def calculate_dos(
self,
estep: float = defaults["dos_estep"],
progress_bar: bool = defaults["print_log"]
):
"""
Args:
estep: The DOS energy step in eV, where smaller numbers give more
accuracy but are more expensive.
progress_bar: Show a progress bar for DOS calculation.
"""
emin = np.min([np.min(spin_eners) for spin_eners in self.energies.values()])
emax = np.max([np.max(spin_eners) for spin_eners in self.energies.values()])
epoints = int(round((emax - emin) / (estep * units.eV)))
energies = np.linspace(emin, emax, epoints)
dos_weight = 1 if self._soc or len(self.spins) == 2 else 2
logger.debug("DOS parameters:")
log_list(
[
"emin: {:.2f} eV".format(emin / units.eV),
"emax: {:.2f} eV".format(emax / units.eV),
"dos weight: {}".format(dos_weight),
"n points: {}".format(epoints),
]
)
logger.debug("Generating tetrahedral DOS:")
t0 = time.perf_counter()
emesh, dos = self.tetrahedral_band_structure.get_density_of_states(
energies=energies, progress_bar=progress_bar
)
log_time_taken(t0)
num_electrons = self.num_electrons if self.is_metal else None
self.dos = FermiDos(
self.intrinsic_fermi_level,
emesh,
dos,
self.structure,
atomic_units=True,
dos_weight=dos_weight,
num_electrons=num_electrons,
)
def get_dos(
self,
kpoint_mesh: Union[float, int, List[int]],
energy_cutoff: Optional[float] = None,
scissor: Optional[float] = None,
bandgap: Optional[float] = None,
estep: float = defaults["dos_estep"],
symprec: float = defaults["symprec"],
atomic_units: bool = False,
) -> Union[Dos, FermiDos]:
"""Calculates the density of states using the interpolated bands.
Args:
kpoint_mesh: The k-point mesh as a 1x3 array. E.g.,``[6, 6, 6]``.
Alternatively, if a single value is provided this will be
treated as a reciprocal density and the k-point mesh dimensions
generated automatically.
energy_cutoff: The energy cut-off to determine which bands are
included in the interpolation. If the energy of a band falls
within the cut-off at any k-point it will be included. For
metals the range is defined as the Fermi level ± energy_cutoff.
For gapped materials, the energy range is from the VBM -
energy_cutoff to the CBM + energy_cutoff.
scissor: The amount by which the band gap is scissored. Cannot
be used in conjunction with the ``bandgap`` option. Has no
effect for metallic systems.
bandgap: Automatically adjust the band gap to this value. Cannot
be used in conjunction with the ``scissor`` option. Has no
effect for metallic systems.
estep: The energy step, where smaller numbers give more
accuracy but are more expensive.
symprec: The symmetry tolerance used when determining the symmetry
inequivalent k-points on which to interpolate.
atomic_units: Whether to return the DOS in atomic units. If False, the
unit of energy will be eV.
Returns:
The density of states.
"""
if isinstance(kpoint_mesh, numeric_types):
logger.info("DOS k-point length cutoff: {}".format(kpoint_mesh))
else:
str_mesh = "x".join(map(str, kpoint_mesh))
logger.info("DOS k-point mesh: {}".format(str_mesh))
structure = self._band_structure.structure
tri = not self._soc
(
ir_kpts,
_,
full_kpts,
ir_kpts_idx,
ir_to_full_idx,
tetrahedra,
*ir_tetrahedra_info,
) = get_kpoints_tetrahedral(kpoint_mesh,
structure,
symprec=symprec,
time_reversal_symmetry=tri)
energies, efermi, vb_idx = self.get_energies(
ir_kpts,
scissor=scissor,
bandgap=bandgap,
energy_cutoff=energy_cutoff,
atomic_units=atomic_units,
return_efermi=True,
return_vb_idx=True,
)
if not self._band_structure.is_metal():
