Print Page - problem in saving seperate bias nc file with different bias

QuantumATK => General Questions and Answers => Topic started by: yhzhou80 on June 1, 2013, 07:35

Title: problem in saving seperate bias nc file with different bias
Post by: yhzhou80 on June 1, 2013, 07:35

Dir sir:
   I come across the problem when i want to have seperate bias.nc file under different bias. The attacted files are the message of errors in calculation and the input file, respectively.
   thanks
   your sincerely.
   zhou

Title: Re: problem in saving seperate bias nc file with different bias
Post by: Anders Blom on June 1, 2013, 23:13

Small typo in the script. Change inUnitsof to inUnitsOf (case matters).

Title: Re: problem in saving seperate bias nc file with different bias
Post by: yhzhou80 on June 3, 2013, 04:01

Dear sir,
   Thank you very much for your help. Then i come across another problem in the calculation. The total energy， the transmission spectra, etc, can be calculated, howerer, the calculations stopped in the device density of states. In other word, the device density of states can't be caculated. In the input.py, exchange_correlation = LSDA.PZ is set. Then what is the problem? Thank you again for your help!
   Sincerely,
   zhou.

Title: Re: problem in saving seperate bias nc file with different bias
Post by: Anders Blom on June 3, 2013, 21:58

"Stopped" as in it's hanging, or it terminated with error message?

Title: Re: problem in saving seperate bias nc file with different bias
Post by: yhzhou80 on June 4, 2013, 00:51

the calculation "Stopped" as in it's hanging: Calculating DOS :===========

Title: Re: problem in saving seperate bias nc file with different bias
Post by: Anders Blom on June 4, 2013, 01:49

I'm afraid I have no logical explanation for that... You will first need to check if it happens every time, or maybe it was just an accident once.

Title: Re: problem in saving seperate bias nc file with different bias
Post by: yhzhou80 on June 4, 2013, 03:27

No，I check it several times.

Title: Re: problem in saving seperate bias nc file with different bias
Post by: Anders Blom on June 4, 2013, 03:38

In that case my best guess is that it runs out of memory. You can try to rerun only the device DOS calculation (using nlread - there is no need to rerun the calculation again, once it has converged), while making sure the machine is not busy with too many other tasks.

Btw, which version of ATK?

Title: Re: problem in saving seperate bias nc file with different bias
Post by: yhzhou80 on June 4, 2013, 15:28

Thanks for your help. It isn't promblem of the memory. First, my twoprobe system is not big, every time it stopped in the place of DDOS when i change the order of calculation contents; second, many calculations can be done in this cpu with atk of 2008 version. In the calculation, we use the version of ATK 11.2.0.

Title: Re: problem in saving seperate bias nc file with different bias
Post by: Anders Blom on June 4, 2013, 15:50

I'm afraid I can't help much more. It may be a particular feature in this system that causes it. Given that it occurs half-way through, maybe there is a particular energy that has problems, so try a different number of energy points to avoid it. But this ATK version is old, and not even the correct 11.2 version (we released 3 bug fix versions, the final one was 11.2.3), but it's also an unsupported version by now which we don't offer support on.

Title: Re: problem in saving seperate bias nc file with different bias
Post by: yhzhou80 on June 4, 2013, 17:35

Thank you very much for your help. And i find it's the energy point just several minutes ago. In fact, our team have made a plan to update the version soon.

Title: Re: problem in saving seperate bias nc file with different bias
Post by: Anders Blom on June 4, 2013, 17:43

I recall we fixed some bugs related to symmetry finding, this could have caused it. I can't promise 11.2.3 would have fixed it, but generally it's important to stay updated on our website or here on the Forum regarding bug-fix releases.

Title: Re: problem in saving seperate bias nc file with different bias
Post by: yhzhou80 on June 11, 2013, 05:32

Dear Blom, when i run the sript, it did output seperate bias.nc file, for example 0.1.nc, 0.2.nc, but in these nc files, there is nothing execpt several configurations. What's wrong with the input file.

Title: Re: problem in saving seperate bias nc file with different bias
Post by: Anders Blom on June 11, 2013, 05:44

I don't have your full script, but the main idea is to do

Code

device_configuration = geometry
for bias in some voltages:
    set bias on calculator
    device_configuration.update()
    nlsave("file bias.nc", device_configuration)
    transmission = TransmissionSpectum()
    nlsave("file bias.nc", transmission)
    some other quantity = Analysis()
    nlsave("file bias.nc", some other quantity)

If you are only seeing the configuration in the NC file, you are probably missing the indentation, so you are doing

Code

device_configuration = geometry
for bias in some voltages:
    set bias on calculator
    device_configuration.update()
    nlsave("file bias.nc", device_configuration)
transmission = TransmissionSpectum()
nlsave("file bias.nc", transmission)
some other quantity = Analysis()
nlsave("file bias.nc", some other quantity)

See the difference? The last lines are not part of the loop.

QuantumATK Forum

QuantumATK => General Questions and Answers => Topic started by: yhzhou80 on June 1, 2013, 07:35