I wrote that part of Inelastica a few years ago. Without looking in the code I would say that the “bond currents” are in units of transmission, i.e., in the low bias limit the current is given by G0 I_ij V (Go: conductance quantum, I_ij: bond currents, and V: bias). If you want you could divide it into orbital currents by changing the code without too much trouble.
As already said, adding them up over a plane separating the two leads should give the total transmission. There are similar quantities calculated for IETS signals although those are more complicated to explain since those depend on energy ... -Magnus ----------------------------------------------- Magnus Paulsson Associate Professor Dept. of Physics and Electrical Engineering Linnaeus University Phone: +46-480-446308 Mobile: +46-70-6942987 On 30 Aug 2016, at 09:19, Gabriele Penazzi <g.pena...@gmail.com<mailto:g.pena...@gmail.com>> wrote: Hi Luk, I jump in. You can define, instead of the bond currents, something like a bond transmission. Such quantity, integrated on a plane, will give you the total transmission (a good sanity check). Therefore it will not depend on Vext and will be unitless. Some refers to it as transmission pathways. I'm not familiar with Inelastica but from the formula this is what is calculated there. If you look in the supplementary material of the Nature Chemistry by G.Solomon you refer to in your first mail you can find something along those lines (see eq.11 in suppl.mat.). Anyway even though these object are very useful to get a physical insight, you should be a bit careful in giving a strict physical interpretation akin to current densities. As you see in the work from Todorov, these quantities are, differently from the total current, basis-dependent and therefore they are not uniquely defined. Best, Gabriele On 08/29/2016 08:09 PM, Luk Keh wrote: Dear Nick, thanks for your reply, that paper was very insightful. I still have some questions to get more understanding if you hopefully don't mind. So, in the EigenChannels.py subroutine, the bond current for atoms (ij) is calculated via 4 * pi * Im[H_ij * D_ji] where D is the DOS of the considered electrode which is obtained by the resp. spectral function. As I can see, this corresponds to eq. 100 or 101 (not sure here). In both cases I don't see the prefactor G0 (G0*e) and for eq. 101, the external bias Vext as prefactor misses too - which would yield zero bond currents for Vext = 0. What I'm trying to understand is which unit the bond currents have and why they exist for Vext = 0 which I tried out. Thanks alot and best regards, Luk 2016-08-29 16:42 GMT+02:00 Nick Papior <nickpap...@gmail.com<mailto:nickpap...@gmail.com> <mailto:nickpap...@gmail.com>>: Sorry, it is the proxy used. Here: http://dx.doi.org/10.1088/0953-8984/14/11/314 <http://dx.doi.org/10.1088/0953-8984/14/11/314> 2016-08-29 16:30 GMT+02:00 Luk Keh <lukke...@gmail.com<mailto:lukke...@gmail.com> <mailto:lukke...@gmail.com>>: Dear Nick, thanks for your instant reply. It seems that I need a login for the link you provided. Can you provide another mirror or the paper's title? Thanks alot, Luk 2016-08-29 16:18 GMT+02:00 Nick Papior <nickpap...@gmail.com<mailto:nickpap...@gmail.com> <mailto:nickpap...@gmail.com>>: This paper is excellent in explaining the details concerning bond-currents: http://dx.doi.org.globalproxy.cvt.dk/10.1088/0953-8984/14/11/314 <http://dx.doi.org.globalproxy.cvt.dk/10.1088/0953-8984/14/11/314> PS. In the next release of siesta, transiesta/tbtrans also enables the calculation of bond-currents. 2016-08-29 16:11 GMT+02:00 Luk Keh <lukke...@gmail.com<mailto:lukke...@gmail.com> <mailto:lukke...@gmail.com>>: Dear users and developers, could somebody tell me which unit the 'bond currents' in the .curr files produced by Inelastica have? Are those in fact transmissions (since they don't vanish without bias, i.e. [f_L - f_R] = 0 => I_mn = 0) or actual currents (in Ampere)? Also I would like to know how those currents are calculated. I found some papers, e.g. http://www.nature.com/nchem/journal/v2/n3/full/nchem.546.html <http://www.nature.com/nchem/journal/v2/n3/full/nchem.546.html> but I'm not sure about this. Thanks and best regards, Luk -- Kind regards Nick -- Kind regards Nick -- Gabriele Penazzi mobile: +49 (0) 151 19650383 skype: gabriele.penazzi