# if not a metal, set the Fermi level to the top of the valence band.
efermi = np.max(
[np.max(e[vb_idx[spin]]) for spin, e in energies.items()])
full_energies = {s: e[:, ir_to_full_idx] for s, e in energies.items()}
tetrahedral_band_structure = TetrahedralBandStructure(
full_energies, full_kpts, tetrahedra, structure, ir_kpts_idx,
ir_to_full_idx, *ir_tetrahedra_info)
emin = np.min([np.min(spin_eners) for spin_eners in energies.values()])
emax = np.max([np.max(spin_eners) for spin_eners in energies.values()])
epoints = int(round((emax - emin) / estep))
energies = np.linspace(emin, emax, epoints)
_, dos = tetrahedral_band_structure.get_density_of_states(energies)
return FermiDos(efermi,
energies,
dos,
structure,
atomic_units=atomic_units)
class AmsetData(MSONable):
def __init__(
self,
structure: Structure,
energies: Dict[Spin, np.ndarray],
vvelocities_product: Dict[Spin, np.ndarray],
velocities: Dict[Spin, np.ndarray],
kpoint_mesh: np.ndarray,
kpoints: np.ndarray,
ir_kpoints_idx: np.ndarray,
ir_to_full_kpoint_mapping: np.ndarray,
tetrahedra: np.ndarray,
ir_tetrahedra_info: np.ndarray,
efermi: float,
num_electrons: float,
is_metal: bool,
soc: bool,
vb_idx: Optional[Dict[Spin, int]] = None,
):
self.structure = structure
self.velocities_product = vvelocities_product
self.kpoint_mesh = kpoint_mesh
self.intrinsic_fermi_level = efermi
self.ir_to_full_kpoint_mapping = ir_to_full_kpoint_mapping
self._soc = soc
self.num_electrons = num_electrons
self.is_metal = is_metal
self.vb_idx = vb_idx
self.spins = list(energies.keys())
self.velocities = {
s: v.transpose((0, 2, 1))
for s, v in velocities.items()
}
self.dos = None
self.scattering_rates = None
self.scattering_labels = None
self.doping = None
self.temperatures = None
self.fermi_levels = None
self.electron_conc = None
self.hole_conc = None
self.conductivity = None
self.seebeck = None
self.electronic_thermal_conductivity = None
self.mobility = None
self.overlap_calculator = None
self.mrta_calculator = None
self.fd_cutoffs = None
self.grouped_ir_to_full = groupby(np.arange(len(kpoints)),
ir_to_full_kpoint_mapping)
self.tetrahedral_band_structure = TetrahedralBandStructure.from_data(
energies, kpoints, tetrahedra, structure, ir_kpoints_idx,
ir_to_full_kpoint_mapping, *ir_tetrahedra_info)
logger.info("Initializing momentum relaxation time factor calculator")
self.mrta_calculator = MRTACalculator.from_data(
kpoints, self.velocities)
@property
def energies(self):
return self.tetrahedral_band_structure.energies
@property
def kpoints(self):
return self.tetrahedral_band_structure.kpoints
@property
def ir_kpoints(self):
return self.kpoints[self.tetrahedral_band_structure.ir_kpoints_idx]
@property
def ir_kpoints_idx(self):
return self.tetrahedral_band_structure.ir_kpoints_idx
def set_overlap_calculator(self, overlap_calculator):
if overlap_calculator is not None:
equal = check_nbands_equal(overlap_calculator, self)
if not equal:
raise RuntimeError(
"Overlap calculator does not have the correct number of bands\n"
"If using wavefunction coefficients, ensure they were generated "
"using the same energy_cutoff (not encut)")
self.overlap_calculator = overlap_calculator
def calculate_dos(
self,
estep: float = defaults["dos_estep"],
progress_bar: bool = defaults["print_log"],
):
"""
Args:
estep: The DOS energy step in eV, where smaller numbers give more
accuracy but are more expensive.
progress_bar: Show a progress bar for DOS calculation.
"""
emin = np.min(
[np.min(spin_eners) for spin_eners in self.energies.values()])
emax = np.max(
[np.max(spin_eners) for spin_eners in self.energies.values()])
epoints = int(round((emax - emin) / (estep * ev_to_hartree)))
energies = np.linspace(emin, emax, epoints)
dos_weight = 1 if self._soc or len(self.spins) == 2 else 2
logger.info("DOS parameters:")
log_list([
"emin: {:.2f} eV".format(emin * hartree_to_ev),
"emax: {:.2f} eV".format(emax * hartree_to_ev),
"dos weight: {}".format(dos_weight),
"n points: {}".format(epoints),
])
logger.info("Generating tetrahedral DOS:")
t0 = time.perf_counter()
emesh, dos = self.tetrahedral_band_structure.get_density_of_states(
energies=energies, progress_bar=progress_bar)
log_time_taken(t0)
num_electrons = self.num_electrons if self.is_metal else None
self.dos = FermiDos(
self.intrinsic_fermi_level,
emesh,
dos,
self.structure,
atomic_units=True,
dos_weight=dos_weight,
num_electrons=num_electrons,
)
def set_doping_and_temperatures(self, doping: np.ndarray,
temperatures: np.ndarray):
if not self.dos:
raise RuntimeError(
"The DOS should be calculated (AmsetData.calculate_dos) before "
"setting doping levels.")
