PAOFLOW.boltzmann.do_Boltz_tensors#

Attributes#

Functions#

do_Boltz_tensors(data_controller, smearing, temp, ene, ...)

Compute the Boltzmann transport tensors L0, L1, and L2.

do_Boltz_tensors_hall(data_controller, smearing, temp, ...)

Compute the anomalous (Hall) Boltzmann transport tensors L0_hall and L1_hall.

get_tau(data_controller, temp, channels, weights)

Compute the relaxation time τ(n, k) from a list of scattering channels.

L_loop(data_controller, temp, smearing, ene, velkp, ...)

Inner BZ summation loop for one Lu1d45 transport tensor.

L_loop_hall(data_controller, temp, smearing, ene, ...)

Inner BZ summation loop for the Hall transport tensor.

Module Contents#

PAOFLOW.boltzmann.do_Boltz_tensors.comm[source]#
PAOFLOW.boltzmann.do_Boltz_tensors.rank[source]#
PAOFLOW.boltzmann.do_Boltz_tensors.do_Boltz_tensors(data_controller, smearing, temp, ene, velkp, ispin, channels, weights)[source]#

Compute the Boltzmann transport tensors L0, L1, and L2.

The three generalized transport tensors are defined by the BZ integral

\[\begin{split}L^{(\\alpha)}_{ij}(\\varepsilon) = \\sum_{n\\mathbf{k}} \\tau_{n\\mathbf{k}}\\, v^i_{n\\mathbf{k}}\\, v^j_{n\\mathbf{k}}\\, \\left(-\\frac{\\partial f}{\\partial\\varepsilon}\\right) (E_{n\\mathbf{k}} - \\varepsilon)^\\alpha\end{split}\]

for alpha = 0, 1, 2. From these, the electrical conductivity (u03c3 ~ L0), Seebeck coefficient (S ~ L1/L0), and electronic thermal conductivity (u03ba ~ L2 - L1u00b2/L0) are derived.

When smearing is None the Fermi-Dirac derivative 1/(4T cosh\u00b2(...)) is evaluated analytically. When adaptive smearing is enabled ('gauss' or 'm-p') the integration is performed on an extended energy grid and convolved with the Fermi window via Simpson quadrature.

Only the upper-triangular components of each tensor are computed; the lower-triangular entries are filled by symmetry before returning.

Parameters:
  • data_controller (DataController) – Provides E_k, deltakp, scattering_tau, nkpnts, bnd.

  • smearing (str or None) – Smearing type: None (Fermi-Dirac), 'gauss', or 'm-p'.

  • temp (float) – Temperature in eV (ku0432T).

  • ene (ndarray, shape (ne,)) – Chemical-potential (energy) grid in eV.

  • velkp (ndarray, shape (nkpts_local, 3, bnd, nspin)) – Band velocities (diagonal elements of the momentum matrix).

  • ispin (int) – Spin channel index.

  • channels (list) – Scattering channel identifiers (strings or TauModel objects).

  • weights (list of float) – Harmonic-sum weights for each scattering channel.

Returns:

L0, L1, L2 – Transport tensors on rank 0; (None, None, None) on all other ranks.

Return type:

ndarray, shape (3, 3, ne), or (None, None, None)

PAOFLOW.boltzmann.do_Boltz_tensors.do_Boltz_tensors_hall(data_controller, smearing, temp, ene, velkp, ispin, channels, weights)[source]#

Compute the anomalous (Hall) Boltzmann transport tensors L0_hall and L1_hall.

Evaluates the rank-3 Hall conductivity kernel

\[\begin{split}L^{\\rm Hall}_{ijp}(\\varepsilon) = \\sum_{n\\mathbf{k}} \\tau^2_{n\\mathbf{k}} \\sum_{qr} \\epsilon_{pqr}\\, v^i_{n\\mathbf{k}}\\, v^r_{n\\mathbf{k}}\\, M^{-1}_{jq,n\\mathbf{k}}\\, \\left(-\\frac{\\partial f}{\\partial\\varepsilon}\\right)\end{split}\]

where \(\\epsilon_{pqr}\) is the Levi-Civita symbol and \(M^{-1}_{jq}\) is the inverse effective-mass tensor from arry['d2Ed2k'].

L0_hall is the zeroth energy moment of this kernel and L1_hall its first moment, weighted by \((E_{n\\mathbf{k}} - \\varepsilon)\).

When smearing is None the Fermi-Dirac derivative 1/(4T cosh\u00b2(...)) is evaluated analytically and each moment is obtained from a separate L_loop_hall() call, selected through its alpha argument. When adaptive smearing is enabled ('gauss' or 'm-p') the kernel is evaluated once on an extended energy grid and both moments are obtained by convolution with the Fermi window via Simpson quadrature.

Parameters:
  • data_controller (DataController) – Provides E_k, deltakp, d2Ed2k, scattering_tau, nkpnts, bnd, nspin.

  • smearing (str or None) – Smearing type: None (Fermi-Dirac), 'gauss', or 'm-p'.

  • temp (float) – Temperature in eV (ku0432T).

  • ene (ndarray, shape (ne,)) – Chemical-potential grid in eV.

  • velkp (ndarray, shape (nkpts_local, 3, bnd, nspin)) – Band velocities.

  • ispin (int) – Spin channel index.

  • channels (list) – Scattering channel identifiers.

  • weights (list of float) – Harmonic-sum weights for each scattering channel.

Returns:

L0_hall, L1_hall – Hall transport tensor on rank 0; (None, None) on all other ranks.

Return type:

ndarray, shape (3, 3, 3, ne), or (None, None)

PAOFLOW.boltzmann.do_Boltz_tensors.get_tau(data_controller, temp, channels, weights)[source]#

Compute the relaxation time τ(n, k) from a list of scattering channels.

