storing and handling Wannier90 input and output files

class wannierberri.w90files.Wannier90data(seedname='wannier90', read_npz=True, write_npz_list=('mmn', 'eig', 'amn'), write_npz_formatted=True, overwrite_npz=False, formatted=(), readfiles=())

Bases: object

A class to describe all input files of wannier90, and to construct the Wannier functions

Parameters:

• seedname – str the prefix of the file (including relative/absolute path, but not including the extensions, like .chk, .mmn, etc)

• formatted – tuple(str) list of files which should be read as formatted files (uHu, uIu, etc)

• read_npz – bool if True, try to read the files converted to npz (e.g. wanier90.mmn.npz instead of wannier90.mmn)

• write_npz_list – list(str) for which files npz will be written

• write_npz_formatted – bool write npz for all formatted files

• overwrite_npz – bool overwrite existing npz files (incompatinble with read_npz)

• read_chk – bool if True, read the CheckPoint file, otherwise create a “bare” CheckPoint object and prepare for wannierisation

• kmesh_tol – float see CheckPoint

• bk_complete_tol – float see CheckPoint

chk

the checkpoint file

Type:

~wannierberri.w90files.CheckPoint or ~wannierberri.w90files.CheckPoint_bare

seedname

the prefix of the file (including relative/absolute path, but not including the extensions, like .chk, .mmn, etc)

Type:

str

wannierised

if True, the data is already wannierised (so, it can be used to create a System_w90 object)

Type:

bool

read_npz

if True, try to read the files converted to npz (e.g. wanier90.mmn.npz instead of wannier90.

Type:

bool

write_npz_list

for which files npz will be written

Type:

set(str)

write_npz_formatted

write npz for all formatted files

Type:

bool

formatted_list

list of files which should be read as formatted files (uHu, uIu, etc)

Type:

list(str)

readfiles

the list of files to be read durint the initialization. Others may be read later.

Type:

list(str)

__init__(seedname='wannier90', read_npz=True, write_npz_list=('mmn', 'eig', 'amn'), write_npz_formatted=True, overwrite_npz=False, formatted=(), readfiles=())

select_bands(win_min=-inf, win_max=inf, band_start=None, band_end=None, selected_bands=None, allow_again=False)

exclude some bands from the data

Parameters:

• win_min (float) – the minimum and maximum energy of the window. Only the bands that are ENTIRELY below win_min or ENTIRELY above win_max are EXCLUDED

• win_max (float) – the minimum and maximum energy of the window. Only the bands that are ENTIRELY below win_min or ENTIRELY above win_max are EXCLUDED

• band_start (int) – the range of bands to be included (the numeration in the current system, not in the original (if bands are selected more than once)) note : the pythonic indexing is used, so the included bands are [band_start, band_start+1, … band_end-1]

• band_end (int) – the range of bands to be included (the numeration in the current system, not in the original (if bands are selected more than once)) note : the pythonic indexing is used, so the included bands are [band_start, band_start+1, … band_end-1]

• selected_bands (array((NB,), dtype=bool) or list of int) – if present, the those bands are selected, and the other parameters are ignored

• allow_again (bool) – if False, but bands were already selected, raise an error.

Returns:

new_selected_bands – the indices of bands that were actually selected. If data files are added to the dataset later, one should manually select those bands before adding, using the select_bands method of those objects

Return type:

list of int

set_random_symmetric_projections()

Set random symmetric projections for the system, according to the symmetrizer object

write(seedname, files=None)

writes the files on disk

Parameters:

• seedname (str) – the seedname of the files (including relative/absolute path, but not including the extensions, like .chk, .mmn, etc)

• files (list(str)) – the list of files extensions to be written. If None, all files are written

class wannierberri.w90files.w90file.W90_file(autoread=False, **kwargs)

Bases: SavableNPZ

Abstract class for the files of wannier90

Parameters:

