QuantumATK Forum

QuantumATK => General Questions and Answers => Topic started by: jdgayles16 on July 5, 2010, 22:03

Title: errors
Post by: jdgayles16 on July 5, 2010, 22:03

I get this error

Code

** Back Engine Exception 21 : incompatible size of electrode and central region density in A direction
** Location : twoprobeutils.cpp:502

I assume it means my system is not configured properly, however it seems my bulk converged nicely.

Also I calculated the DOS for a bulk device and which i was quite pleased with except for the dips i see just above 0 eV and below 0.5 eV. Also there seem to be a little dip just before the peak above zero eV. I attache the picture. Does anyone have any suggestions?

Title: Re: errors
Post by: Nordland on July 5, 2010, 22:38

Could you show your configuration or is it top secret?

Title: Re: errors
Post by: jdgayles16 on July 6, 2010, 00:05

yeah the config is kind of secret for the two probe setup (the first error) but its okay if i show the electrode on which i did the bulk dos calculation on. its attached.

and here are the coordinates

Code

# Set up lattice
vector_a = [30.0, 0.0, 0.0]*Angstrom
vector_b = [0.0, 20.0, 0.0]*Angstrom
vector_c = [0.0, 0.0, 9.844]*Angstrom
lattice = UnitCell(vector_a, vector_b, vector_c)

# Define elements
elements = [Carbon, Carbon, Carbon, Carbon, Carbon, Carbon, Carbon, Carbon,
            Carbon, Carbon, Carbon, Carbon, Carbon, Carbon, Carbon, Carbon,
            Carbon, Carbon, Carbon, Carbon, Carbon, Carbon, Carbon, Carbon,
            Carbon, Carbon, Carbon, Carbon, Carbon, Carbon, Carbon, Carbon,
            Carbon, Carbon, Carbon, Carbon, Carbon, Carbon, Carbon, Carbon,
            Carbon, Carbon, Carbon, Carbon, Carbon, Carbon, Carbon, Carbon]

# Define coordinates
cartesian_coordinates = [[  9.9962115,  10.       ,   0.361443 ],
                         [ 20.0033235,  10.       ,   0.36186  ],
                         [ 11.4021825,  10.       ,   0.808086 ],
                         [ 14.2817305,  10.       ,   0.893618 ],
                         [ 15.7180385,  10.       ,   0.893629 ],
                         [ 18.5973595,  10.       ,   0.808277 ],
                         [ 20.8764155,  10.       ,   1.469241 ],
                         [  9.1233085,  10.       ,   1.468799 ],
                         [ 12.0353665,  10.       ,   2.099487 ],
                         [ 13.5147845,  10.       ,   2.099576 ],
                         [ 16.4851035,  10.       ,   2.099529 ],
                         [ 17.9644105,  10.       ,   2.099633 ],
                         [  9.1232435,  10.       ,   2.730952 ],
                         [ 20.8767565,  10.       ,   2.731411 ],
                         [ 11.4020705,  10.       ,   3.390884 ],
                         [ 14.2817135,  10.       ,   3.305606 ],
                         [ 15.7181005,  10.       ,   3.305556 ],
                         [ 18.5975435,  10.       ,   3.39106  ],
                         [  9.9961815,  10.       ,   3.838215 ],
                         [ 20.0034125,  10.       ,   3.838556 ],
                         [ 12.2063675,  10.       ,   4.560441 ],
                         [ 13.6111745,  10.       ,   4.560555 ],
                         [ 16.3885035,  10.       ,   4.560544 ],
                         [ 17.7933005,  10.       ,   4.560551 ],
                         [  9.9962115,  10.       ,   5.283445 ],
                         [ 20.0033235,  10.       ,   5.283863 ],
                         [ 11.4021825,  10.       ,   5.730088 ],
                         [ 14.2817305,  10.       ,   5.81562  ],
                         [ 15.7180385,  10.       ,   5.815631 ],
                         [ 18.5973595,  10.       ,   5.730279 ],
                         [ 20.8764155,  10.       ,   6.391244 ],
                         [  9.1233085,  10.       ,   6.390802 ],
                         [ 12.0353665,  10.       ,   7.021489 ],
                         [ 13.5147845,  10.       ,   7.021578 ],
                         [ 16.4851035,  10.       ,   7.021531 ],
                         [ 17.9644105,  10.       ,   7.021636 ],
                         [  9.1232435,  10.       ,   7.652954 ],
                         [ 20.8767565,  10.       ,   7.653413 ],
                         [ 11.4020705,  10.       ,   8.312887 ],
                         [ 14.2817135,  10.       ,   8.227608 ],
                         [ 15.7181005,  10.       ,   8.227559 ],
                         [ 18.5975435,  10.       ,   8.313062 ],
                         [  9.9961815,  10.       ,   8.760218 ],
                         [ 20.0034125,  10.       ,   8.760559 ],
                         [ 12.2063675,  10.       ,   9.482443 ],
                         [ 13.6111745,  10.       ,   9.482557 ],
                         [ 16.3885035,  10.       ,   9.482547 ],
                         [ 17.7933005,  10.       ,   9.482553 ]]*Angstrom

# Set up configuration
bulk_configuration = BulkConfiguration(
    bravais_lattice=lattice,
    elements=elements,
    cartesian_coordinates=cartesian_coordinates
    )

Title: Re: errors
Post by: Nordland on July 6, 2010, 07:36

Okay, all I need is the cell from the central region together with the electrode.

