QuantumATK Forum

QuantumATK => General Questions and Answers => Topic started by: Dhirendra on April 8, 2016, 12:59

Title: Vacuum Region and Density Mesh Cutoff
Post by: Dhirendra on April 8, 2016, 12:59

Hi,

Is there any relation between the density mesh cutoff and the length of vacuum padding in a slab simulation?
I mean in two calculations, the one with 10A of vacuum and the other with 50A, do I need to increase the density mesh cutoff for the latter one?

Title: Re: Vacuum Region and Density Mesh Cutoff
Post by: Anders Blom on April 8, 2016, 13:27

No.

Title: Re: Vacuum Region and Density Mesh Cutoff
Post by: Dhirendra on April 8, 2016, 19:14

Thanks Anders, so in that case, increasing the vacuum should not increase the computational cost. Is it so?

Title: Re: Vacuum Region and Density Mesh Cutoff
Post by: Anders Blom on April 8, 2016, 22:34

Increasing the vacuum size means more grid points so there is an increase in computational time and memory,  but it's not very large if you use FFT and unless of course you make it huge.

Title: Re: Vacuum Region and Density Mesh Cutoff
Post by: Dhirendra on April 8, 2016, 23:19

Thanks Anders.

Title: Re: Vacuum Region and Density Mesh Cutoff
Post by: Dhirendra on April 9, 2016, 13:06

Hi Anders, did this check. Similar to InAs tutorial, I did Germanium slab. With 3.3nm Ge slab, and 10A, 30A and 50A vacuum padding on both sides, dmc=75 Hartree, Kmesh=5x5x1. MGGA functional c=0.94.
For 10A vacuum it too 41 iterations, 11m, for 30A vacuum 39 iterations 12m and for 50A vacuum, 31 iterations 8 min. It seems that the larger vacuum can actually lead to a faster convergence.

Title: Re: Vacuum Region and Density Mesh Cutoff
Post by: Anders Blom on April 9, 2016, 22:20

That probably indicates that the vacuum is too short in the smaller calculation.