I have recently started calculating transport properties using Transiesta and TBTrans. However, I have noticed that:

  1. the TS.Contours.Eq.Pole value recommended is the default. In the last version of siesta, this is 1.5eV but the value printed by tselecs.sh is 2.5eV. Does anyone know where this difference comes from?
  2. In TBTrans options, the TBT.Contours default is 'line'. I have seen online that also 'neq' is a possibility. However, in the manual, I haven't found any explanation of the difference between the two. Does anyone have any documentation about this?
  3. Lastly, is there any recommendation on the values of TBT.T.Eig and the Eta (I guess 1e-4 should be enough)?


  • 2
    $\begingroup$ It'd be better if you were to split this into three focused questions. $\endgroup$
    – Anyon
    Nov 23, 2022 at 15:43
  • $\begingroup$ Thanks for the comment. I was avoiding posting multiple questions, but if I don't receive any answer here I will follow your advice $\endgroup$
    – Laura
    Nov 23, 2022 at 15:46

1 Answer 1

  1. On testing a few systems it showed to not depend too much on these two values. The reason is that typical systems use 300 K as the electronic temperature, and the number of poles for a fixed pole energy scales as $2N kT\pi\ge E_{pole} $ with N being the lowest integer. So a lower energy means fewer integration points at the possible cost of accuracy.
  2. There is no difference. Just allowing users to use 2 words,should have been documented... Sorry.
  3. TBT.T.Eig should be an integer such that the lowest transmitting eigenchannels is approx 0, but it depends on what you are interested in. This number is only used to determine which transmission eigenvalues to save, and thus the result does not depend on the integer. If you take too high a number it will be cut to the maximum allowed. Do note that the disk space required for the output file depends on this number.
  4. The eta value should generally be as low as possible. A too high value will smear out the states a lot and will not require a dense energy grid. A too low value might require your energy grid to be very dense to capture the states. Recall that the Eta value is the smearing in a Lorentzian distribution. Note that this applies to both TranSiesta and Tbtrans runs.
  • $\begingroup$ Thank you Nick for your answer. Just to be sure before solving the question, in point 1, if I understood correctly, the previous default value of 2.5eV could give more accurate results but more expensive calculations. Right? While, in point 3, I would understand that high numbers should be used but at the cost of disk space. However, is there a range of values that I could expect to give good results? Thanks! $\endgroup$
    – Laura
    Nov 25, 2022 at 9:28
  • $\begingroup$ 1. Yes, as with any other integration, the finer the mesh, the more accurate the result is. 3.see the updated answer. $\endgroup$
    – nickpapior
    Nov 25, 2022 at 12:36
  • $\begingroup$ Thanks :) question solved! $\endgroup$
    – Laura
    Nov 25, 2022 at 13:53

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .