Error problem... Please help me~

[hieido]

 I'm calculating the Spectrum transmission by ATK package, but after 1hour calculate, the calculation was failed.
in the log file they said:
                                                                     
** Back Engine Exception : Diagonalization error, overlap matrix not positive definite. This may be caused by atoms that are too close to each other or situated in equivalent positions, or (in the Extended Huckel model) a too low value of interaction_max_range
** Location of Exception : mathutils.cpp:1084

This application has requested the Runtime to terminate it in an unusual way.
Please contact the application's support team for more information.
 Please help me~ as soon as possible.
Thanks in advance~

[kstokbro]

please post your input file, such that we can check if some of the problems suggested in the error message are present,

[hieido]

The input file is attached at below~
Please help me check the error~
thank you so much!

[Anders Blom]

Your hydrogen atom with index 345 should not be there, it's very close to a Si atom, and you should not have any H atoms inside the nanowire. Probably the passivation added it by mistake, it's always a good idea to check the outcome of that function carefully. In this case I used the tool "Select Tools>Close Neighbors" in VNL 12.2 and immediately found this problem.

[Anders Blom]

Sorry, let me correct that somewhat; it's not the H-Si distance which is the problem, although it's relevant too.

Actually, there are 2 H atoms on top of each other (one of them is 345). You must delete both of them to avoid having H atoms inside the nanowire.

This system is extremely difficult to auto-passivate because of the Si atoms with an unusual bonding character to the S atoms. You should passivate them manually (note that the other Si atom, without the H atom problem, only has 2 bonds, not 4).

[hieido]

thanks for your advice,
I tried to fix the error by:
- the layer Silicon connect directly with gold layer was passivated by Hydrogen, each Silicon bond with 1 Hydrogen.
- try to passivated by Hydrogen by manually way.
- then opt with max steps: 100, max size: 1Angstron, max force:  0.05, max stress: 0.05
But it still error. When I tried to make the bulk device, between bonding Si-Au also have bonding between Au-H.
Can somebody help me?

[kstokbro]

The script runs on my laptop with 4 giga byte.

Did you run it on a parallel machine? If yes check if it runs in seriel and read the parallel guide on our website to make sure you use the code properly in parallel mode.
 
I changed the script a title to improve the quality of the calculation, i.e providing periodic boundary conditions and constraining gold atoms during relaxation.

[hieido]

Dear Sir,
Thanks a lot for your advice, I'm using PC to calculate aularge.py file, however there are some error. Therefor, I decided to reduce the number of gold in the structure, just use only 1 layer Gold in the both left and right sight to make it's more easier, and then constrained all the gold before optimize.
Could You please help me to check the Py* file?
and can you answer my question:
1- how many bonding between Au and Si is possible? (2 or 4)
2- if my Silicon layer were passivated by H, so the layer which connected directly with Au was passivated by H or not? if passivated so how many H for each Si? (1 or 2)
3- Usually, Silicon have 4 bonds, so if Si connect with Sulfur, there is how many bonding will be formed?
Thanks you in advance.
 

[kstokbro]

I took a look at your structure,

a) In the terminating (111) surface you have 3 dangling bonds. When you cleave silicon in the (111) direction you always get the (111) surface with only 1 dangling bond. Thus, you should correct that.

b) Similar with the contact to gold, use terminating surfaces with only 1 dangling bond.

c) you have a finite silicon cluster, why dont you make the structure as a gold-silicon interface, you can set it up using the interface builder in 12.2

d) I would not relax it as a molecule, I would relax it using bulk, but a very thin gold slab, and fix the position of the gold atoms.

[hieido]

Dear kstokbro,
Thank you so much for your advice.

By using the new version ATK 12.2.0 I have already finish in building the structure,
1- the interface of Au and Si
2- Si passivated by H
Then I'm using on a parallel machine to calculate, but I have some question about parallel machine: Could You help me.... because i'm just an amateur~
1-  after calculation is finish, i just got the (*log) file... so how and where can I save it on the NetCDF (*.nc) file?
2- because the parallel machine base on Linux program. so how to convert it to open on the window program? 
-------------------
I tested a simple calculation:
Module load: module load atk/atk-12.2.0
#> mpiexec -n 6 atkpython test.py >test.nc
-------------------
After 50mins the calculate finished and I got 1 output file: test.nc (but it's seem be *log file)

[kstokbro]

 You directed your output into the test.nc file, thus it gets corrupted
use the command:
mpiexec -n 6 atkpython test.py >test.log

this will direct output to the file test.log

[hieido]

thanks kstokbro.
I got the *nc file after the *log file finish.
but another problem is I can not run this *nc file on the win version. How can convert file from linux version to win-version?
could you please teach me~ thanks