In this thesis we present methodologies that have been developed to efficiently solve for the non-equilibrium Green’s Function (NEGF) of interacting electron systems, and demonstrate their applications to several different condensed matter systems. These solvers are based on the numerical integration of the equations of motion for the NEGF, which are known as the Kadanoff-Baym Equations (KBE). The KBE are a general set of equations which allow for the study of a wide range of systems that are of interest to condensed matter physicists, including superconductivity, nuclear matter, molecules and atoms, and qubits, among many others. They are also useful for solving impurity models and systems in the dynamical mean-field theory (DMFT) approximation through their ability to incorporate dynamical hybridization functions. The generality of these equations comes from their connection to time-dependent Many-Body Perturbation Theory (MBPT), which introduces non-Markovian dynamical correlations through the self-energy functional. MBPT ensures important conservation laws, such as those for density and energy, are obeyed no matter the type of time-dependent perturbations that are applied to the Hamiltonian. The non-Markovian correlations manifest in a kernel that couples the dynamics to the entire history of the system. This coupling leads to computational costs that become prohibitively expensive as longer-time dynamics are probed, this is especially true for systems which decohere slowly, leading to undamped dynamics and kernels. Due to the large amount of interest in non-equilibrium quantum dynamics, the availability of efficient solvers for the KBE are of great importance. This thesis will focus on the implementation and analysis of integration methods for the KBE, including compression algorithms which lower computational complexity and adaptive integration techniques which allow for resolution of dynamics which vary over wide ranges of timescales. Chapter 1 gives an introduction to quantum systems and the problems we will be studying. Chapter 2 provides a derivation of the NEGF and its representation on the Keldysh contour. Chapter 3 derives the KBE from MBPT and introduces the self- energy and its conservation laws. Chapter 4 introduces the numerical techniques used to integrate the KBE, as well as an analysis of its causality structure. In Chapter 5, we use compression techniques to lower the complexity of the KBE, and provide details on its implementation. In Chapter 6 we apply the compression algorithm to study the Higgs mode in a superconducting system. In Chapter 7, we couple our solver to an adaptive integrator and analyze its performance against fixed-timestep integrators. Finally, Chapter 8 provides a summary of the results and an outlook for future work on efficient solvers.
Understanding the dynamics of nonequilibrium quantum many-body systems is an important research topic in a wide range of fields across condensed matter physics, quantum optics, and high-energy physics. However, numerical studies of large-scale nonequilibrium phenomena in realistic materials face serious challenges due to intrinsic high-dimensionality of quantum many-body problems. The nonequilibrium properties of many-body systems can be described by the dynamics of the Green's function of the system, whose time evolution is given by a high-dimensional system of integro-differential equations, known as the Kadanoff-Baym equations (KBEs). The time-convolution term in KBEs, which needs to be recalculated at each time step, makes it difficult to perform long-time simulations. In this paper, we develop an operator-learning framework based on Recurrent Neural Networks (RNNs) to address this challenge. The proposed framework utilizes RNNs to learn the nonlinear mapping between Green's functions and convolution integrals in KBEs. By using the learned operators as a surrogate model in the KBE solver, we obtain a general machine-learning scheme for predicting the dynamics of nonequilibrium Green's functions. This new methodology reduces the temporal computational complexity from
O(Nt3) to
O(Nt), where
Nt is the total time steps taken in a simulation, thereby making it possible to study large many-body problems which are currently infeasible with conventional KBE solvers. Through different numerical examples, we demonstrate the effectiveness of the operator-learning based approach in providing accurate predictions of physical observables such as the reduced density matrix and time-resolved photoemission spectra. Moreover, our framework exhibits clear numerical convergence and can be easily parallelized, thereby facilitating many possible further developments and applications.
A time-stepping scheme with adaptivity in both the step size and the integration order is presented in the context of non-equilibrium dynamics described via Kadanoff-Baym equations. The accuracy and effectiveness of the algorithm are analysed by obtaining numerical solutions of exactly solvable models. We find a significant reduction in the number of time-steps compared to fixed-step methods. Due to the at least quadratic scaling of Kadanoff-Baym equations, reducing the amount of steps can dramatically increase the accessible integration time, opening the door for the study of long-time dynamics in interacting systems. A selection of illustrative examples is provided, among them interacting and open quantum systems as well as classical stochastic processes. An open-source implementation of our algorithm in the scientific-computing language Julia is made available.
