that you have a high number of contour points on the non-equilibrium
 contour, you can try and increase the Eta value.

 Again, the article of Brandbyge et al. has many details explained and I
 encourage you to read (and re-read it :) ).
 DOI: 10.1103/PhysRevB.65.165401

 Kind regards Nick


 2013/8/28 mahmoud ali vas...@gmail.com

 Dear users,
 As you know in usual siesta calculations we have mesh cutoff and k
 points convergence tests, my concern if there are any convergence tests in
 transiesta or tbtrans calculations.
 Is the choice of contour parameters makes any difference in the results?

 Regards,

[SIESTA-L] transiesta convergence tests

[mahmoud ali]

Dear users,
As you know in usual siesta calculations we have mesh cutoff and k points
convergence tests, my concern if there are any convergence tests in
transiesta or tbtrans calculations.
Is the choice of contour parameters makes any difference in the results?

Regards,

Re: [SIESTA-L] transiesta convergence tests

[mahmoud ali]

Thank you Nick for your answer.
I have found somewhere saying to start the contour bellow the lowest
eigenvalue of the system, but as you know for each k point there are a
bunch of energy values. So which energy value should I take?

M


On Wed, Aug 28, 2013 at 12:53 AM, Nick Papior Andersen nickpap...@gmail.com
 wrote:

 Dear Mahmoud Ali

 First notice that tbtrans does not do any self-consistency loops. However,
 typically the required number of k-points for converging the transmission
 is much larger than that needed for converging the density matrix in
 transiesta. Hence you should converge your transmission as a function of
 number of k-points.

 On your question about transiesta. Yes, transiesta convergence is
 dependent on the contour parameters.

 Currently you don't have a high degree of control of how the integral in
 the complex plane is performed, but if you don't see any change in the SCF
 loop, then you most probably have chosen converged parameters.

 A general notice is that for the equilibrium contour you typically don't
 need too many contour points (as we have a smooth integral in the complex
 plane).

 But as the non-equilibrium contour has to be located close to the real
 axis you need a lot of contour points. The article of Brandbyge et al.
 emphasizes these points above, so a close read should guide you.

 Lastly, the small imaginary part is very important for placing the
 non-equilibrium contour. A large Eta means that the DOS terms in the
 non-equilibrium contour will be smoothened out, whereas a small Eta will
 not alter the DOS terms in the non-equilibrium contour greatly.
 Hence, if you have a very hard time on converging for V /= 0, and you feel
 that you have a high number of contour points on the non-equilibrium
 contour, you can try and increase the Eta value.

 Again, the article of Brandbyge et al. has many details explained and I
 encourage you to read (and re-read it :) ).
 DOI: 10.1103/PhysRevB.65.165401

 Kind regards Nick


 2013/8/28 mahmoud ali vas...@gmail.com

 Dear users,
 As you know in usual siesta calculations we have mesh cutoff and k points
 convergence tests, my concern if there are any convergence tests in
 transiesta or tbtrans calculations.
 Is the choice of contour parameters makes any difference in the results?

 Regards,

[SIESTA-L] TranSiesta Convergence

[Manjeet Bhatia]

Dear Siesta User,

I have problem in TranSiesta Conversgence. At zero bias it converges but at
higher voltage does not converge even after 120 iterations. I am using
these parameters.
Please help me.

PAO.BasisSizeDZP
PAO.EnergyShift  0.005 Ry

XC.functionalLDA
XC.authors   CA

LatticeConstant  1. Ang
%block LatticeVectors
   10.9467890.00 0.00
0.0010.0797180.00
0.000.00 50.664718
%endblock LatticeVectors
ElectronicTemperature  600 K
MeshCutoff  300.0 Ry
MaxSCFIterations 300
DM.MixingWeight  0.01
DM.NumberPulay   5
DM.Tolerance 1.d-5