Dear Artem,
some short comments:

1. You are right, it is a very good idea FIRST get some idea
about band structure, well converge things, AND THEN
select k-points and functions for plotting, not other way around.
2. In k-points other than Gamma, the wave functions are complex;
you should figure out how to plot them in a meaningful way;
it may seem cumbersome, though.
3. The number of occupied states in a nonmagnetic dielectric equals
(number of valence electrons)/2, i.e. exactly given by chemical
composition, prior to any calculation.
If magnetic moment is known, getting HOMO and LUMO for each spin channel
is equally straightforward. In a metal, this is of course more tricky,
as values from one k-point to another.

Best regards

Andrei Postnikov


> Dear Andrei,
>
> Thank you very much for your assistance. It seems that I overlooked the
> option that you pointed out.
>
> But, still there is a problem. WaveFuncKPoints block does allow to specify
> the k-points at which wavefunctions are calculated. This is how we can
> control this aspect at the stage of main Siesta program. But what if I
> simply do not know the whole band structure before running Siesta which
> means that I do not know the numbers of wavefunctions that I am
> potentially
> interested in (e.g. I am interested in the bands in vicinity of Fermi
> level
> but I do not know its value and numbers and values of the corresponding
> eigenstates). Also, before having the whole band structure calculated it
> is
> hard to determine in which k-points I want the wavefunctions to be
> calculated.
> For this reason, I thought that we could first let Siesta calculate the
> whole picture, then visualize the Band Structure and only after that
> decide
> which k-points and which eigenstates we are interested in. In other words,
> we could perform the control at the stage of using Denchar utility without
> reducing the outcome of calculations from Siesta. Does it make sense?
>
> Besides this issue I am not quite sure that I understand the meaning of
> option NumberOfEigenStates (*integer*). There is a restriction on that
> integer number (greater than the number of occupied states and less than
> the
> number of basis functions). How can I know the number of occupied states
> (even at T=0) before the calculation of Fermi level and "HOMO" band are
> done? I suppose that this question is related to the above mentioned.
>
> I would be very obliged if you could explain this aspect and help me with
> my
> initial problem.
>
> Best regards,
>
> Artem
>
>
> On Sat, May 28, 2011 at 3:52 AM, <[email protected]> wrote:
>
>>
>> >
>> > Now MY QUESTION: how to control k-point in which I want the
>> wavefunctions
>> > to
>> > be calculated, and also how to control the number of eigenstate (fist,
>> > 31st,
>> > etc), for a given k-point, for which I want the wavefunction to be
>> > calculated?  Does anybody know how to perform that controlling?
>>
>> Dear Artem,
>> I think, the description of the block
>>
>> WaveFuncKPoints
>>
>> contains the information you searches?
>> (k-points AND wavefunc. numbers to store)
>>
>> Best regards
>>
>> Andrei Postnikov
>>
>> >
>> > Many thanks for any help and your time,
>> >
>> > Artem
>> >
>>
>>
>

Responder a