Some doubt about MPSH result

[zhangguangping]

Hi,as I kown,the MPSH can be regarded as the value of the selected atoms under the bias and inlcuded the effect of the surroundings.Am I right?
But how to explain the following phenomenon:

The selected atoms have 124 electrons,and should have 62 occupied orbitals if treated as isolated.And the MPSH HOMO orbital of the selected atoms should at 62 or so,and LUMO at 61 or so.

After the convergency,there are 124.096 (Mulliken) electrons on the selected atoms.So I expected the 61th (for the quntum number is started with zero) MPSH orbital is HOMO,and 62th MPSH orbital is LUMO.
And as I kown,the smallest positive number is LUMO and the biggest negative number is HOMO,as atk put the fermi energy to zero.

But the MPSH result shows:
58       -2.51662803866
59       -2.10063289295
60       -2.02243292869
61       -1.78048389083
62       -1.44400824774
63        0.936283743779
64        1.46552142409
65        1.87921738475
66        1.89000122501
67        2.33287791883
68        2.59222292105
69        2.79361757147

Since there are 63 occupied orbitals in the MPSH ,there should be 126 electrons on the selected atoms.
But in fact there are only 124.096.

I expcted the 61th (ie.62) MPSH is the HOMO,for there is 124 electrons.But now the extra 0.096 electron makes the 62th (ie.63) MPSH HOMO.
Is this so?
or
What is wrong?

[jdgayles16]

If your using ATK 2008.10 this link has a script to calculate the occupation of each level

http://quantumwise.com/forum/index.php?topic=50.0

10.8.... series does it for you =)

[zhangguangping]

If your using ATK 2008.10 this link has a script to calculate the occupation of each level

http://quantumwise.com/forum/index.php?topic=50.0

10.8.... series does it for you =)

Dear jdgayles16:
Thank you very much.I will try(that is a script for molecular system and mine is a two probe system).And does the calculation of the occupation use the fermi function to smearing it?I think it is so.
And I have found to calculate a system in equilibrium is very fast,but when calculate it under bias,it takes so long time compared to equilibrium to finish a scf of  TwoProbe self-consisten calculation.Is this normal?Or is there something to speed up it. I calculate the bias on the previous one.
Thanks.

[kaypu]

in the new vension 2010, atk can calculate the occupancies for Twoprobe Projected Hamiltonian energy spectrum, i want to ask about how to get the HOMO and LUMO?  is that the HOMO as the last negative energy and the LUMO as the first positive energy?  or  identify the HOMO-LUMO according to the occupancies?

[jdgayles16]

I always identify according to occupancy.

[zh]

It is more reliable to identify the HOMO-LUMO by means of the occupancies.

[zhangguangping]

It is more reliable to identify the HOMO-LUMO by means of the occupancies.

However,there is no functionality to do occupation analysis with the MPSH spectrum in 2008.10 version.

[zhangguangping]

I always identify according to occupancy.

Does the script you provide me can do it for two probe systems?

[jdgayles16]

I dont see why it shouldnt I would just test the script and play around with the calculateEigenstateOccupations() function

[zhangguangping]

I dont see why it shouldnt I would just test the script and play around with the calculateEigenstateOccupations() function

I think this script is for a molecule system,I don't think it work for the MPSH eigenstates spectrum.
If it works,how to use it?
change:
self_consistent_calculation = restoreSelfConsistentCalculation('c6h4s2.nc')
to
self_consistent_calculation = restoreSelfConsistentCalculation('your_twoprobe.nc')
But your_twoprobe.nc contains the consistent density for the whole two probe not only for the selected MPSH atoms.How can the script kown we want to do the occupation analysis on the MSPH eigenstate?
 

[zh]

The charge obtained by Mulliken population analysis is not unique, and it may depend on the choice of basis set. Since the molecular region selected for the MPSH analysis is not isolated in a two-probe system anymore, charge transfer may occur between the molecule and the surface screening layers. Therefore, sometimes it may fail to identify the so-called "HOMO" and "LUMO" in MPSH analysis by counting the total electrons that are estimated from the isolated molecule or from the Mulliken population analysis.  

In the MPSH analysis of a two-probe, the MSPH eigenvalues are given with respect to the Fermi level, which is set as the average fermi energy of the electrode at a given bias and is also taken as the zero of energy. If the occupations of energy levels are not available in the MPSH analysis, the so-called "HOMO" and "LUMO" in MPSH analysis could be identified by checking the positions of MSPH eigenvalues relative to the Fermi level.

[zhangguangping]

The charge obtained by Mulliken population analysis is not unique, and it may depend on the chosen basis set. Since the molecular region selected for the MPSH analysis is not isolated in a two-probe system anymore, charge transfer may occur between the molecule and the surface screening layers. Therefore, sometimes it may fail to identify the so-called "HOMO" and "LUMO" in MPSH analysis by counting the total electrons that are estimated from the isolated molecule or from the Mulliken population analysis. 

In the MPSH analysis of a two-probe, the MSPH eigenvalues are given with respect to the Fermi level, which is set as the average fermi energy of the electrode at a given bias and is also taken as the zero of energy. If the occupations of energy levels are not available in the MPSH analysis, the so-called "HOMO" and "LUMO" in MPSH analysis could be identified by checking the positions of MSPH eigenvalues relative to the Fermi level.

Dear zh,
I think 'The charge obtained by Mulliken population analysis is not unique' is the issue for this question.
So if a occupation is available, it is convenient to see this.But 08.10 version can not do this task althogh 10.8 can do it.
Thank you very much for reminding me the role of the Mulliken population analysis.

Best regards.