Title: Re: errors
Post by: jdgayles16 on July 6, 2010, 08:25

Sorry I lost the geometry of the twoprobe. I was fiddling around and I replaced my script file but the new one seemed to work fine.  ;D


Do you have any suggestions about the Bulk DOS.


Also I was trying to cleave the electrode and I was getting problems saying that it must be a cubic bravis lattice. any ideas?


Thanks for the help
Jacob

Title: Re: errors
Post by: Anders Blom on July 6, 2010, 08:57

If you have the NC file, it contains the geometry too.

Yeah, only cubic lattices can be cleaved for now, but if your electrode is set up by hand in the proper way already, there's no need to cleave it. Note that the new way of making two-probes does not require "cleaving", you just need the 3 periodic systems for left and right electrodes, and the central cell.

All will be a bit clearer as soon as we publish the upgrade guide, should be online later today, I expect (look on the Download page, in the 10.8 section).

Title: Re: errors
Post by: jdgayles16 on July 6, 2010, 11:55

My job aborted before it could write an NC file :(.

thanks for the other tip

Title: Re: errors
Post by: Anders Blom on July 6, 2010, 12:02

Check for "py" files in /tmp (Linux) or your user's TEMP directory (on Windows). The Job Manager will place temporary Python files there for running; if it crashed, chances are it didn't manage to remove them afterwards (as it will on success).

On Windows you can find out where your TEMP dir is by

echo $TEMP (in Cygwin)

echo %TEMP% (in "cmd")

One of those scripts might be the one you ran, and so you can get your geometry from there. If you ran via the Job Manager, that is.

Title: Re: errors
Post by: jdgayles16 on July 6, 2010, 21:17

Thanks i ran into a similar problem with another config. I will send it to you if you have time to look

Title: Re: errors
Post by: Anders Blom on July 6, 2010, 22:13

Can you post it here? Just replace all elements by Hydrogen or remove the "elements" variables.

Title: Re: errors
Post by: jdgayles16 on July 6, 2010, 23:28

I tried to but the script was too long....


Any suggestions about the bulk DOS in the first panel?

Title: Re: errors
Post by: Anders Blom on July 6, 2010, 23:36

Too long to attached, because of Forum rules on max size? That must be a large script... :)
Send it by mail then.

DOS: May be a feature in the band structure, or a result of too few k-points? Try to correlate the band structure to the DOS in those areas.

Title: Re: errors
Post by: jdgayles16 on July 7, 2010, 01:44

here is the attached

Title: Re: errors
Post by: jdgayles16 on July 7, 2010, 01:45

Also what is the command for printing only for the master node

Title: Re: errors
Post by: Nordland on July 7, 2010, 07:30

If you use nlprint, it is MPI safe.

If you want more advanced printing functionality, then you will have to use the command processIsMaster() to check which node is the master node.

Title: Re: errors
Post by: Nordland on July 7, 2010, 07:33

Was there a problem with this geometry? because it works fine for me with hydrogens.

Title: Re: errors
Post by: Anders Blom on July 7, 2010, 08:46

Note that processIsMaster() is not available in the beta-version 10.8.b1, but it is in 10.8, to be released today (fingers crossed).

Code: python

nlprint("my string")

is your best option, in general.

Title: Re: errors
Post by: Anders Blom on July 7, 2010, 08:52

Please send the real geometry by private mail. I don't think this is a geometric issue, but something else.

Am I right that the error does not appear immediately, but after some time of calculations? How long? Does it make it through one iteration of the two-probe SCF? Do you have the full output log?

Title: Re: errors
Post by: jdgayles16 on July 7, 2010, 09:18

Yeah It dies at the end of the equivalent bulk calc. just before the device calc.

Title: Re: errors
Post by: jdgayles16 on July 7, 2010, 09:19

Thanks for the nlprint suggestion it worked  :)

Title: Re: errors
Post by: Nordland on July 7, 2010, 18:45

The issues has been resolved now. You can also fix it manually, if you wnet to get a head start.
In the script you will shut have to set the grid-mesh-cutoff to be the same for the electrodes and the central region.

Title: Re: errors
Post by: jdgayles16 on July 7, 2010, 19:38

Thanks  ;D