Abstract Understanding the dynamics of nonequilibrium quantum many-body systems is an important research topic in a wide range of fields across condensed matter physics, quantum optics, and high-energy physics. However, numerical studies of large-scale nonequilibrium phenomena in realistic materials face serious challenges due to intrinsic high-dimensionality of quantum many-body problems and the absence of time-invariance. The nonequilibrium properties of many-body systems can be described by the dynamics of the correlator, or the Green’s function of the system, whose time evolution is given by a high-dimensional system of integro-differential equations, known as the Kadanoff–Baym equations (KBEs). The time-convolution term in KBEs, which needs to be recalculated at each time step, makes it difficult to perform long-time numerical simulation. In this paper, we develop an operator-learning framework based on recurrent neural networks (RNNs) to address this challenge. We utilize RNNs to learn the nonlinear mapping between Green’s functions and convolution integrals in KBEs. By using the learned operators as a surrogate model in the KBE solver, we obtain a general machine-learning scheme for predicting the dynamics of nonequilibrium Green’s functions. Besides significant savings per each time step, the new methodology reduces the temporal computational complexity from O ( N t 3 ) to O ( N t ) where Nt is the number of steps taken in a simulation, thereby making it possible to study large many-body problems which are currently infeasible with conventional KBE solvers. Through various numerical examples, we demonstrate the effectiveness of the operator-learning based approach in providing accurate predictions of physical observables such as the reduced density matrix and time-resolved photoemission spectra. Moreover, our framework exhibits clear numerical convergence and can be easily parallelized, thereby facilitating many possible further developments and applications.
Abstract Understanding the dynamics of nonequilibrium quantum many-body systems is an important research topic in a wide range of fields across condensed matter physics, quantum optics, and high-energy physics. However, numerical studies of large-scale nonequilibrium phenomena in realistic materials face serious challenges due to intrinsic high-dimensionality of quantum many-body problems and the absence of time-invariance. The nonequilibrium properties of many-body systems can be described by the dynamics of the correlator, or the Green's function of the system, whose time evolution is given by a high-dimensional system of integro-differential equations, known as the Kadanoff-Baym equations (KBEs). The time-convolution term in KBEs, which needs to be recalculated at each time step, makes it difficult to perform long-time numerical simulation. In this paper, we develop an operator-learning framework based on Recurrent Neural Networks (RNNs) to address this challenge. We utilize RNNs to learn the nonlinear mapping between Green's functions and convolution integrals in KBEs. By using the learned operators as a surrogate model in the KBE solver, we obtain a general machine-learning scheme for predicting the dynamics of nonequilibrium Green's functions. Besides significant savings per each time step, the new methodology reduces the temporal computational complexity from
O(Nt3) to
O(Nt) where
Nt is the number of steps taken in a simulation, thereby making it possible to study large many-body problems which are currently infeasible with conventional KBE solvers. Through various numerical examples, we demonstrate the effectiveness of the operator-learning based approach in providing accurate predictions of physical observables such as the reduced density matrix and time-resolved photoemission spectra. Moreover, our framework exhibits clear numerical convergence and can be easily parallelized, thereby facilitating many possible further developments and applications.
The time evolution in quantum many-body systems after external excitations is attracting high interest in many fields, including dense plasmas, correlated solids, laser-excited materials, or fermionic and bosonic atoms in optical lattices. The theoretical modeling of these processes is challenging, and the only rigorous quantum-dynamics approach that can treat correlated fermions in two and three dimensions is nonequilibrium Green functions (NEGF). However, NEGF simulations are computationally expensive due to their
T3 scaling with the simulation duration
T. Recently,
T2 scaling was achieved with the generalized Kadanoff-Baym ansatz (GKBA), for the second-order Born (SOA) self energy, which has substantially extended the scope of NEGF simulations. In a recent Letter [Schl\"unzen et al., Phys. Rev. Lett. 124, 076601 (2020).] we demonstrated that GKBA-NEGF simulations can be efficiently mapped onto coupled time-local equations for the single-particle and two-particle Green functions on the time diagonal, hence the method has been called the G1-G2 scheme. This allows one to perform the same simulations with order
T1 scaling, both for SOA and
GW self energies giving rise to a dramatic speedup. Here we present more details on the G1-G2 scheme, including derivations of the basic equations including results for a general basis, for Hubbard systems, and for jellium. Also, we demonstrate how to incorporate initial correlations into the G1-G2 scheme. Further, the derivations are extended to a broader class of self energies, including the
T matrix in the particle-particle and particle-hole channels and the dynamically-screened-ladder approximation. Finally, we demonstrate that, for all self energies, the CPU-time scaling of the G1-G2 scheme with the basis dimension
Nb can be improved compared to our first report: The overhead compared to the original GKBA is not more than an additional factor
Nb, even for Hubbard systems.