Thank you Nick for the good answer. So I should make the convergence tests on zero bias.
Regards, Mahmoud On Thursday, September 26, 2013, Nick Papior Andersen wrote: > Dear Mahmoud > > Generally I would advice to first do the 0 bias transiesta calculation and > then continue with that DM for the calculation with a bias. It should (and > most probably will) converge faster, although you can do it the way you > propose. > Secondly, if your bias is "very large", i.e. above .5 Volt it will > probably also converge smoother by doing intermediate bias calculations. > Say you want 1 V, then calculations at 0 V, 0.5 V and then the 1 V. (or if > it is hard to converge then in steps of 0.25 V). > All of this is of course system dependent. > > Kind regards Nick > > > 2013/9/25 mahmoud ali <[email protected] <javascript:_e({}, 'cvml', > '[email protected]');>> > >> Dear siesta users, >> In the example listed here about graphene nanoribbon, the authors said >> transiesta runs for a given bias and it uses the converged density matrix >> of the previous bias step for the next bias step. Now if I wish to run >> transiesta for a given bias only, do I need to do it first for the zero >> bias then use the DM for the given bias? >> >> Regards >> > >
