QuantumATK Forum

QuantumATK => General Questions and Answers => Topic started by: BlackBarrel on April 3, 2012, 14:09

Title: Question about initial state
Post by: BlackBarrel on April 3, 2012, 14:09

Recently I noticed that if I use previously converged structures stored in .nc file as the initial state to start a new SCF cycle (with probably some changes, say, in the voltages of metallic regions), the electrodes' calculation will be skipped. Is it reasonable?

Particularly, if the bias voltages (and some parameters such as gate voltages) are changed while the structure is unchanged, is it still a good way to converge with previous converged states as initials?

Title: Re: Question about initial state
Post by: kstokbro on April 3, 2012, 14:34

The electrode calculation is independent of the central region calculation, thus, the electrode scf loop can be skipped when you change something in the central region. Fortunately, ATK has build in logic which determines when it is required to update the electrodes