Thanks, Anton! Creating a fake system is a good idea, but it seems that it is still not a finalized system as it returns the following error:
"TypeError: The system needs to be finalized." Here is the piece of code I use: syst_c=[] for item in finilized_sys.graph: if finalized_sys.sites[item[0]].family in [a1, b1] : syst_c.append syst_slice = SimpleNamespace(sites=syst_c)densities, box = kwant.plotter.interpolate_density(syst=syst_slice, density=densities) Best,Hadi On Friday, August 16, 2019, 9:16:29 AM GMT+1, Anton Akhmerov <anton.akhmerov...@gmail.com> wrote: Hi Hadi, The intended way would be to extend the density vector with np.nan, so that it has a correct length. However since you mention performance, I'm not sure if this would achieve what you are looking for. Inspecting the source of interpolate_density [1], we see that the only way the system is used is to get sites = np.array([s.pos for s in syst.sites]) Therefore we can mock the necessary properties of the system by using the following code: from types import SimpleNamespace # make a fake system that only contains the correct sites and isn't actually a system syst_slice = SimpleNamespace(sites=(site for site in syst.sites if where(site))) It is certainly a hack, so a more reliable solution would be to take over the bits of the source of interpolate_density and the low level function that it uses [2] and re-purpose for your needs. [1]: https://gitlab.kwant-project.org/kwant/kwant/blob/v1.4.1/kwant/plotter.py#L1793-1872 [2]: https://gitlab.kwant-project.org/kwant/kwant/blob/v1.4.1/kwant/plotter.py#L1637-1701 On Fri, 16 Aug 2019 at 10:03, Hadi Zahir <zahir....@gmail.com> wrote: > > Dear Kwant developers, > > I have a finalized closed system ''sys_f'' for which I computed the local > charge densities "ch_densities", which is basically a vector of > length=len(sys_f.sites). > > For complexity reason, I'd like to feed a slice of sys_f to the > interpolate_density() as: > > >> kwant.plotter.interpolate_density(syst=sys_p, density=ch_densities_p) > > where sys_p is a slice of sys_f and ch_densities_p is the corresponding > density vector. > Could you please let me know how to do it? > > Best regards, > Hadi
