Dear EPW community,
I am working on a project about MgB2, and I want to extract the band resolved anisotropic Eliashiberg spectral function (a2f) in terms of pi-band and sigma-band as shown in the Fig. 3(d) in this PRL paper: https://journals.aps.org/prl/pdf/10.1103/PhysRevLett.94.227005.
I am wondering if EPW is able to calculate this result, any suggestion would be useful. Thank you very much!
Shunshun
University of Virginia
Department of Materials Science and Engineering
Band resolved anisotropic Eliashiberg spectral function
Moderator: stiwari
Re: Band resolved anisotropic Eliashiberg spectral function
Hi Shunshun,
While a2f itself is calculated within the evaluate_a2f_lambda function in EPW (located in EPW/src/supercond.f90), calculations like those shown in FIG. 3 of the PRL paper cannot be done using EPW. If by 'a2f decomposed by band index' you mean in the literal sense, this can be computed by modifying evaluate_a2f_lambda. However, since the band index in Bloch states is simply a numbering of eigenvalues at each k-point, decomposing the contribution by band index is different from decomposing contributions from σ-bands or π-bands, as done in the PRL paper.
Fundamentally, π-bonding is a classification of covalent bonds based on a real-space representation, whereas the formula we use for a2f is based on the Bloch representation, as described in equations (33)-(35) of the 2016 paper [S. Poncé et al., Comput. Phys. Commun. 209, 116 (2016)]. In the case of MgB2, π-bonding refers to bonds primarily formed by the B-pz orbitals. Therefore, to extract the contribution from the π-bands, you would need to extract the contribution from the B-pz orbitals using some kind of projection from the physical quantities calculated in the Bloch representation. However, considering the current code structure of EPW, extracting the contribution of specific atomic orbitals within evaluate_a2f_lambda would be challenging. While you could, of course, modify EPW yourself, please understand that the public version of EPW generally does not support what you are trying to achieve.
Best,
Hitoshi
While a2f itself is calculated within the evaluate_a2f_lambda function in EPW (located in EPW/src/supercond.f90), calculations like those shown in FIG. 3 of the PRL paper cannot be done using EPW. If by 'a2f decomposed by band index' you mean in the literal sense, this can be computed by modifying evaluate_a2f_lambda. However, since the band index in Bloch states is simply a numbering of eigenvalues at each k-point, decomposing the contribution by band index is different from decomposing contributions from σ-bands or π-bands, as done in the PRL paper.
Fundamentally, π-bonding is a classification of covalent bonds based on a real-space representation, whereas the formula we use for a2f is based on the Bloch representation, as described in equations (33)-(35) of the 2016 paper [S. Poncé et al., Comput. Phys. Commun. 209, 116 (2016)]. In the case of MgB2, π-bonding refers to bonds primarily formed by the B-pz orbitals. Therefore, to extract the contribution from the π-bands, you would need to extract the contribution from the B-pz orbitals using some kind of projection from the physical quantities calculated in the Bloch representation. However, considering the current code structure of EPW, extracting the contribution of specific atomic orbitals within evaluate_a2f_lambda would be challenging. While you could, of course, modify EPW yourself, please understand that the public version of EPW generally does not support what you are trying to achieve.
Best,
Hitoshi