Ladder equation for the three-particle vertex and its approximate solution

Patrick Kappl , Tin Ribic, Anna Kauch, Karsten Held · 2024

Read Ladder equation for the three-particle vertex and its approximate solution by Patrick Kappl , Tin Ribic, Anna Kauch, Karsten Held (2024) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2412-01848.txt.

Introduction

Patrick Kappl and co-authors extend a classic equation system from pairs of particles to triples, in a 2024 paper. The Bethe-Salpeter equations govern how two interacting particles bind and scatter. Their three-particle vertex, the hub where three lines meet, needs its own ladder equations. The authors generalize the three known pair ladders into ten triple ladders, an exact framework in principle.

Exactness stalls on a missing block: the part of the vertex irreducible in each channel. They approximate it by stitching together pair-level irreducible pieces. Tested against an exact solution for a nonlinear response function, the shortcut fares only qualitatively, and only for weak interactions. The honest verdict maps where cheap three-particle physics can be trusted.

Catalog record

CREATORPatrick Kappl , Tin Ribic, Anna Kauch, Karsten Held
YEAR2024
COLLECTIONPAPERS
FORMATTXT
FILE SIZE136 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2412-01848

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Ladder equation for the three-particle vertex and its approximate solution Patrick Kappl , Tin Ribic, Anna Kauch, Karsten Held Abstract We generalize the three two-particle Bethe-Salpeter equations to ten three-particle ladders. These equations are exact and yield the exact three-particle vertex, if we knew the three-particle vertex irreducible in one of the ten channels. However, as we do not have this three-particle irreducible vertex at hand, we approximate this building block for the ladder by the sum of two-particle irreducible vertices each connecting two fermionic lines. The comparison to the exact solution shows that this approximation is only good for rather weak interactions and even than only qualitatively — at least for the non-linear response function analyzed. \svgpath ./diagrams/./diagrams/3p_ladders I Introduction Feynman diagrams are at the heart of the theory of strongly correlated electrons. Textbooks Abrikosov et al. 1975 study diagrams for one- and two-particle Green’s functions which provide renormalizations (and life-times) of the quasiparticles and susceptibilities, respectively. One-particle Green’s functions and self-energies also form the basis of dynamical mean field theory (DMFT) Metzner and Vollhardt 1989 ; Georges and Kotliar 1992 ; Georges et al. 1996 ; Pavarini et al. 2014 ; whereas two-particle Green’s functions and vertices build the fundament of diagramamtic extensions of DMFT Toschi et al. 2007 ; Rubtsov et al. 2008 ; Rohringer et al. 2018 . Electronic correlations on the next level of three-particle Feynman diagrams are hitherto hardly explored. A calculation of selected three-particle contributions shows that these may become relevant for diagrammatic extensions of DMFT Ribic et al. 2017a ; Ribic et al. 2017b 1 1 1 In Ref. Hafermann et al. 2009 only minor corrections have been reported, but this calculation was also in another parameter regime. . Similarly, for the functional renormalization group the next level, hitherto largely neglected, is the three-particle level Metzner et al. 2012 . This calls for a better understanding and for developing methods to actually calculate three-particle Green’s functions and vertices more systematically. Physics-wise, three-particle Green’s functions are relevant for calculating Raman and Hall responses Jorio et al. 2011 as well as for non-linear response functions Kubo 1957 ; Rostami et al. 2017 ; Rostami et al. 2021 ; Michishita and Peters 2021 ; Kappl et al. 2023 . Traditionally such calculations have been restricted to calculating the three-particle correlator from bubble contributions with quasiparticle-renormalizations of the Green’s functions but without vertex corrections; or more recently by including a two-particle vertex between the electron-hole pair of one of the bosonic fields/repsonses Rostami et al. 2021 . The three-particle Green’s function has also recently beed used to obtain electronic spectra Riva et al. 2022 ; Riva et al. 2023 In related research fields, three-particle physics is also relevant for trions in semiconductors, which are typically treated as a three-particle Hamiltonian Lampert 1958 ; Combescot et al. 2004 . In quantum chromodynamics (QCD), three-particle extensions of the Bethe-Salpeter equations, the so-called Faddeev equations Faddeev 1961 , are employed for solving the three-particle problem of e.g. quarks forming protons or neutrons. Approximations such as the rainbow-ladder