lp:~nickpapior/siesta/4.1-test-upgrade-format0.92

Branch merges

Propose for merging

No branches dependent on this one.

Related source package recipes

No recipes using this branch. (?)

Create packaging recipe

Related snap packages

Related bugs

Related blueprints

756. By Nick Papior on 2017-08-29

Fixed a sign convention error in bond-currents

I still have to check that dH does it correctly but
I cannot see how it shouldn't.

Note that since sisl 0.8.6 this is also fixed.
The mistake was that in sisl 0.8.5 and prior the signs
were wrong, so effectively all previous 4.1 releases and sisl
0.8.5 showed the correct bond-currents. But now we have to correct
the mistake.

755. By Nick Papior on 2017-08-29

Added simple case of pivoting 1 no regions

Now rgn2trimat works for 1-orbital systems.

754. By Nick Papior on 2017-08-28

Removed unused variables in tbtrans

753. By Nick Papior on 2017-08-28

Enabled tbtrans calculations with 1 orbital systems

Now we may run calculations on *any* size systems. Before
it could only run on systems with *at least* 2 orbitals per
block. This has been leveraged now.

752. By Nick Papior on 2017-08-25

Added new-line when using dH and changed tbtrans: to tbt:

There is no code change.

751. By Nick Papior on 2017-08-23

Changed internal Gamma array to be i (Sigma - Sigma^\dagger)^T

Generically the scattering matrix (Gamma) is defined as:
  Gamma = i ( Sigma - Sigma^\dagger)
however, before this commit the convention was:
  Gamma = ( Sigma - Sigma^\dagger) ^ T

Now we use this convention:
  Gamma = i ( Sigma - Sigma^\dagger) ^ T

1. The complex number i makes the scattering matrix Hermitian
2. The transpose allows certain optimizations when calculating the
   transport (i.e. dot instead of matrix multiplications)

I have re-runned all k-point sampled versions of MUMPS, full, BTD
with and without bias (only bias should be important) and I have
also checked TBtrans.
The only thing missing for testing is the molecular projections
for tbtrans. This should be tested (TODO). However, there it shouldn't
matter because the Gamma are not re-calculated.

I also found a bug when writing correction eigenvalues to ASCII,
now it correctly only writes the k-averaged ones in AVCEIG.

750. By Nick Papior on 2017-08-22

Merged 4.0 r532-534

749. By Nick Papior on 2017-08-22

Added total energy contribution from an open-boundary calculation.

The total energy contribution is (grand canonical ensemble):
  Etot = Etot - e \sum_i N_i \mu_i
where N_i is the number of particles being occupied from the i'th chemical
potential, with chemical potential \mu_i.
Sadly this value depends on the choice of reference energy.
For choices:
  mu_1 = -V/2
  mu_2 = V/2
and
  mu_1 = 0
  mu_2 = V
one will find two _very_ different results. But this is a matter of choosing the
zero potential.
Perhaps we should make the "effective" chemical potentials be subtracted the
average chemical potential to have a constant potential for varying bias'.

748. By Nick Papior on 2017-08-18

Fixed wrong documentation in Diag

747. By Nick Papior on 2017-08-07

Removed unneed m_diagon_opt file