• seedname (str) – the prefix of the file (including relative/absolute path, but not including the extensions, like .chk, .mmn, etc)

• ext (str) – the extension of the file, e.g. ‘mmn’, ‘eig’, ‘amn’, ‘uiu’, ‘uhu’, ‘siu’, ‘shu’, ‘spn’

• tags (list(str)) – the tags to be saved in the npz file

• read_npz (bool) – if True, try to read the files converted to npz (e.g. wanier90.mmn.npz instead of wannier90.mmn)

• write_npz (bool) – if True, write the files to npz

• kwargs (dict) – the keyword arguments to be passed to the constructor of the file see MMN, EIG, AMN, UIU, UHU, SIU, SHU, SPN, WIN for more details

npz_tags

the tags to be saved/loaded in the npz file

Type:

list(str)

class wannierberri.w90files.AMN(seedname='wannier90', npar=2, **kwargs)

Bases: W90_file

Class to store the projection of the wavefunctions on the initial Wannier functions AMN.data[ik, ib, iw] = <u_{i,k}|w_{i,w}>

Parameters:

• seedname (str) – the prefix of the file (including relative/absolute path, but not including the extension .amn)

• npar (int) – the number of parallel processes to be used for reading

Notes

NB

number of bands

Type:

int

NW

number of Wannier functions

Type:

int

NK

number of k-points

Type:

int

data

the data projections

Type:

numpy.ndarray( (NK, NB, NW), dtype=complex)

from_w90_file(seedname, npar)

abstract method to read the necessary data from Wannier90 file

select_bands(selected_bands)

Select the bands to be used in the calculation, the rest are excluded

Parameters:

selected_bands (list of int or bool) – the indices of the bands to be used, or a boolean mask

spin_order_block_to_interlace()

If you are using an old VASP version, you should change the spin_ordering from block to interlace

spin_order_interlace_to_block()

the reverse of spin_order_block_to_interlace

class wannierberri.w90files.CheckPoint(real_lattice=None, num_wann=None, num_bands=None, num_kpts=None, wannier_centers_cart=None, wannier_spreads=None, v_matrix=None, kpt_latt=None, mp_grid=None, kmesh_tol=1e-07, bk_complete_tol=1e-05)

Bases: object

A class to store the data about wannierisation, written by Wannier90

Parameters:

• seedname (str) – the prefix of the file (including relative/absolute path, but not including the extension .chk)

• kmesh_tol (float) – tolerance to distinguish different/same k-points

• bk_complete_tol (float) – tolerance for the completeness relation for finite-difference scheme

get_AABB_qb(mmn, transl_inv=False, eig=None, phase=None, sum_b=False)

Returns the matrix elements AA or BB(if eig is not Flase) in the Wannier gauge

Parameters:

• mmn (~wannierberri.w90files.MMN) – the overlap matrix elements between the Wavefunctions at neighbouring k-points

• transl_inv (bool) – if True, the band-diagonal matrix elements are calculated using the Marzari & Vanderbilt translational invariant formula

• eig (~wannierberri.w90files.EIG) – the eigenvalues of the Hamiltonian, needed to calculate BB (if None, the matrix elements are AA)

• phase (np.ndarray(shape=(num_wann, num_wann, nnb), dtype=complex)) – the phase factors to be applied to the matrix elements (if None, no phase factors are applied)

• sum_b (bool) – if True, the matrix elements are summed over the neighbouring k-points. Otherwise, the matrix elements are stored in a 5D array of shape (num_kpts, num_wann, num_wann, nnb, 3)

Returns:

• np.ndarray(shape=(num_kpts, num_wann, num_wann, nnb, 3), dtype=complex) (if sum_b=False)

• or np.ndarray(shape=(num_kpts, num_wann, num_wann, nnb, 3), dtype=complex) (if sum_b=True) – the q-resolved matrix elements AA or BB in the Wannier gauge

get_AA_q(mmn, transl_inv=False)

