QuantumATK Forum

QuantumATK => General Questions and Answers => Topic started by: gradius on May 24, 2011, 05:07

Title: ATKError: inverse(DZMatrix const &) : Could not LU factorize!
Post by: gradius on May 24, 2011, 05:07

Dear all:
    After Electrodes and Equivalent Bulk Calculation, ATK 2008.10 give two errors:

1]# Pulay mixing inversion failed. Using only last step.

2]ATKError: inverse(DZMatrix const &) : Could not LU factorize!

I already look up previous posts which talk about this error and try almost all the methods to solve this problem but failed. There are two attatchments about the two-probe structure and scf calculation input file.

Title: Re: ATKError: inverse(DZMatrix const &) : Could not LU factorize!
Post by: nori on May 30, 2011, 02:55

I could get the SCF converged with graphene2.py and graphene2_scf2.py.
Although I've changed several parameters in order to reduce calculation cost, changing twoProbeAlgorithmParameters() would be the most important for better SCF convergence.

By the way, Can you access the latest version(11.2.3)?
If so, I strongly recommend that you use ATK 11.2.3 because a SCF convergence is more stable and an accuracy is improved.

Title: Re: ATKError: inverse(DZMatrix const &) : Could not LU factorize!
Post by: gradius on June 1, 2011, 07:25

Thank you very much! But I only have ATK-2008.10