| many-body perturbation theory | |
|---|---|
| Name | Many-body perturbation theory |
| Caption | Feynman diagrammatic representation of interacting fermions |
| Field | Quantum physics |
| Introduced | 1950s |
| Pioneers | Schwinger, Feynman, Landau, Bohm |
many-body perturbation theory
Many-body perturbation theory (MBPT) is a collection of analytic and diagrammatic methods used to treat interacting quantum systems by expanding around a solvable reference, typically a noninteracting or mean-field state. It provides systematic corrections to observables such as energies, correlation functions and response functions, and underpins quantitative predictions in condensed matter physics and nuclear physics. MBPT matters because it bridges exactly solvable models and realistic interacting systems, enabling computation of quasiparticle properties, excitation spectra, and screening effects.
Many-body perturbation theory formulates the dynamics of ensembles of identical particles (fermions or bosons) using perturbative expansions in an interaction parameter such as the Coulomb coupling or nucleon-nucleon potential. The approach connects to concepts in statistical mechanics, quantum field theory, and atomic physics by employing second quantization, creation/annihilation operators, and time-ordered products. Foundational results include the Dyson equation for propagators, the concept of the self-energy, and classification of diagrams into reducible and irreducible subsets. MBPT is applied to electrons in solids, nucleons in nuclei, cold atomic gases, and quantum chemistry problems where correlation beyond mean-field Hartree–Fock is important.
The formal MBPT formalism uses perturbative expansion of the S-matrix or time-evolution operator in the interaction representation. Wick's theorem reduces time-ordered operator products to sums of contractions, which are represented graphically by Feynman diagrams. Diagrams are organized by topology and order in the coupling constant; linked-cluster theorems guarantee extensivity of the energy. Diagrammatic resummations, such as ladder and ring series, capture collective modes and screening. Key historical developments include the diagrammatic techniques introduced by Feynman and formal operator methods by Schwinger; later refinements arose in works by Van Hove and Gell-Mann and Low.
Single-particle and two-particle Green's functions (propagators) are central objects in MBPT. Time-ordered Green's functions G(x,x') encode occupation and spectral information and satisfy the Dyson equation G = G0 + G0 Σ G, where Σ is the self-energy that encapsulates exchange and correlation effects beyond the noninteracting propagator G0. The self-energy is generally nonlocal in space and time and energy-dependent, producing finite lifetimes and renormalized quasiparticle energies. Two-particle Green's functions lead to Bethe–Salpeter equations for optical excitations. Important named constructs include the Dyson equation, the Bethe–Salpeter equation, and vertex functions such as the irreducible vertex Γ used to enforce conservation laws and compute response functions.
Several approximation schemes derived from MBPT are widely used. The GW approximation computes Σ ≈ i G W where W is the screened interaction often obtained from the random phase approximation (RPA); GW yields improved quasiparticle energies in many semiconductors and metals and is associated with work by Hedin. The RPA sums ring diagrams to capture long-range screening and collective plasmons. The T-matrix approximation resums ladder diagrams, important for strong scattering, pairing, and near-resonant interactions in ultracold atoms and nuclear matter. Combining GW with the Bethe–Salpeter equation yields accurate optical spectra in solids. Conserving approximations such as those derived from the Luttinger–Ward functional or Baym–Kadanoff formalism ensure thermodynamic consistency.
In condensed matter, MBPT predicts band structures, quasiparticle lifetimes, excitation spectra, and dielectric functions for materials studied at institutions like Bell Labs, Argonne National Laboratory, and university research groups. GW and GW+BSE approaches are standard for computing bandgaps of semiconductors and insulators; RPA underlies van der Waals interactions and screening in two-dimensional materials like graphene. In nuclear physics, MBPT and resummations such as particle-particle ladders are used to compute nuclear matter equations of state, shell-model effective interactions, and pairing gaps, with implementations at facilities including Brookhaven National Laboratory and TRIUMF. Applications extend to quantum chemistry methods like perturbative Møller–Plesset (MP2) theory and coupled-cluster theory connections.
Practical MBPT requires numerical evaluation of multidimensional integrals, analytic continuation, and solution of integral equations. Implementations appear in electronic structure codes such as VASP, Quantum ESPRESSO, ABINIT, and Yambo for GW and BSE calculations. Algorithms exploit fast Fourier transforms, Wannier interpolation, Monte Carlo sampling for diagrammatic expansions, and diagrammatic Monte Carlo for nonperturbative summations. Finite-temperature formulations use Matsubara frequency techniques; real-frequency methods apply Pade or maximum-entropy analytic continuation. High-performance computing and parallelization are crucial for treating large unit cells and fine k-point meshes.
MBPT is limited by convergence issues in strongly correlated regimes, truncation errors, and dependence on reference states. Extensions include nonperturbative resummation techniques, functional renormalization group, dynamical mean-field theory (DMFT) combined with diagrammatic methods (DFT+DMFT, GW+DMFT), and bold-line diagrammatic Monte Carlo. MBPT is formally a specialization of quantum field theory applied to many-particle condensed matter and nuclear systems, sharing renormalization concepts, Ward identities, and path-integral formulations. Ongoing research aims to rigorously connect MBPT approximations with exact conservation laws and to improve scalability for predictive materials and nuclear modeling.