if doping is None:
# Generally this is for metallic systems; here we use the intrinsic Fermi
# level
self.doping = [0]
print("doping is none")
else:
self.doping = doping * (1 / cm_to_bohr)**3
self.temperatures = temperatures
self.fermi_levels = np.zeros((len(doping), len(temperatures)))
self.electron_conc = np.zeros((len(doping), len(temperatures)))
self.hole_conc = np.zeros((len(doping), len(temperatures)))
fermi_level_info = []
tols = np.logspace(-5, 0, 6)
for n, t in np.ndindex(self.fermi_levels.shape):
for i, tol in enumerate(tols):
# Finding the Fermi level is quite fickle. Enumerate multiple
# tolerances and use the first one that works!
try:
if self.doping[n] == 0:
self.fermi_levels[
n, t] = self.dos.get_fermi_from_num_electrons(
self.num_electrons,
temperatures[t],
tol=tol / 1000,
precision=10,
)
else:
(
self.fermi_levels[n, t],
self.electron_conc[n, t],
self.hole_conc[n, t],
) = self.dos.get_fermi(
self.doping[n],
temperatures[t],
tol=tol,
precision=10,
return_electron_hole_conc=True,
)
break
except ValueError:
if i == len(tols) - 1:
raise ValueError(
"Could not calculate Fermi level position."
"Try a denser k-point mesh.")
else:
pass
fermi_level_info.append((doping[n], temperatures[t],
self.fermi_levels[n, t] * hartree_to_ev))
table = tabulate(
fermi_level_info,
headers=("conc [cm⁻³]", "temp [K]", "E_fermi [eV]"),
numalign="right",
stralign="center",
floatfmt=(".2e", ".1f", ".4f"),
)
logger.info("Calculated Fermi levels:")
logger.info(table)
def calculate_fd_cutoffs(
self,
fd_tolerance: Optional[float] = 0.01,
cutoff_pad: float = 0.0,
max_moment: int = 2,
mobility_rates_only: bool = False,
):
energies = self.dos.energies
vv = {
s: v.transpose((0, 3, 1, 2))
for s, v in self.velocities_product.items()
}
_, vvdos = self.tetrahedral_band_structure.get_density_of_states(
energies, integrand=vv, sum_spins=True, use_cached_weights=True)
vvdos = tensor_average(vvdos)
# vvdos = np.array(self.dos.get_densities())
# three fermi integrals govern transport properties:
# 1. df/de controls conductivity and mobility
# 2. (e-u) * df/de controls Seebeck
# 3. (e-u)^2 df/de controls electronic thermal conductivity
# take the absolute sum of the integrals across all doping and
# temperatures. this gives us the energies that are important for
# transport
if fd_tolerance:
def get_min_max_cutoff(cumsum):
min_idx = np.where(cumsum < fd_tolerance / 2)[0].max()
max_idx = np.where(cumsum > (1 - fd_tolerance / 2))[0].min()
return energies[min_idx], energies[max_idx]
min_cutoff = np.inf
max_cutoff = -np.inf
for n, t in np.ndindex(self.fermi_levels.shape):
ef = self.fermi_levels[n, t]
temp = self.temperatures[t]
dfde = -dfdde(energies, ef, temp * boltzmann_au)
for moment in range(max_moment + 1):
weight = np.abs((energies - ef)**moment * dfde)
weight_dos = weight * vvdos
weight_cumsum = np.cumsum(weight_dos)
weight_cumsum /= np.max(weight_cumsum)
cmin, cmax = get_min_max_cutoff(weight_cumsum)
min_cutoff = min(cmin, min_cutoff)
max_cutoff = max(cmax, max_cutoff)
# import matplotlib.pyplot as plt
# ax = plt.gca()
# plt.plot(energies / units.eV, weight / weight.max())
# plt.plot(energies / units.eV, vvdos / vvdos.max())
# plt.plot(energies / units.eV, weight_dos / weight_dos.max())
# plt.plot(energies / units.eV, weight_cumsum / weight_cumsum.max())
# ax.set(xlim=(4, 7.5))
# plt.show()
else:
min_cutoff = energies.min()
max_cutoff = energies.max()
if mobility_rates_only:
