Computation Kernel for Feynman Diagrams
Daria Gazizova, Rayan Farid, B. D. E. McNiven, I. Assi, Ethan G. Armstrong, J. P. F. LeBlanc · 2025
Read Computation Kernel for Feynman Diagrams by Daria Gazizova, Rayan Farid, B. D. E. McNiven, I. Assi, Ethan G. Armstrong, J. P. F. LeBlanc (2025) 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-2503-08444.txt.
Introduction
Gazizova and five co-authors present a universal computation kernel for Feynman diagrams in many-body theory, in a paper from twenty twenty-five. Many-body perturbation theory evaluates diagrams full of temperature and frequency sums. Projecting the single-particle Green's function to an auxiliary space transforms the problem. Any Feynman diagram then converts into a standard universal kernel form.
The kernel holds all explicit temperature and frequency dependence of the diagram. It contains no problem-specific information once constructed, by design. Diagram results become linear combinations of kernel pieces with problem-specific coefficients. A low-order self-energy diagram demonstrates the construction concretely. Distinct physical problems reuse the same kernel for their solutions afterward.
Catalog record
| CREATOR | Daria Gazizova, Rayan Farid, B. D. E. McNiven, I. Assi, Ethan G. Armstrong, J. P. F. LeBlanc |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 35 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2503-08444
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Computation Kernel for Feynman Diagrams Daria Gazizova, Rayan Farid, B. D. E. McNiven, I. Assi, Ethan G. Armstrong, J. P. F. LeBlanc Abstract We present a general representation for solving problems in many-body perturbation theory. By defining an overcomplete representation for the single-particle Green’s function we show how one can convert an arbitrary Feynman graph to a universal kernel representation. Once constructed, the computation kernel contains no problem specific information yet contains all explicit temperature and frequency dependence of the diagram. This computation kernel is problem agnostic, and valid for any physical problem that would normally leverage the Matsubara formalism of many-body perturbation theory. The result of any diagram can be written as a linear combination of these computation kernel elements with coefficients given by a sum over products of known tensor elements that are themselves problem specific and represent spatial degrees of freedom. We probe the efficacy of this approach by generating the computation kernel for a low order self-energy diagram which we then use to construct solutions to distinct problems. I Introduction The need for precise evaluation of correlated electron problems drives the development of numerous numerical approaches and serves as a catalyst for advancing quantum computational hardware and algorithms. Despite advancements in algorithms for quantum hardware, the need for purely classical approaches remains undiminished, as they have accounted for virtually all useful computations of correlated electron systems to date. Wu et al. 2024 ; Louvet et al. 2024 On classical hardware, quantum algorithm development spans several research areas, including the study of equilibrium and non-equilibrium systems, methods for finite temperature and ground state properties, and approaches using wavefunctions versus correlation functions. All these avenues are worth pursuing, with the shared goal of solving challenging quantum systems on classical or quantum hardware. Qin et al. 2022 Among the methods for finite temperature systems in equilibrium is many-body perturbation theory that is formulated in terms of Feynman diagrams. While not a new concept, Feynman diagrams remain a cornerstone of computational physics and chemistry. In quantum chemistry, the lowest-order perturbations, Hartree-Fock theory, serve as the foundation for many approaches and can be formulated in terms of diagrams. Going beyond Hartree-Fock, diagrammatic methods are valuable not only due to their broad applicability but also for their potential to provide a path to finite-temperature results for dynamic properties while offering the flexibility to evaluate a variety of observables. These methods are, however, at least exponential in complexity as the order of the interaction increases, Rossi et al. 2016 ; Rossi 2017 ; Šimkovic and Kozik 2019 and self-consistent approaches have shown evidence of false convergence Kozik et al. 2015 ; Eßl et al. 2025 ; McNiven et al. 2021 . At the same time there have been algorithmic advancements intended to drastically improve computation time. In the case of real and imaginary time