A wrapper for get_AA_qb with sum_b=True see get_AA_qb() for more details

get_AA_qb(mmn, transl_inv=False, phase=None, sum_b=False)

A wrapper for get_AABB_qb with eig=None see get_AABB_qb() for more details

get_BB_qb(mmn, eig, phase=None, sum_b=False)

a wrapper for get_AABB_qb to evaluate BB matrix elements. (transl_inv is disabled) see get_AABB_qb() for more details

get_CCOOGG_qb(mmn, uhu, antisym=True, phase=None, sum_b=False)

Returns the matrix elements CC, OO or GG in the Wannier gauge

Parameters:

• mmn (MMN) – the overlap matrix elements between the Wavefunctions at neighbouring k-points

• uhu (UHU or UIU) – the matrix elements uhu or uiu produced by pw2wannier90

• antisym (bool) – if True, the antisymmetric piece of the matrix elements is calculated. Otherwise, the full matrix is calculated

• phase (np.ndarray(shape=(num_wann, num_wann, nnb), dtype=complex)) – the phase factors to be applied to the matrix elements (if None, no phase factors are applied)

• sum_b (bool) – if True, the matrix elements are summed over the neighbouring k-points. Otherwise, the matrix elements are stored in a 6D array of shape (num_kpts, num_wann, num_wann, nnb, nnb, 3)

Returns:

• np.ndarray(shape=(num_kpts, num_wann, num_wann, nnb, nnb, 3), dtype=complex) (if sum_b=False)

• or np.ndarray(shape=(num_kpts, num_wann, num_wann, nnb, nnb, 3), dtype=complex) (if sum_b=True) – the q-resolved matrix elements CC, OO or GG in the Wannier gauge

get_CC_qb(mmn, uhu, phase=None, sum_b=False)

A wrapper for get_CCOOGG_qb with antisym=True see get_CCOOGG_qb() for more details

get_GG_qb(mmn, uiu, phase=None, sum_b=False)

A wrapper for get_CCOOGG_qb with antisym=False see get_CCOOGG_qb() for more details

get_HH_q(eig)

Returns the Hamiltonian matrix in the Wannier gauge

Parameters:

eig (~wannierberri.w90files.EIG) – the eigenvalues of the Hamiltonian

Returns:

the Hamiltonian matrix in the Wannier gauge

Return type:

np.ndarray

get_OO_qb(mmn, uiu, phase=None, sum_b=False)

A wrapper for get_CCOOGG_qb with antisym=False see get_CCOOGG_qb() for more details (actually, the same as get_CC_qb())

get_SHA_q(shu, mmn, phase=None, sum_b=False)

SHA or SA (if siu is used instead of shu)

get_SHR_q(spn, mmn, eig=None, phase=None)

SHR or SR(if eig is None)

get_SS_q(spn)

Returns the spin matrix in the Wannier gauge

Parameters:

spn (~wannierberri.w90files.SPN) – the spin matrix

Returns:

the spin matrix in the Wannier gauge

Return type:

np.ndarray

get_wannier_centers(mmn, spreads=False)

calculate wannier centers with the Marzarri-Vanderbilt translational invariant formula and optionally the spreads

Parameters:

• mmn (MMN) – the overlap matrix elements between the Wavefunctions at neighbouring k-points

• spreads (bool) – if True, the spreads are calculated

Returns:

• np.ndarray(shape=(num_wann, 3), dtype=float) – the wannier centers

• np.ndarray(shape=(num_wann,), dtype=float) – the wannier spreads (in Angstrom^2) (if spreads=True)

property kpt_latt_int

Returns the k-points in the lattice basis

spin_order_block_to_interlace()

If the chk was obtain from a block ordering (like in old VASP versions), the ordering should be changed to interlace

spin_order_interlace_to_block()

