QuantumATK Forum

QuantumATK => General Questions and Answers => Topic started by: Sabyasachi Sen on January 12, 2013, 07:50

Title: Iteration control parameters not effective to remove a convergence problem
Post by: Sabyasachi Sen on January 12, 2013, 07:50

Dear ATK Forum

I attach herewith an input file where iteartion control parametrs have been included. The output is too large to be attached.

Sabyasachi Sen

Title: Re: Iteration control parameters not effective to remove a convergence problem
Post by: Anders Blom on January 15, 2013, 11:19

Is there any particular reason why you run this a spin-polarized? I don't see any reason to expect any spin-dependence in this system, none of the elements involved should exhibit any magnetic behavior. Although I may be wrong of course... But perhaps it's at least faster to try different convergence tests without spin, as it reduces time-consumption.

The structure is reasonable (heavy, of course), but it's worth noticing that the Au inside the central region is a bit thin. If you have 3 layers in the electrode, I always make sure to have at least one more in the central region, so 4 in your case.

Finally, the SingleZeta basis sets are often really not useful, and having it for Gold here is probably the primary reason for the bad convergence. I understand SingleZetaPolarized will increase the computation time quite a bit, but it is most likely necessary.