vbm = get_vbm_energy(self.energies, self.vb_idx)
cbm = get_cbm_energy(self.energies, self.vb_idx)
mid_gap = (cbm + vbm) / 2
if np.all(self.doping < 0):
# only electron mobility so don't calculate valence band rates
min_cutoff = max(min_cutoff, mid_gap)
elif np.all(self.doping < 0):
# only hole mobility so don't calculate conudction band rates
max_cutoff = min(max_cutoff, mid_gap)
min_cutoff -= cutoff_pad
max_cutoff += cutoff_pad
logger.info("Calculated Fermi–Dirac cut-offs:")
log_list([
"min: {:.3f} eV".format(min_cutoff * hartree_to_ev),
"max: {:.3f} eV".format(max_cutoff * hartree_to_ev),
])
self.fd_cutoffs = (min_cutoff, max_cutoff)
def set_scattering_rates(self, scattering_rates: Dict[Spin, np.ndarray],
scattering_labels: List[str]):
for spin in self.spins:
s = (len(self.doping), len(
self.temperatures)) + self.energies[spin].shape
if scattering_rates[spin].shape[1:] != s:
raise ValueError(
"Shape of scattering_type rates array does not match the "
"number of dopings, temperatures, bands, or kpoints")
if scattering_rates[spin].shape[0] != len(scattering_labels):
raise ValueError(
"Number of scattering_type rates does not match number of "
"scattering_type labels")
self.scattering_rates = scattering_rates
self.scattering_labels = scattering_labels
def fill_rates_outside_cutoffs(self, fill_value=None):
if self.scattering_rates is None:
raise ValueError(
"Scattering rates must be set before being filled")
min_fd, max_fd = self.fd_cutoffs
snt_fill = fill_value
for spin, spin_energies in self.energies.items():
mask = (spin_energies < min_fd) | (spin_energies > max_fd)
any_in_mask = np.any(
~mask) # sometimes no rates calculated for spin
rate_info = defaultdict(list)
for s, n, t in np.ndindex(self.scattering_rates[spin].shape[:3]):
if fill_value is None and any_in_mask:
# get average log rate inside cutoffs
snt_fill = np.log(self.scattering_rates[spin][s, n, t,
~mask])
snt_fill = np.exp(snt_fill.mean())
elif fill_value is None:
# no rates have been calculated for this spin channel so cannot
# average them. In this rare case we use a rate of 10^14
snt_fill = 1e14
rate_info[self.scattering_labels[s]].append(snt_fill)
self.scattering_rates[spin][s, n, t, mask] = snt_fill
if len(self.spins) == 1:
logger.info(
"Filling scattering rates [s⁻¹] outside FD cutoffs with:")
else:
logger.info(
"Filling {} scattering rates [s⁻¹] outside FD cutoffs "
"with:".format(spin_name[spin]))
headers = ["conc [cm⁻³]", "temp [K]"]
headers += ["{}".format(s) for s in self.scattering_labels]
rate_table = []
for i, (n, t) in enumerate(np.ndindex(self.fermi_levels.shape)):
col = [
self.doping[n] * (1 / bohr_to_cm)**3, self.temperatures[t]
]
col += [rate_info[s][i] for s in self.scattering_labels]
rate_table.append(col)
table = tabulate(
rate_table,
headers=headers,
numalign="right",
stralign="center",
floatfmt=[".2e", ".1f"] +
[".2e"] * len(self.scattering_labels),
)
logger.info(table)
def set_transport_properties(
self,
conductivity: np.ndarray,
seebeck: np.ndarray,
electronic_thermal_conductivity: np.ndarray,
mobility: Optional[np.ndarray] = None,
):
self.conductivity = conductivity
self.seebeck = seebeck
self.electronic_thermal_conductivity = electronic_thermal_conductivity
self.mobility = mobility
def to_dict(self, include_mesh=defaults["write_mesh"]):
data = {
"doping": (self.doping * cm_to_bohr**3).round(),
"temperatures": self.temperatures,
"fermi_levels": self.fermi_levels * hartree_to_ev,
