Two Probe calculation converge trivially

[Sandra]

I am carrying out an electron density calculation using the atk code (2008.10 version).

I began with 3 and 4 layers of Au in each side. I have increased the size of K-point grid  from(5,5,50)to (7,7,50), the value of mesh cut-off (from 180 to 300 Ry.), the number of screening layers ( 5 in each side). I have played with the contour integration parameters (increasing the integral lower bound (from 3 to 5 and 7 Ry) and the Circle Points (from 30 to 50 point). I also have reduced the Diagonal Mixing Parameter to 0.05…unsuccessfully!! I have kept the temperature at 300k. With all possible combination of parameters, the charge in the TwoProbe calculation still is zero.
Can someone help me??
I have attached one of the tests.
Thanks

[Anders Blom]

2008.10 version is not supported at all anymore.

[Sandra]

Thank you very much!!

[Anders Blom]

It's not that we don't want to help, but actually for some legal reasons, we no longer have access ourselves to 2008.10, since that was formally developed by Atomistix,  not QuantumWise.

But also, there are not many places you can expect to get support for an 8 year old software package... We'll be happy to offer a nice price for an upgrade to the latest ATK which is sometimes 10x faster than 2008.10 and contains a huge amount of new features.

[Sandra]

Do not worry!!!
The problem with the trivial convergence in ATK  is exasperating!
I am going to use the new version soon, I have some questions about the building a molecular junction in this new version. Can I ask about that in this forum??

Thank you!!

[Anders Blom]

Sure. One thing that's improved in the new version is we now use a double contour which stabilizes convergence. We rarely ever see the q=0 problem any more.