Dear Joe, Dear Christoph, For testing I have adopted quantum_wire.py. The "critical" part looks like this: local_dos = kwant.ldos(sys, energy) # check size print local_dos.shape D2D = local_dos.reshape(NL,NW) kwant.plotter.map(sys, local_dos) # create dictionary to localize [i,j] ldos_dict = {tuple(site.tag): ldos for site, ldos in zip(sys.sites, local_dos)} for i in xrange(NL): for j in xrange(NW): D2D[i,j] = ldos_dict[i,j] pyplot.imshow(D2D.T, cmap=pyplot.cm.hot) pyplot.colorbar() kwant.plotter.map(sys, local_dos) I am using 64 bit Python2.7 windows version. Jerzy

"Jerzy Wrobel" writes:

...

Interestingly enough, when kwant.plotter.map(sys, ldos_array) is called again, AFTER dictionary creation, the resulting picture is changed to a random patchwork of colourful rectangles.

The creation of ldos_dict should not have any side-effects. Can you show us a script that demonstrates the problem?

Best, Christoph

On 30/06/14 13:37, Jerzy Wrobel wrote:

means probably that D2D is not a new matrix by a new name for the same thing.

Actually, `reshape` creates a new *view* into the `local_dos` array. The `D2D` and `local_dos` arrays are not the same Python object; you can verify that id(D2D) != id(local_dos) The `D2D` array (view), however, *does* reference the same memory as the local_dos array, as can be verified by checking that D2D.base is local_dos Joe