[Smeagol-discuss] End of File reached
Ivan Rungger
runggeri at tcd.ie
Wed Mar 25 13:43:00 GMT 2009
Hello,
I think there is a bit of misunderstanding. The following applies for
systems that are periodic in the x-y plane, for wires always use 1
k-point only. There is no constraint in the code that forces the number
of k-points in the leads calculation to be equal to the one of the
scattering region. So in principle you can use any value, however in the
following I write what I suggest to do:
A smeagol calculation for the extended molecule (EM) and periodic
systems consists of two parts:
1) the convergence of the charge density matrix (DM):
for this part one should always use the same number of x-y k-points in
leads and scattering region, because that will help convergence. If the
x-y k-points are different an artificial scattering is created at the
place where the leads are joined to the EM, which leads to convergence
problems. So although it is not strictly necessary, it is better to
always use the same number of x-y kpoints as in the leads.
2) the calculation of the transmission coefficient and current for the
converged DM:
Usually step 1) is the difficult/time consuming one, while 2) is much
faster. Moreover it is often the case (e.g. in magnetic tunnel
junctions) that the DM can be converged with few xy kpoints (e.g. 7x7 or
10x10), but that the transmission can only be calculated correctly if
many kpoints are used (e.g. 100x100). Therefore one can use a kgrid of
few kpoints for step 1), and then setup a separate calculation for the
TRC, where a much finer k-grid is used (using the converged DM and
setting Maxscfiterations 1, so that no self-consistency is performed).
In this case there is no problem if the number of kpoints does not match
the leads.
Hope this helps,
Ivan
Xiaohong Zheng wrote:
> Hello, Sagar and Ivan,
>
> Since calculations with different k meshes in the lead and scattering
> region go on very well, perhaps the block in the scattering region like
>
> %block kgrid_Monkhorst_Pack
> 4 0 0 0.0
> 0 4 0 0.0
> 0 0 1 0.0
> %endblock kgrid_Monkhorst_Pack
>
> is just for the transmission, not for the density matrix. The k mesh
> for the density matrix will be obtained from the lead calculation
> automatically? If this is the case, then we do not have to give the same
> k point number along x and y directions in the lead and transport file
> for the kgrid block. I do not know. Please Ivan give us the correct answer.
>
> Xiaohong
>
>
> ambavale sagar wrote:
>
>> Dear Xiaohong,
>> You are correct. There is no constraint that your k-point mesh for
>> transport calculation in scattering file and k-point mesh for density
>> matrix calculation in the same scattering file must be same.
>> But I could not see the any special parameter for defining different
>> k-point mesh for density matrix and transport calculation in the same
>> scattering file.
>> On the other way it is necessary to have k-pt mesh in x-y direction
>> same in leads file and for density matrix calculation in scattering
>> file. So I answered like that..
>>
>> Hope Ivan will correct me If there is something wrong in my
>> understanding..
>>
>>
>> Regards
>> Sagar Ambavale
>>
>> --- On *Thu, 19/3/09, Xiaohong Zheng /<exhzheng at gmail.com>/* wrote:
>>
>>
>> From: Xiaohong Zheng <exhzheng at gmail.com>
>> Subject: Re: [Smeagol-discuss] Smeagol-discuss Digest, Vol 15, Issue 9
>> To: "ambavale sagar" <sagarambavale at yahoo.co.in>
>> Cc: smeagol-discuss at lists.tchpc.tcd.ie
>> Date: Thursday, 19 March, 2009, 6:40 PM
>>
>> Hello, Sagar,
>>
>> I have a different idea about setting up the k points for the lead
>> and transport calculations. Several days ago, I tried to set
>> different k points for the lead and transport along x and y
>> directions, the calculations went on very well. Then how to
>> understand this if it is a requirement for the numbers of k points
>> along x and y directions to be the same in both lead and transport
>> calculations?
