k2gamma assertion error: complex MO coefficients

There are several points in the k2gamma code (specifically in the mo_k2gamma function) where there are assertions with no description of what causes the issue or how to resolve it. Mostly these are related to matrices that should be real, having significant imaginary components. When I come across one of these assertion errors, what should I do to reduce the errors so that my system/calculation can pass the assertion?

Here is an example that triggers an assertion error:

from pyscf.pbc import scf
from pyscf.pbc.gto import Cell
import numpy as np


# ----- Construct a 1-dimensional O2 chain unit cell ----- #
cell = Cell()
cell.atom = [
    ["O", (0,0,0)],
    ["O", (2,0,0)]
]

cell.dimension = 1
cell.basis = '6-31g'
cell.a = np.eye(3)*[4,10,10]
cell.verbose = 4
cell.unit = 'A'
cell.build()

kpts = cell.make_kpts([3,1,1])

# -------- KHF calculation ------- #
mf = scf.KRHF(cell, kpts).density_fit()
e_hf = mf.kernel()


from pyscf.pbc.tools.k2gamma import mo_k2gamma
mo_k2gamma(cell, mf.mo_energy, mf.mo_coeff, mf.kpts, kmesh=None)

[tberkel]

Hi all, Another way to say Zhihao’s point: at the noninteracting level, this system is a metal (it’s equally-spaced O atoms). An RHF calculation of the supercell that Max suggested probably doesn’t converge. The converged k-point solution, which breaks lattice symmetry as Zhihao said, is a CDW insulator.

On Tue, Jan 11, 2022 at 5:58 PM Max @.***> wrote:

Okay that very interesting! I naively assumed that since the supercell calculation is more “flexible” (has fewer symmetry translations assumed), any k-point solution could also be exactly translated to a supercell solution. But I guess this is only true if you leave the orbitals complex. Thanks a lot

— Reply to this email directly, view it on GitHub https://github.com/pyscf/pyscf/issues/1161#issuecomment-1010441801, or unsubscribe https://github.com/notifications/unsubscribe-auth/ADGAJC6M2FWEX4RFJ3EOUD3UVSYZRANCNFSM5LXGEQZA . Triage notifications on the go with GitHub Mobile for iOS https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Android https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub.

You are receiving this because you are subscribed to this thread.Message ID: @.***>

[zhcui]

Hi Max,

There exists a set of orbitals in the supercell describing the same occupied space, but cannot be realized.

I think we can always get a set of real orbitals from a supercell calculation, but that solution would be different from the one from the above k-point calculation.