[Smeagol-discuss] Smeagol examples

[Ivan Rungger]

Hello 寇良志,

  there are two problems:
1)the difference in the energy: if you are using the pgf compiler, it is
probably due to the fact that with that compiler if you use -O3 or some
similar optimization, part of the energy is wrong. I think it is enough
just to recompile atom.f with -O0, and then the energy should be OK.
2)The second problem is more serious, i.e. your TRC is definitely wrong.
The reason is that you converge to a wrong solution. I compared my
buffer.out and *CHR file, and they are identical to yours for the first
few iterations, but then they start to differ (you converge to a
solution with only 53 electrons instead of 93). What code exactly are
you using, is it smeagol-1.0b with the two patches applied (and which
siesta version is it based on)? If yes, try to rerun the calculation
with DM.NumberPulay         1 (and remove any saved DM files). And also:
try to compile the whole code with -O0 (i.e. no optimization), and see
what you get.

I think once you converge to the right solution, you should also get the
same TRC and CUR.

Cheers,

 Ivan


寇良志 wrote:
> Dear Ivan and every smeagol user
>  
> I have run the smeagol examples. However, there are some differences
> comparing with your attached output files.
>  
> For the first two examples (Au_wire_1kp and Au_wire_4kp), the results
> (including the *.TRC, *.CUR etl) are almost same to yours. But we note
> that, there are some discrepancys for the following energy parameters
> in the "out" file:
> Our results are:
> siesta: Ena     =        15.344394
> siesta: Eharris =     -1647.976798
> siesta: Etot    =     -1648.041394
> siesta: FreeEng =     -1648.041394
> Your results are:
> siesta: Ena     =       731.942850
> siesta: Eharris =      -930.824171
> siesta: Etot    =      -922.648096
> siesta: FreeEng =      -922.648096
>  
> However, the serious problems occure in the third example. The datas
> in the *.TRC and *.CUR files we obtained are at least 3 or 4 orders of
> magnitude lower than your datas. In addition, the energy differents
> described above still exist.
>  
> I attached the five output files of the third example we obtained.
>  
> Could you please provide some suggestions, and tell us what is the
> problem?
>  
> Thanks a lot !
>  
> PS: All of the parameters and pseudopotentials we used are exactly
> those you provided. The soft version to run the examples is
> siesta13flp+smeagol1.1.0.b
>  
> 2009-04-17
> ------------------------------------------------------------------------
> Kou Liangzhi
> Institute of NanoScience, NanJing University of Aeronautics and
> Astronautics, NanJing, China.
> ------------------------------------------------------------------------
> *发件人：* Ivan Rungger
> *发送时间：* 2009-04-14  23:31:28
> *收件人：* smeagol-discuss
> *抄送：*
> *主题：* [Smeagol-discuss] Smeagol examples
> Hello,
>    since different smeagol users have asked me to provide some simple
> example systems with which to check whether the output of the code they
> compiled is correct, here I post a tar archive with 3 different test
> systems:
> 1)Au_wire_1kp: A gold wire with 8 gold atoms
> 2)Au_wire_1kp: A gold wire with 8 gold atoms, 4 k-points are used in the
> plane perpendicular to the transport. There is no need to use k-points
> for a wire, here I just added them for testing the k-points routines.
> The results must be identical to the ones without k-points.
> 3)Au_Au3_Au: a gold cacitor with 3 gold atoms joining the two gold
> surfaces. The conductance of this wire at the Fermi energy is about one.
> Whereas the bias was applied in an artificial way in the first two
> examples (which are modified versions of those provided with the smeagol
> source code), this example represents a real physical system.
> All the directories contain the necessary leads and scattering-region
> input files, and also the used Au pseudopotential. These systems are
> intended for small tests, the aim is not to simulate real physical
> systems (especially the gold wires are completely unphysical at finite
> bias). All the calculations were run with the parallel version of the
> code (I suggest using that one). In the output directory there is the
> main output of the code. It contains the following files:
> 1)out: standard output
> 2)buffer.out
> 3).CHR file
> 4).TRC: transmission coefficient. There is also a rescaled *.TRC file
> called *.TRC.rescaled, where the transmission has been divided by the
> number of k-points,  the Fermi energy has been set to 0 and the units of
> the energy axis are in eV.
> 5).CUR file for the current.
> These output files can be used as reference to check the calculated results.
> Cheers,
>  Ivan
> -- 
> =================================================
> Ivan Rungger,
> School of Physics and CRANN,  
> Trinity College Dublin,  
> Dublin 2,  IRELAND  
> Phone: +353-1-8968454  
> Email: runggeri at tcd.ie
> =================================================
> _______________________________________________
> Smeagol-discuss mailing list
> Smeagol-discuss at lists.tchpc.tcd.ie
> http://lists.tchpc.tcd.ie/listinfo/smeagol-discuss


-- 
=================================================
Ivan Rungger,

School of Physics and CRANN,  
Trinity College Dublin,  
Dublin 2,  IRELAND  
Phone: +353-1-8968454  
Email: runggeri at tcd.ie
=================================================