If the chk was obtain from an interlace ordering, one may want to change the ordering to block

wannier_gauge(mat, ik1, ik2)

Returns the matrix elements in the Wannier gauge

Parameters:

• mat (np.ndarray) – the matrix elements in the Hamiltonian gauge

• ik1 (int) – the indices of the k-points

• ik2 (int) – the indices of the k-points

Returns:

the matrix elements in the Wannier gauge

Return type:

np.ndarray

class wannierberri.w90files.EIG(seedname='wannier90', **kwargs)

Bases: W90_file

from_w90_file(seedname)

abstract method to read the necessary data from Wannier90 file

select_bands(selected_bands)

abstract method to select the bands from the data

Select the bands to be used in the calculation, the rest are excluded

Parameters:

selected_bands (list of int) – the list of bands to be used in the calculation

class wannierberri.w90files.MMN(seedname='wannier90', npar=2, **kwargs)

Bases: W90_file

class to store overlaps between Bloch functions at neighbouring k-points the MMN file of wannier90

MMN.data[ik, ib, m, n] = <u_{m,k}|u_{n,k+b}>

Parameters:

• seedname (str) – the prefix of the file (including relative/absolute path, but not including the extension .mmn)

• npar (int) – the number of parallel processes to be used for reading the file

data

the overlap matrix elements between the Wavefunctions at neighbouring k-points

Type:

np.ndarray(shape=(NK, NNB, NB, NB), dtype=complex)

neighbours

the indices of the neighbouring k-points

Type:

np.ndarray(shape=(NK, NNB), dtype=int)

G

the reciprocal lattice vectors connecting the k-points

Type:

np.ndarray(shape=(NK, NNB, 3), dtype=int)

from_w90_file(seedname, npar)

abstract method to read the necessary data from Wannier90 file

property n_neighb

number of nearest neighbours indices

select_bands(selected_bands)

abstract method to select the bands from the data

Select the bands to be used in the calculation, the rest are excluded

Parameters:

selected_bands (list of int) – the list of bands to be used in the calculation

class wannierberri.w90files.SHU(seedname='wannier90', formatted=False, **kwargs)

Bases: SXU

SHU.data[ik, ib, m, n, ipol] = <u_{m,k}|S_ipol*H(k)|u_{n,k+b}>

Parameters:

• seedname (str) – the prefix of the file (including relative/absolute path, but not including the extensions, like .sHu, sIu)

• formatted (bool) – if True, the file is expected to be formatted, otherwise it is binary

• kwargs (dict(str, Any)) – the keyword arguments to be passed to the parent constructor, see ~wannierberri.system.w90_files.SXU for more details

class wannierberri.w90files.SIU(seedname='wannier90', formatted=False, **kwargs)

Bases: SXU

SIU.data[ik, ib, m, n, ipol] = <u_{m,k}|S_ipol|u_{n,k+b}>

Parameters:

• seedname (str) – the prefix of the file (including relative/absolute path, but not including the extensions, like .sIu, sHu)

• formatted (bool) – if True, the file is expected to be formatted, otherwise it is binary

• kwargs (dict(str, Any)) – the keyword arguments to be passed to the parent constructor, see ~wannierberri.system.w90_files.SXU for more details

class wannierberri.w90files.SPN(seedname='wannier90', **kwargs)

Bases: W90_file

SPN.data[ik, m, n, ipol] = <u_{m,k}|S_ipol|u_{n,k}>

from_w90_file(seedname='wannier90', formatted=False)

abstract method to read the necessary data from Wannier90 file

select_bands(selected_bands)

abstract method to select the bands from the data

Select the bands to be used in the calculation, the rest are excluded

Parameters:

selected_bands (list of int) – the list of bands to be used in the calculation

class wannierberri.w90files.UHU(seedname='wannier90', **kwargs)

Bases: UXU

UHU.data[ik, ib1, ib2, m, n] = <u_{m,k+b1}|H(k)|u_{n,k+b2}>

class wannierberri.w90files.UIU(seedname='wannier90', **kwargs)