truncation Sanchis-Alepuz et al. 2013 have been developed. The Faddeev equations have also been employed in nuclear physics to study the propagation of two holes and a particle interacting with each other through phonons Barbieri and Dickhoff 2001 . A notable difference is that in solid-state physics our three particles or holes propagate in a background of many more electrons with which they interact strongly. This requires more involved Feynman diagrams where we can have at a given time many more (or less) than three particle or hole excitations. A mapping onto a three-particle Hamiltonian is no longer possible. In this paper, we derive a three-particle analog of the Bethe-Salpeter equation. Irreducibility on the three-particle level is more involved and leads to additional one-particle reducible diagrams that need to be considered for the nine particle-particle-hole channels but not for the particle-particle-particle channel. Further, we solve this generalized Bethe-Salpeter ladder for an approximate irreducible three-particle vertex that consists only of the two-particle vertices connecting all possible pairs of particle (hole) lines. To fulfill crossing symmetries of the vertex, contributions from multiple channels need to be considered. The thus approximated three-particle vertex is then compared to the exact one for an Anderson impurity model (AIM) at DMFT self-consistency. For this AIM, the vertex is local and can be calculated exactly. The outline of the paper is as follows: Section II sets the stage, defining the AIM and recalling some essentials for the two-particle diagrammatics. Section III defines the three-particle Green’s function and vertex, as well as our frequency and spin notation. Section IV is the main part of this paper. Here, we derive the three-particle Bethe-Salpeter-like ladder equations in Section IV.1 and develop an approximation in terms of two particle irreducible vertices in Section IV.2 . Section V presents numerical results for the latter and compares them to exact ones. Finally, Section VI summarizes our results and provides a brief outlook. II Theoretical background In this Section, we define the AIM, the two-particle Green’s function and vertex. We further recall the parquet and Bethe-Salpeter ladder equation as well as the crossing symmetry, which is very helpful for later understanding the extensions to three particles. II.1 Anderson impurity model The AIM, introduced in 1961 by P. W. Anderson to describe magnetic impurities in metals Anderson 1961 , has become a popular model in the field of correlated many-electron systems. Arguably, this is because even though its Hamiltonian, which is described below, is rather simple it already shows effects of strong correlation, more specifically the Kondo effect Kondo 1964 ; Schrieffer and Wolff 1966 . Another important aspect is that its solution is connected to local correlation functions of more complex lattice Hamiltonians in DMFT Georges et al. 1996 and its diagrammatic extensions Rohringer et al. 2018 . For an in-depth review of the physics of the AIM and the Kondo problem see Refs. Hewson 1993 ; Coleman 2015 . The Hamiltonian of the AIM reads \hamiltonian[AIM]=\energy\nq+\interaction\nq_{\up}\nq_{\down}+\sum_{k,\sigma}\energy^{\vphantom{\dagger}}_{k}\cqdag_{k\sigma}\cq_{k\sigma}\!+\!\sum_{k,\sigma}\left(\hybridization^{\vphantom{\dagger}}_{k}\fqdag_{\sigma\vphantom{k}}\cq_{k\sigma}\!+\mathrm{h.c.}\right). (1) The first two terms describe the impurity with one-particle energy \energy . Electrons on the impurity are created and annihilated with \fqdag_{\sigma} and \fq_{\sigma} , respectively. As usual, the density operators are given by \nq=\nq_{\up}+\nq_{\down} , with \nq_{\sigma}=\fqdag_{\sigma}\fq_{\sigma} . If the impurity is doubly occupied, the local Coulomb interaction \interaction comes into play. The third term models a bath of noninteracting electrons with dispersion relation \energy_{k} in terms of creation and annihilation operators \cqdag_{k\sigma} and \cq_{k\sigma} . The last two terms describe hopping (hybridization) from the bath to the impurity ( \hybridization_{k} ) and vice versa (Hermitian conjugate; \hybridization^{*}_{k} ). Here, we employ the AIM with a real, k -independent hybridization \hybridization_{k}\equiv\hybridization\in\mathbb{R} and with a bath density of states (DOS) given by DMFT self-consistency for the Hubbard model Georges et al. 1996 with the same U , hopping t\equiv 1 and inverse temperature \beta=1/T=10 (we set the Boltzmann constant k_{B} to 1 ). The motivation behind this choice is that we are eventually interested in diagrammatic extensions of DMFT which have the local two- or, on the next level, three-particle vertices of such an AIM as a starting point, see Ref. Rohringer et al. 2018 . However, here

Read the complete text