Constructs the total scattering rate as a harmonic sum of individual channel contributions

\[\frac{1}{\tau_{n\mathbf{k}}} = \sum_c \frac{w_c}{\tau^{(c)}_{n\mathbf{k}}}\]

and returns \(\tau = 1 / \sum_c (w_c / \tau^{(c)})\). When no channels are provided, a constant relaxation time of unity is used (constant-τ approximation).

Each channel entry in channels can be a built-in model name string (dispatched to do_tau_models.builtin_tau_model()), a TauModel instance, or a ScatteringChannel (e.g. the electron-phonon TransportSpectralChannel, a ResidualResistivity), whose rate \(1/ au\) (1/s) enters the same harmonic sum with its weight.

Parameters:
  • data_controller (DataController) – Provides E_k, bnd, and tau_dict (model parameters).

  • temp (float) – \(k_BT\) in eV (not kelvin), passed to each model’s evaluate method; scattering channels receive temp / k_B in K and the band energies relative to the Fermi level.

  • channels (list) – Scattering channel identifiers (strings or TauModel objects). Pass an empty list or None for constant-τ.

  • weights (list of float) – Per-channel harmonic-sum weights. Defaults to all-ones when empty.

Returns:

tau – Relaxation time in arbitrary units (consistent with tau_dict).

Return type:

ndarray, shape (nkpts_local, bnd, nspin)

Raises:

Exception – If a required parameter is missing from attr['tau_dict'].

PAOFLOW.boltzmann.do_Boltz_tensors.L_loop(data_controller, temp, smearing, ene, velkp, t_tensor, alpha, ispin)[source]#

Inner BZ summation loop for one Lu1d45 transport tensor.

Evaluates the energy-resolved integral

\[\begin{split}L^{(\\alpha)}_{ij}(\\varepsilon) = \\frac{1}{N_k} \\sum_{n\\mathbf{k}} \\tau_{n\\mathbf{k}}\\, v^i_{n\\mathbf{k}}\\, v^j_{n\\mathbf{k}}\\, \\sigma(E_{n\\mathbf{k}}, \\varepsilon, \\delta_k)\\, (E_{n\\mathbf{k}} - \\varepsilon)^\\alpha\end{split}\]

where \(\\sigma\) is the smearing kernel (Fermi-Dirac derivative, Gaussian, or Methfessel-Paxton depending on smearing).

The loop runs over the k-point slice held by the local MPI rank. Results are subsequently reduced across ranks by the caller.

Parameters:
  • data_controller (DataController) – Provides E_k, deltakp, scattering_tau, nkpnts, bnd.

  • temp (float) – Temperature in eV.

  • smearing (str or None) – None uses the Fermi-Dirac window; 'gauss' or 'm-p' uses adaptive smearing widths from deltakp.

  • ene (ndarray, shape (ne,)) – Energy (chemical-potential) grid.

  • velkp (ndarray, shape (nkpts_local, 3, bnd, nspin)) – Band velocities.

  • t_tensor (ndarray, shape (nc, 2), int) – Tensor component pairs [i, j] to evaluate (upper triangle).

  • alpha (int) – Power of the (E - \u03b5) kernel: 0, 1, or 2.

  • ispin (int) – Spin channel index.

Returns:

L – Local contribution to Lu1d45 from this MPI rank’s k-points.

Return type:

ndarray, shape (3, 3, ne)

PAOFLOW.boltzmann.do_Boltz_tensors.L_loop_hall(data_controller, temp, smearing, ene, velkp, t_tensor, alpha, ispin)[source]#

Inner BZ summation loop for the Hall transport tensor.

Computes the rank-3 Hall kernel

\[\begin{split}L^{\\rm Hall}_{ijp}(\\varepsilon) = \\frac{1}{N_k} \\sum_{n\\mathbf{k}} \\tau^2_{n\\mathbf{k}} \\left(\\sum_{qr} \\epsilon_{pqr}\\, v^i_{n\\mathbf{k}}\\, v^r_{n\\mathbf{k}}\\, M^{-1}_{jq,n\\mathbf{k}}\\right) \\sigma(E_{n\\mathbf{k}}, \\varepsilon, \\delta_k)\\, (E_{n\\mathbf{k}} - \\varepsilon)^\\alpha\end{split}\]

The Levi-Civita symbol \(\\epsilon_{pqr}\) is a static rank-3 tensor; the inverse effective-mass tensor components are read from arry['d2Ed2k'] and assembled into a full 3u00d73 matrix.

The \((E - \\varepsilon)^\\alpha\) factor is applied only in the Fermi-Dirac branch; with adaptive smearing the kernel is returned for alpha = 0 irrespective of the argument, and the energy moment is taken by the caller.

Parameters:
  • data_controller (DataController) – Provides E_k, deltakp, d2Ed2k, scattering_tau, nkpnts, bnd, nspin.

  • temp (float) – Temperature in eV.

  • smearing (str or None) – None uses the Fermi-Dirac window; 'gauss' or 'm-p' uses adaptive smearing widths from deltakp.

  • ene (ndarray, shape (ne,)) – Energy (chemical-potential) grid.

  • velkp (ndarray, shape (nkpts_local, 3, bnd, nspin)) – Band velocities.

  • t_tensor (ndarray, shape (nc, 2), int) – Tensor component pairs (used to unpack the symmetric effective-mass tensor stored as 6 independent components).

  • alpha (int) – Power of the (E - \u03b5) kernel: 0 or 1. Honoured only when smearing is None.

  • ispin (int) – Spin channel index.

Returns:

L_hall – Local contribution to the Hall tensor from this MPI rank’s k-points.

Return type:

ndarray, shape (3, 3, 3, ne)