representations, application of quantics tensor trains have shown promise for reducing the complexity of numerical problems by reducing expressions to tensor products Ritter et al. 2024 ; Erpenbeck et al. 2023 ; Fernández et al. 2024 ; Rohshap et al. 2024 . Other alternate representations for Matsubara Green’s functions show similar promise for application to diagrammatic problems Shinaoka et al. 2022 ; Dirnböck et al. 2024a ; Kaye et al. 2022a ; Kiese et al. 2025 ; Otsuki et al. 2020 . In the case of problems formulated in the Matsubara frequency domain there are also symbolic approaches that have emerged and are able to generate analytic solutions to internal temporal degrees of freedom of arbitrary Feynman diagrams Vučičević and Ferrero 2020 ; Vučičević et al. 2021 ; Kovačević et al. 2025 ; Taheridehkordi et al. 2019 . One such symbolic approach, algorithmic Matsubara integration (AMI), has been applied to a broad spectrum of problems Taheridehkordi et al. 2019 ; Taheridehkordi et al. 2020a ; Taheridehkordi et al. 2020b ; Assi and LeBlanc 2024 ; McNiven et al. 2021 ; McNiven et al. 2022 ; Farid et al. 2023 ; Farid and LeBlanc 2024 ; LeBlanc et al. 2022 ; Gazizova and LeBlanc 2023 ; Burke et al. 2023 . However, the AMI approach is currently limited in its application for a number of reasons. Computationally, the method suffers from a nearly factorial growth of the number of function calls required to evaluate diagrams of increasing order limiting applications to fifth or sixth order in practice LeBlanc et al. 2022 ; McNiven et al. 2021 . The method also requires access to the pole structure of the Green’s function which prevents one from merging AMI with self-consistent schemes or applying AMI to systems with non-diagonal Green’s functions. Alternate representations of Green’s functions provide a possible path forward for the AMI approach. In particular, the discrete Lehmann representation (DLR) is an alternate function representation for Matsubara Green’s functions in both the imaginary time and imaginary frequency domains. The DLR represents the Green’s function in imaginary frequency via a discretized spectral kernel wherein the pole structure is symbolically accessible and this allows one to merge the DLR with AMI to provide self-consistent schemes. This was recently shown to be effective for studying single-band impurity problems in both single-shot and self-consistent schemes. Gazizova et al. 2024 There it was shown that an exponential trade-off in computational expense is accompanied by an exponential convergence of the result to the numerically exact result. However, applying this approach to multi-band or finite systems scales poorly due to the addition of yet another exponential computational scaling on top of existing factorial scalings. In this work, we make substantial progress in applying the DLR combined with AMI to arbitrary systems beyond impurities such as multi-band problems or lattice problems. We consider the implications of applying the DLR to Feynman diagrams with additional internal spatial degrees of freedom and find that there is a factorizable component of the problem that is identical to the case of an impurity problem but depends only on the non-physical space of DLR poles. This realization allows us to rephrase the factorizable component as a kernel for systems with any set of internal spatial degrees of freedom. We call this factorizable component the computation kernel. Once the computation kernel is computed for a given diagram, it can be reused endlessly for new problems regardless of the spatial geometry of the system. We show that diagrams can be decomposed as a linear combination of computation kernel entries with coefficients that are a product of tensor elements that are apriori known from the DLR fit of the single-particle Green’s function. II General Methodology II.1 Discrete Lehmann Representation This work relies heavily on the formulation of the discrete Lehmann representation (DLR) presented in Refs. Kaye et al. 2022a ; Kaye et al. 2022b for single particle Green’s functions. The premise of the DLR is to approximate the spectral decomposition in the truncated form, G(i\omega_{n})=\int\limits_{-\Lambda}^{\Lambda}K(i\omega_{n},x)\rho(x)dx, (1) where \rho is the density and the kernel is given by K(i\omega_{n},x)=\frac{1}{i\omega_{n}-x} when represented in Matsubara frequencies, i\omega_{n} . In the frequency domain, the DLR procedure assigns a set of nodes along the imaginary axis, \{i\omega_{DLR}\} , that are determined uniquely for a given choice of \Lambda for a desired error tolerance at a chosen inverse temperature, \beta . The coefficients