"conductivity": self.conductivity,
"seebeck": self.seebeck,
"electronic_thermal_conductivity":
self.electronic_thermal_conductivity,
"mobility": self.mobility,
}
if include_mesh:
rates = self.scattering_rates
energies = self.energies
vel = self.velocities
ir_rates = {
s: r[..., self.ir_kpoints_idx]
for s, r in rates.items()
}
ir_energies = {
s: e[:, self.ir_kpoints_idx] * hartree_to_ev
for s, e in energies.items()
}
ir_vel = {s: v[:, self.ir_kpoints_idx] for s, v in vel.items()}
mesh_data = {
"energies":
ir_energies,
"kpoints":
self.kpoints,
"ir_kpoints":
self.ir_kpoints,
"ir_to_full_kpoint_mapping":
self.ir_to_full_kpoint_mapping,
"efermi":
self.intrinsic_fermi_level * hartree_to_ev,
"vb_idx":
self.vb_idx,
"num_electrons":
self.num_electrons,
# "dos": self.dos, # TODO: Convert dos to eV
"velocities":
ir_vel, # TODO: convert units
"scattering_rates":
ir_rates,
"scattering_labels":
self.scattering_labels,
"is_metal":
self.is_metal,
"fd_cutoffs": (
self.fd_cutoffs[0] * hartree_to_ev,
self.fd_cutoffs[1] * hartree_to_ev,
),
"structure":
get_angstrom_structure(self.structure),
"soc":
self._soc,
"doping":
data["doping"],
"temperatures":
data["temperatures"],
"fermi_levels":
data["fermi_levels"],
}
data["mesh"] = mesh_data
return data
def to_data(self):
data = []
triu = np.triu_indices(3)
for n, t in np.ndindex(len(self.doping), len(self.temperatures)):
row = [
self.doping[n] * cm_to_bohr**3,
self.temperatures[t],
self.fermi_levels[n, t] * hartree_to_ev,
]
row.extend(self.conductivity[n, t][triu])
row.extend(self.seebeck[n, t][triu])
row.extend(self.electronic_thermal_conductivity[n, t][triu])
if self.mobility is not None:
for mob in self.mobility.values():
row.extend(mob[n, t][triu])
data.append(row)
headers = ["doping[cm^-3]", "temperature[K]", "Fermi_level[eV]"]
ds = ("xx", "xy", "xz", "yy", "yz", "zz")
# TODO: confirm unit of kappa
for prop, unit in [("cond", "S/m"), ("seebeck", "µV/K"),
("kappa", "?")]:
headers.extend(["{}_{}[{}]".format(prop, d, unit) for d in ds])
if self.mobility is not None:
for name in self.mobility.keys():
headers.extend(
["{}_mobility_{}[cm^2/V.s]".format(name, d) for d in ds])
return data, headers
def to_file(
self,
directory: str = ".",
prefix: Optional[str] = None,
write_mesh_file: bool = defaults["write_mesh"],
file_format: str = defaults["file_format"],
suffix_mesh: bool = True,
):
if self.conductivity is None:
raise ValueError(
"Can't write AmsetData, transport properties not set")
if not prefix:
prefix = ""
else:
prefix += "_"
if suffix_mesh:
suffix = "_{}".format("x".join(map(str, self.kpoint_mesh)))
else:
suffix = ""
if file_format in ["json", "yaml"]:
data = self.to_dict()
data = cast_dict_list(data)
filename = joinpath(
directory, "{}transport{}.{}".format(prefix, suffix,
file_format))
dumpfn(data, filename, indent=4)
elif file_format in ["csv", "txt"]:
# don't write the data as JSON, instead write raw text files
data, headers = self.to_data()
filename = joinpath(
directory, "{}transport{}.{}".format(prefix, suffix,
file_format))
np.savetxt(filename, data, header=" ".join(headers))
else:
raise ValueError(
"Unrecognised output format: {}".format(file_format))
if write_mesh_file:
mesh_data = self.to_dict(include_mesh=True)["mesh"]
mesh_filename = joinpath(directory,
"{}mesh{}.h5".format(prefix, suffix))
write_mesh(mesh_data, filename=mesh_filename)
return filename, mesh_filename
else:
return filename
class AmsetData(MSONable):
def __init__(