>>
>> From the discussions of ATK manual, the transport calculation will
>> use the same k mesh as the lead for the density matrix, but a
>> different k mesh for transmission can be specified. Can we or how
>> can we set a different k mesh for the transmission calculations in
>> Smeagol? Shall the k mesh for both the density matrix and the
>> transmission be exactly the same?
>>
>> Best,
>> Xiaohong
>>
>>
>> ambavale sagar wrote:
>> > Dear Serhan,
>> > Yes, number of k-points in z-direction must be 1. While no. of k
>> pts in x and y direction may have any values though they should be
>> same in both lead and transport calculations.
>> >
>> > if you may have k-points for leads
>> > 4 0 0
>> > 0 4 0
>> > 0 0 100
>> >
>> > then transport file must have
>> > 4 0 0
>> > 0 4 0
>> > 0 0 1
>> >
>> >
>> > Regards
>> > Sagar Ambavale
>> >
>> >
>> >
>> > Message: 1
>> > Date: Wed, 18 Mar 2009 06:15:24 +0200
>> > From: Serhan Yamacli <syamacli at gmail.com
>> </mc/compose?to=syamacli at gmail.com>
>> > </mc/compose?to=syamacli at gmail.com
>> </mc/compose?to=syamacli at gmail.com>>>
>> > Subject: [Smeagol-discuss] End of file reached
>> > To: smeagol-discuss at lists.tchpc.tcd.ie
>> </mc/compose?to=smeagol-discuss at lists.tchpc..tcd.ie>
>> > </mc/compose?to=smeagol-discuss at lists.tchpc.tcd.ie
>> </mc/compose?to=smeagol-discuss at lists.tchpc.tcd.ie>>
>> > Message-ID:
>> >
>> <42e7de20903172115r7f29711fk529eb6a761686679 at mail.gmail.com
>> </mc/compose?to=42e7de20903172115r7f29711fk529eb6a761686679 at mail.gmail.com>
>> >
>> </mc/compose?to=42e7de20903172115r7f29711fk529eb6a761686679 at mail.gmail.com
>> </mc/compose?to=42e7de20903172115r7f29711fk529eb6a761686679 at mail.gmail.com>>>
>> > Content-Type: text/plain; charset="iso-8859-1"
>> >
>> > Hi asa,
>> >
>> > Thank you very much for the answer. Must I use the k points as
>> >
>> > 1 0 0
>> > 0 1 0
>> > 0 0 1
>> >
>> > always during transport calculation? But I can take such as
>> >
>> > 1 0 0
>> > 0 1 0
>> > 0 0 100
>> >
>> > during leads calculation? Thanks again.
>> >
>> > Cheers
>> > Serhan
>> >
>> > 2009/3/18 asa aravindh <mails2asa at gmail.com
>> </mc/compose?to=mails2asa at gmail.com>
>> > </mc/compose?to=mails2asa at gmail.com
>> </mc/compose?to=mails2asa at gmail.com>>>
>> >
>> >
>> >
>> >
>> ------------------------------------------------------------------------
>> > Get perfect Email ID for your Resume. Get before others grab.
>> <http://in.rd.yahoo.com/tagline_dbid_9/*http://in.promos.yahoo.com/address>
>>
>> >
>> ------------------------------------------------------------------------
>> >
>> > _______________________________________________
>> > Smeagol-discuss mailing list
>> > Smeagol-discuss at lists.tchpc.tcd.ie
>> </mc/compose?to=Smeagol-discuss at lists.tchpc.tcd.ie>
>> > http://lists.tchpc.tcd.ie/listinfo/smeagol-discuss
>>
>>
>> ------------------------------------------------------------------------
>> Get rid of Add-Ons in your email ID. Get yourname at rocketmail.com. Sign
>> up now!
>> <http://in.rd.yahoo.com/tagline_dbid_11/*http://in.promos.yahoo.com/address>
>>
>
>
> _______________________________________________
> 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
=================================================
More information about the Smeagol-discuss
mailing list