QuantumATK Forum

QuantumATK => General Questions and Answers => Topic started by: vboyz on May 4, 2009, 06:34

Title: error on calculation
Post by: vboyz on May 4, 2009, 06:34

i tried to calculat the spin transport , but when i use
two_probe_algorithm_parameters = twoProbeAlgorithmParameters(
    electrode_constraint = ElectrodeConstraints.Off,
    initial_density_type = InitialDensityType.NeutralAtom
)
and i use verbosity_level = 5
a error has happened:
# TwoProbe Calculation
# ----------------------------------------------------------------
# sc  0 : q =  440.00000 e
Traceback (most recent call last):
  File "c:/docume~1/admini~1/locals~1/temp/tmpslxoie.nl", line 793, in ?
    runtime_parameters = runtime_parameters
ATKError: inverse(DZMatrix const &) : Could not LU factorize!
Terminated Abnormally

I dont know how to deal with it.

Title: Re: error on calculation
Post by: Nordland on May 4, 2009, 07:09

In 9 out of 10 cases, this is due to a bad alignment of the twoprobe.

Check that the systems is correctly setup and the equivalent atoms are correct, and I think you will have solve this problem.

Title: Re: error on calculation
Post by: vboyz on May 4, 2009, 16:15

i did not think i setup uncorrectly, when i change initial_density_type = InitialDensityType.EquivalentBulk,
it was well run. so it shoud be other reasons.
Anyway ,the tutorial of the Fe-Mgo-Fe suggest we use the NeutralAtom.
I did not know the difference of those two sets.
If i choose EquivalentBulk, does the result right?

Title: Re: error on calculation
Post by: Nordland on May 4, 2009, 18:21

Neutral Atom is in general a bad idea unless the system is very unsymmetric, and therefore I would personally always use EquivalentBulk.
In the Fe-MgO-Fe case, I think it is due to the anti-polarized configuration, where the EquivalentBulk is just a waste of time, and you can gain a lot of speed by doing it.

Perhaps you can share your script with us? It would make it easier to spot the error.