self,
structure: Structure,
energies: Dict[Spin, np.ndarray],
vvelocities_product: Dict[Spin, np.ndarray],
velocities: Dict[Spin, np.ndarray],
kpoint_mesh: np.ndarray,
kpoints: np.ndarray,
ir_kpoints: np.ndarray,
ir_kpoints_idx: np.ndarray,
ir_to_full_kpoint_mapping: np.ndarray,
tetrahedra: np.ndarray,
ir_tetrahedra_info: np.ndarray,
efermi: float,
num_electrons: float,
is_metal: bool,
soc: bool,
vb_idx: Optional[Dict[Spin, int]] = None,
):
self.structure = structure
self.energies = energies
self.velocities_product = vvelocities_product
self.kpoint_mesh = kpoint_mesh
self.kpoints = kpoints
self.ir_kpoints = ir_kpoints
self.ir_kpoints_idx = ir_kpoints_idx
self.ir_to_full_kpoint_mapping = ir_to_full_kpoint_mapping
self.intrinsic_fermi_level = efermi
self._soc = soc
self.num_electrons = num_electrons
self.is_metal = is_metal
self.vb_idx = vb_idx
self.spins = self.energies.keys()
self.dos = None
self.scattering_rates = None
self.scattering_labels = None
self.doping = None
self.temperatures = None
self.fermi_levels = None
self.electron_conc = None
self.hole_conc = None
self.conductivity = None
self.seebeck = None
self.electronic_thermal_conductivity = None
self.mobility = None
self.overlap_calculator = None
self.mrta_calculator = None
self.fd_cutoffs = None
self.velocities = {s: v.transpose((0, 2, 1)) for s, v in velocities.items()}
self.grouped_ir_to_full = groupby(
np.arange(len(kpoints)), ir_to_full_kpoint_mapping
)
self.tetrahedral_band_structure = TetrahedralBandStructure(
energies,
kpoints,
tetrahedra,
structure,
ir_kpoints_idx,
ir_to_full_kpoint_mapping,
*ir_tetrahedra_info
)
self.mrta_calculator = MRTACalculator(
self.kpoints, self.kpoint_mesh, self.velocities
)
def set_overlap_calculator(self, overlap_calculator):
nbands_equal = [
self.energies[s].shape[0] == overlap_calculator.nbands[s]
for s in self.spins
]
if not all(nbands_equal):
raise RuntimeError(
"Overlap calculator does not have the correct number of bands\n"
"If using wavefunction coefficients, ensure they were generated using"
"the same energy_cutoff (not encut)"
)
self.overlap_calculator = overlap_calculator
def calculate_dos(self, estep: float = defaults["dos_estep"]):
"""
Args:
estep: The DOS energy step in eV, where smaller numbers give more
accuracy but are more expensive.
"""
emin = np.min([np.min(spin_eners) for spin_eners in self.energies.values()])
emax = np.max([np.max(spin_eners) for spin_eners in self.energies.values()])
epoints = int(round((emax - emin) / (estep * units.eV)))
energies = np.linspace(emin, emax, epoints)
dos_weight = 1 if self._soc or len(self.spins) == 2 else 2
logger.debug("DOS parameters:")
log_list(
[
"emin: {:.2f} eV".format(emin / units.eV),
"emax: {:.2f} eV".format(emax / units.eV),
"dos weight: {}".format(dos_weight),
"n points: {}".format(epoints),
]
)
logger.debug("Generating tetrahedral DOS:")
t0 = time.perf_counter()
emesh, dos = self.tetrahedral_band_structure.get_density_of_states(
energies=energies, progress_bar=True
)
log_time_taken(t0)
num_electrons = self.num_electrons if self.is_metal else None
self.dos = FermiDos(
self.intrinsic_fermi_level,
emesh,
dos,
self.structure,
atomic_units=True,
dos_weight=dos_weight,
num_electrons=num_electrons,
)
def set_doping_and_temperatures(self, doping: np.ndarray, temperatures: np.ndarray):
if not self.dos:
raise RuntimeError(
"The DOS should be calculated (AmsetData.calculate_dos) before "
"setting doping levels."
