move h5 ref page from dft_tools and update

doc/documentation.rst

@@ -68,6 +68,18 @@ in the corresponding guide page of the DFTTools documentation:
      - ✗
      - `conv_generalhk <https://triqs.github.io/dft_tools/latest/guide/conv_generalhk.html>`_
 
+Resulting hdf5 archive
+======================
+
+All converters write into the same, standardized hdf5 layout that is read by
+:py:class:`SumkDFT <triqs_dft_tools.sumk_dft.SumkDFT>`. The groups and datasets
+are documented in :ref:`hdfstructure`.
+
+.. toctree::
+   :maxdepth: 2
+
+   h5structure
+
 Tutorials
 =========
 

python/triqs_dftkit/vasp/converter.py

@@ -548,7 +548,9 @@ def convert_symmetry_input(self, ctrl_head, orbits, symm_subgrp):
         n_atoms = 1
         perm = [0]
         n_orbits = len(orbits)
-        SP = ctrl_head['ns']
+# Note the difference in name conventions: 'ns' counts the spin channels,
+# whereas SP is a flag, as in convert_dft_input() above
+        SP = ctrl_head['ns'] - 1
         SO = ctrl_head['nc_flag']
         time_inv = [0]
         mat = [numpy.identity(1)]

python/triqs_dftkit/wannier90/converter.py

@@ -763,8 +763,9 @@ def read_all_wannier90_data(n_spin_blocks, dim_corr_shells, w90_seed, add_lambda
         Number of bands
     k_mesh_from_umat : np.ndarray[n_k, 3] of float
         The k points as used in wannier for consistency. None if not bloch_basis
-    centres: np.ndarray[n_spin_blocks, 3, 3] of float or None
-        Centres of wannier functions
+    centres: np.ndarray[n_spin_blocks, n_wannier, 3] of float or None
+        Centres of the wannier functions in cartesian coordinates (Angstrom).
+        None if seedname_centres.xyz is not present.
     """
     spin_w90name = ['_up', '_down']
     wannier_hr = []