Solver does not initialize indices of G0_iw properly

[mravlje]

Solver initialization creates Green's function indices as an empty list, which then prevents e.g. fit_tail acting on the GF to work properly.

[mravlje]

The issue is related to branch 1.2dev.

[pseth]

See here for a minimal example: https://gist.github.com/pseth/168fed2220bb6b49c898

[tayral]

Hi Jernej,
which solver are you referring to? cthyb or cthyb_segment ? each of them
have their dedicated issue lists.

Thomas

Le 24/07/14 11:43, mravlje a écrit :

The issue is related to branch 1.2dev.

—
Reply to this email directly or view it on GitHub
#108 (comment).

[pseth]

This is for cthyb, but the problem is related to the indicesR that Michel looked at some time ago.

[pseth]

I should also add that the problem is resolved by defining a new gf explicitly with indices as follows:

g_w = GfImFreq(indices = [0], beta=beta)
for name, gblock in S.G_tau:
      g_w <<= g_w + 0.5 * (1.0/num_orbitals) * Fourier(gblock)
g_w.fit_tail([[[0.0,0.0,1.0]]],5,0.8*(2*1025)*3.14/beta,0.95*(2*1025)*3.14/beta)

[parcollet]

The issue is that C++ created gf have by default an empty list of indices,
which is sort of strange indeed.
@mferrero : Why did we do this ?