)
if doping is None:
# Generally this is for metallic systems; here we use the intrinsic Fermi
# level
self.doping = [0]
print("doping is none")
else:
self.doping = doping * (1 / cm_to_bohr) ** 3
self.temperatures = temperatures
self.fermi_levels = np.zeros((len(doping), len(temperatures)))
self.electron_conc = np.zeros((len(doping), len(temperatures)))
self.hole_conc = np.zeros((len(doping), len(temperatures)))
fermi_level_info = []
tols = np.logspace(-5, 0, 6)
for n, t in np.ndindex(self.fermi_levels.shape):
for i, tol in enumerate(tols):
# Finding the Fermi level is quite fickle. Enumerate multiple
# tolerances and use the first one that works!
try:
if self.doping[n] == 0:
self.fermi_levels[n, t] = self.dos.get_fermi_from_num_electrons(
self.num_electrons,
temperatures[t],
tol=tol / 1000,
precision=10,
)
else:
(
self.fermi_levels[n, t],
self.electron_conc[n, t],
self.hole_conc[n, t],
) = self.dos.get_fermi(
self.doping[n],
temperatures[t],
tol=tol,
precision=10,
return_electron_hole_conc=True,
)
break
except ValueError:
if i == len(tols) - 1:
raise ValueError(
"Could not calculate Fermi level position."
"Try a denser k-point mesh."
)
else:
pass
fermi_level_info.append(
"{:.2g} cm⁻³ & {} K: {:.4f} eV".format(
doping[n], temperatures[t], self.fermi_levels[n, t] / units.eV
)
)
logger.info("Calculated Fermi levels:")
log_list(fermi_level_info)
def calculate_fd_cutoffs(
self, fd_tolerance: Optional[float] = 0.01, cutoff_pad: float = 0.0
):
energies = self.dos.energies
# three fermi integrals govern transport properties:
# 1. df/de controls conductivity and mobility
# 2. (e-u) * df/de controls Seebeck
# 3. (e-u)^2 df/de controls electronic thermal conductivity
# take the absolute sum of the integrals across all doping and
# temperatures. this gives us the energies that are important for
# transport
weights = np.zeros(energies.shape)
for n, t in np.ndindex(self.fermi_levels.shape):
ef = self.fermi_levels[n, t]
temp = self.temperatures[t]
dfde = -dFDde(energies, ef, temp * units.BOLTZMANN)
sigma_int = np.abs(dfde)
seeb_int = np.abs((energies - ef) * dfde)
ke_int = np.abs((energies - ef) ** 2 * dfde)
# normalize the transport integrals and sum
nt_weights = sigma_int / sigma_int.max()
nt_weights += seeb_int / seeb_int.max()
nt_weights += ke_int / ke_int.max()
weights = np.maximum(weights, nt_weights)
if not self.is_metal:
# weights should be zero in the band gap as there will be no density
vb_bands = [
np.max(self.energies[s][: self.vb_idx[s] + 1]) for s in self.spins
]
cb_bands = [
np.min(self.energies[s][self.vb_idx[s] + 1 :]) for s in self.spins
]
vbm_e = np.max(vb_bands)
cbm_e = np.min(cb_bands)
weights[(energies > vbm_e) & (energies < cbm_e)] = 0
weights /= np.max(weights)
cumsum = np.cumsum(weights)
cumsum /= np.max(cumsum)
if fd_tolerance:
min_cutoff = energies[cumsum < fd_tolerance / 2].max()
max_cutoff = energies[cumsum > 1 - fd_tolerance / 2].min()
else:
min_cutoff = energies.min()
max_cutoff = energies.max()
min_cutoff -= cutoff_pad
max_cutoff += cutoff_pad
logger.info("Calculated Fermi–Dirac cut-offs:")
log_list(
[
"min: {:.3f} eV".format(min_cutoff / units.eV),
"max: {:.3f} eV".format(max_cutoff / units.eV),
]
)
self.fd_cutoffs = (min_cutoff, max_cutoff)
def set_scattering_rates(
self, scattering_rates: Dict[Spin, np.ndarray], scattering_labels: List[str]
):
for spin in self.spins:
s = (len(self.doping), len(self.temperatures)) + self.energies[spin].shape
if scattering_rates[spin].shape[1:] != s:
raise ValueError(
"Shape of scattering_type rates array does not match the "
"number of dopings, temperatures, bands, or kpoints"
)
if scattering_rates[spin].shape[0] != len(scattering_labels):