Bases: UXU

UIU.data[ik, ib1, ib2, m, n] = <u_{m,k+b1}|u_{n,k+b2}>

class wannierberri.w90files.WIN(seedname='wannier90', data=None, autoread=False)

Bases: object

Class to read and store the wannier90.win input file

Parameters:

seedname (str) – the prefix of the file (including relative/absolute path, but not including the extensions, like .win)

name

the name of the file

Type:

str

parsed

the parsed data

Type:

dict(str, Any)

get_kpoints()

Get the kpoints in reciprocal coordinates

Returns:

the kpoints in reciprocal coordinates

Return type:

numpy.ndarray(float, shape=(NK, 3))

get_projections()

Get the projections

Returns:

the projections in the wannier90 format

Return type:

list(str)

get_unit_cell_cart_ang()

Get the unit cell in Angstrom in Cartesian coordinates

Returns:

the unit cell in Angstrom in Cartesian coordinates

Return type:

numpy.ndarray(float, shape=(3, 3))

update(dic)

Update the parameters with a dictionary

Parameters:

dic(dict) : the dictionary with the new parameters

write(seedname=None, comment='written by WannierBerri')

Write the wannier90.win file

Parameters:

• seedname (str) – the prefix of the file (including relative/absolute path, but not including the extensions, like .win) if None, the file is written to self.seedname + “.win”

• comment (str) – the comment to be written at the beginning of the file

class wannierberri.w90files.xxu.SXU(autoread=False, **kwargs)

Bases: W90_file

Read and setup sHu or sIu object. pw2wannier90 writes data_pw2w90[n, m, ipol, ib, ik] = <u_{m,k}|S_ipol * X|u_{n,k+b}> in column-major order. (X = H for SHU, X = I for SIU.) Here, we read to have data[ik, ib, m, n, ipol] = <u_{m,k}|S_ipol * X|u_{n,k+b}>.

from_w90_file(seedname='wannier90', formatted=False, suffix='sHu', **kwargs)

Read the sHu or sIu file

Parameters:

• seedname (str) – the prefix of the file (including relative/absolute path, but not including the extensions, like .sHu, sIu)

• formatted (bool) – if True, the file is expected to be formatted, otherwise it is binary

• suffix (str) – the suffix of the file, e.g. ‘sHu’, ‘sIu’

• kwargs (dict(str, Any)) – the keyword arguments to be passed to the constructor of the file see ~wannierberri.system.w90_files.SIU, ~wannierberri.system.w90_files.SHU for more details

Raises:

AssertionError – if the file does not conform with the other files

Sets

self.datanumpy.ndarray(complex, shape=(NK, NNB, NB, NB, 3)

the data of the file

property n_neighb

number of nearest neighbours indices

select_bands(selected_bands)

abstract method to select the bands from the data

Select the bands to be used in the calculation, the rest are excluded

Parameters:

selected_bands (list of int) – the list of bands to be used in the calculation

class wannierberri.w90files.xxu.UXU(autoread=False, **kwargs)

Bases: W90_file

Read and setup uHu or uIu object. pw2wannier90 writes data_pw2w90[n, m, ib1, ib2, ik] = <u_{m,k+b1}|X|u_{n,k+b2}> in column-major order. (X = H for UHU, X = I for UIU.) Here, we read to have data[ik, ib1, ib2, m, n] = <u_{m,k+b1}|X|u_{n,k+b2}>.

from_w90_file(seedname='wannier90', suffix='uXu', formatted=False)

abstract method to read the necessary data from Wannier90 file

property n_neighb

number of nearest neighbours indices

select_bands(selected_bands)

abstract method to select the bands from the data

Select the bands to be used in the calculation, the rest are excluded

Parameters:

selected_bands (list of int) – the list of bands to be used in the calculation