Projected bandstructure cannot mix colors

[bugaboo]

If I understand the current implementation correctly, for any (k, E) point of the bandstructure the color will be determined solely by the highest projection component. For example, if a point is 0.31 s character and 0.30 p character, then it will be only red, but 0.30 s and 0.31 p point will be colored blue.

[lllangWV]

No, the projections are summed together. The sum of all projections should equal to 1 or close to it.

For instance if you do

pyprocar.bandsplot(... orbitals=[0,3])

This will sum the projection contribution from all atoms, spins, s orbitals, and the p_z orbitals.

if you do

pyprocar.bandsplot(... orbitals=[0,3], atoms=[0])

This will sum the projection contribution from the first atom, spins, s orbitals, and the p_z orbitals.

Does this answer your question or am I misunderstanding?

[bugaboo]

When I do bandsplot(... mode="overlay_orbitals"), the code plots s, p, d line segments separately without combining the colors, as far as I can tell. In the case when the d component is the largest, this would make the line segment purely green.
Also, the projections only add up to ~1 if the states are well described by l=0,1,2 orbitals centered on atoms.

[lllangWV]

In the case when the d component is the largest, this would make the line segment purely green.

Yes, this is what happens

For overlay_orbitals we separate the weights by l orbitals s, p, d, f. Essentially, we are adding 4 totals band structures. We achieve the effect by scaling the color and linewidth by the weight and choosing a sequential colormap that increases the lightness value monotonically.

Here is a summary of the implementation in eps_plot:452-505

For SrVO3 weight has a length of 3 for s, p, d

for iweight, weight in enumerate(weights):
    ....
    for ispin in spins:
         ...
         for iband in range(self.ebs.nbands):
               segments = np.delete(segments, np.where(x[1:] == x[:-1])[0], axis=0)
                    lc = LineCollection(
                        segments,
                        cmap=plt.get_cmap(color_map[iweight]),
                        norm=norm,
                        alpha=self.config.opacity[ispin],
                    )
                    lc.set_array(weight[:, iband, ispin]) # scaling of the color happens here
                    lc.set_linewidth(weight[:, iband, ispin] * linewidth[ispin]) # Scaling of the linewidth happens here
                    handle = self.ax.add_collection(lc)