raise ValueError(
"Number of scattering_type rates does not match number of "
"scattering_type labels"
)
self.scattering_rates = scattering_rates
self.scattering_labels = scattering_labels
def set_transport_properties(
self,
conductivity: np.ndarray,
seebeck: np.ndarray,
electronic_thermal_conductivity: np.ndarray,
mobility: Optional[np.ndarray] = None,
):
self.conductivity = conductivity
self.seebeck = seebeck
self.electronic_thermal_conductivity = electronic_thermal_conductivity
self.mobility = mobility
def to_dict(self, include_mesh=defaults["write_mesh"]):
data = {
"doping": self.doping,
"temperatures": self.temperatures,
"fermi_levels": self.fermi_levels,
"conductivity": self.conductivity,
"seebeck": self.seebeck,
"electronic_thermal_conductivity": self.electronic_thermal_conductivity,
"mobility": self.mobility,
}
if include_mesh:
rates = self.scattering_rates
energies = self.energies
vv = self.velocities_product
ir_rates = {s: r[..., self.ir_kpoints_idx] for s, r in rates.items()}
ir_energies = {s: e[:, self.ir_kpoints_idx] for s, e in energies.items()}
ir_vv = {s: v[..., self.ir_kpoints_idx] for s, v in vv.items()}
mesh_data = {
"energies": cast_dict_list(ir_energies),
"kpoints": self.kpoints,
"ir_kpoints": self.ir_kpoints,
"ir_to_full_kpoint_mapping": self.ir_to_full_kpoint_mapping,
"efermi": self.intrinsic_fermi_level,
"vb_idx": cast_dict_list(self.vb_idx),
"dos": self.dos,
"velocities_product": cast_dict_list(ir_vv),
"scattering_rates": cast_dict_list(ir_rates),
"scattering_labels": self.scattering_labels,
"is_metal": self.is_metal,
"fd_cutoffs": self.fd_cutoffs,
"structure": self.structure,
"soc": self._soc,
}
data.update(mesh_data)
return data
def to_data(self):
data = []
triu = np.triu_indices(3)
for n, t in np.ndindex(len(self.doping), len(self.temperatures)):
row = [self.doping[n], self.temperatures[t], self.fermi_levels[n, t]]
row.extend(self.conductivity[n, t][triu])
row.extend(self.seebeck[n, t][triu])
row.extend(self.electronic_thermal_conductivity[n, t][triu])
if self.mobility is not None:
for mob in self.mobility.values():
row.extend(mob[n, t][triu])
data.append(row)
headers = ["doping[cm^-3]", "temperature[K]", "Fermi_level[eV]"]
ds = ("xx", "xy", "xz", "yy", "yz", "zz")
# TODO: confirm unit of kappa
for prop, unit in [("cond", "S/m"), ("seebeck", "µV/K"), ("kappa", "?")]:
headers.extend(["{}_{}[{}]".format(prop, d, unit) for d in ds])
if self.mobility is not None:
for name in self.mobility.keys():
headers.extend(["{}_mobility_{}[cm^2/V.s]".format(name, d) for d in ds])
return data, headers
def to_file(
self,
directory: str = ".",
prefix: Optional[str] = None,
write_mesh: bool = defaults["write_mesh"],
file_format: str = defaults["file_format"],
suffix_mesh: bool = True,
):
if self.conductivity is None:
raise ValueError("Can't write AmsetData, transport properties not set")
if not prefix:
prefix = ""
else:
prefix += "_"
if suffix_mesh:
suffix = "_{}".format("x".join(map(str, self.kpoint_mesh)))
else:
suffix = ""
if file_format in ["json", "yaml"]:
data = self.to_dict(include_mesh=write_mesh)
filename = joinpath(
directory, "{}amset_data{}.{}".format(prefix, suffix, file_format)
)
dumpfn(data, filename)
elif file_format in ["csv", "txt"]:
# don't write the data as JSON, instead write raw text files
data, headers = self.to_data()
filename = joinpath(
directory, "{}amset_transport{}.{}".format(prefix, suffix, file_format)
)
np.savetxt(filename, data, header=" ".join(headers))
if write_mesh:
logger.warning("Writing mesh data as txt or csv not supported")
else:
raise ValueError("Unrecognised output format: {}".format(file_format))
return filename