Development of low-scaling methods to calculate ground state energies and analytical gradients based on the adiabatic-connection fluctuation-dissipation theorem
New methods to efficiently calculate energetics and first order-properties for mean-field and correlated electronic structure theories are presented. In linear-scaling short-range hybrid calculations new integral screening criteria exploiting the locality of the attenuated Coulomb operator are introduced that allow to significantly increase the performance of these density functionals. This enables short-range hybrid calculations at a similar cost as pure semi-local density functional theory (DFT) calculations with increased accuracy due to the admixture of exchange. At the level of correlated electronic structure theory, the realm of systems which can be calculated with a linear-scaling random-phase-approximation (RPA) method is extended with a novel multi-node parallel algorithm. Furthermore, more efficient quadrature schemes are used based on the equivalence of the employed integral transforms with the Fourier transform of the non-interacting polarizability. In combination with a new way to introduce Cholesky orbitals this results in a more effient and numerically more accurate linear-scaling RPA correlation energy method with improved memory requirements. Combining these techniques further with an approach to calculate analytical first order properties using quantities from many body perturbation theory, allows to present a low-scaling method that gives access to RPA gradients for molecular systems with several hundred atoms. Moreover linear- and low-scaling methods to calculate different beyond RPA correlation energies are devised, exploiting the locality of exchange type contractions. This enables to apply these more accurate RPA methods to significantly larger systems, which allows to demonstrate the gain in accuracy for large, dispersion dominated systems. The newly developed methods extend the scope of RPA and RPA with exchange correlation energies, and first order RPA properties significantly, while the accuracy is under full numerical control. Furthermore, there is no overhead to the respective canonical algorithm, making the presented methods competitive also for small system sizes. Finally, a new beyond RPA scheme is presented that combines the benefits of plain RPA and RPA with second-order screened exchange, leading to a more balanced, highly accurate post Kohn-Sham method.
New methods to efficiently calculate energetics and first order-properties for mean-field and correlated electronic structure theories are presented. In linear-scaling short-range hybrid calculations new integral screening criteria exploiting the locality of the attenuated Coulomb operator are introduced that allow to significantly increase the performance of these density functionals. This enables short-range hybrid calculations at a similar cost as pure semi-local density functional theory (DFT) calculations with increased accuracy due to the admixture of exchange. At the level of correlated electronic structure theory, the realm of systems which can be calculated with a linear-scaling random-phase-approximation (RPA) method is extended with a novel multi-node parallel algorithm. Furthermore, more efficient quadrature schemes are used based on the equivalence of the employed integral transforms with the Fourier transform of the non-interacting polarizability. In combination with a new way to introduce Cholesky orbitals this results in a more effient and numerically more accurate linear-scaling RPA correlation energy method with improved memory requirements. Combining these techniques further with an approach to calculate analytical first order properties using quantities from many body perturbation theory, allows to present a low-scaling method that gives access to RPA gradients for molecular systems with several hundred atoms. Moreover linear- and low-scaling methods to calculate different beyond RPA correlation energies are devised, exploiting the locality of exchange type contractions. This enables to apply these more accurate RPA methods to significantly larger systems, which allows to demonstrate the gain in accuracy for large, dispersion dominated systems. The newly developed methods extend the scope of RPA and RPA with exchange correlation energies, and first order RPA properties significantly, while the accuracy is under full numerical control. Furthermore, there is no overhead to the respective canonical algorithm, making the presented methods competitive also for small system sizes. Finally, a new beyond RPA scheme is presented that combines the benefits of plain RPA and RPA with second-order screened exchange, leading to a more balanced, highly accurate post Kohn-Sham method.
First-principles methods are a valuable tool to study the electronic structure of systems and explore the quantum-mechanical mechanisms driving them on the microscopic scale. Semi-local Density-Functional Approximations (DFAs) are still very popular due to the good balance between accuracy and computational efficiency. However, the instantaneous electronic interactions are described with an averaged potential in these methods, which is often insufficient to describe charge transfers or subtle many-body correlations. A promising way to improve the description of such interactions is the usage of advanced correlation methods which explicitly treat the electronic interactions in a many-body picture. Well known methods form this class are e.g. the Random- Phase Approximation (RPA), 2nd Order Møller-Plesset Perturbation Theory (MP2) and Coupled-Cluster Theory (CC). Unfortunately, the time and memory requirements for all these methods suffer from a steep scaling with system size in their canonical formulations. Several lower-scaling reformulations exist for these methods, but the at least cubic memory requirements and the increased algorithmic complexity prevented any efficient parallelization of these methods so far. In my thesis I developed a new lower-scaling technique for advanced correlation methods to address this challenge. The key components of this technique are the Laplace-Transformation (LT) and a new local Resolution of Identity (RI) strategy, named RI-LVL. Together, they significantly reduce the memory requirement without sacrificing accuracy and allow for an efficient parallelization by means of à priori known sparsity patterns. In combination with established integral screening techniques, the computational scaling can also be reduced considerably. The potential of this new approach is demonstrated at the example of an in-memory LT-MP2 with a MPI/OpenMP hybrid parallelization. It features an at worst cubic computational scaling, a quadratic memory scaling and a high parallel efficiency. Based on the Numeric Atom-Centered Orbital (NAO) framework used in the Fritz Haber Institute ab initio molecular simulations (FHI-aims) code, this implementation can handle both molecular and periodic systems with the same numerical footing. I will also discuss in detail how the combination of the LT and RI- LVL can be applied to RPA calculations. The accuracy and performance of my LT- MP2 is demonstrated in detail for both molecular and periodic systems. In addition, the convergence of the MP2 level adsorption energy of water on a rutile titanium dioxide (TiO2) surface in the low coverage limit is explored. This new LT-MP2 recovers more than 99.9% of the canonical MP2 correlation energy and can handle TiO2 surfaces with up to 270 atoms in the unit cell.
We have calculated the correlation energy of the homogeneous electron gas (HEG) and the dissociation energy curves of molecules with covalent bonds from an efficient implementation of the adiabatic connection fluctuation dissipation expression including the exact-exchange (EXX) kernel. The EXX kernel is defined from first-order perturbation theory and used in the Dyson equation of time-dependent density-functional theory. Within this approximation (RPAx), the correlation energies of the HEG are significantly improved with respect to the random phase approximation (RPA) up to densities of the order of ${r}_{s}\ensuremath{\approx}10$. However, beyond this value, the RPAx response function exhibits an unphysical divergence and the approximation breaks down. Total energies of molecules at equilibrium are also highly accurate, but we find a similar instability at stretched geometries. Staying within an exact first-order approximation to the response function, we use an alternative resummation of the higher-order terms. This slight redefinition of RPAx fixes the instability in total energy calculations without compromising the overall accuracy of the approach.
One of the major obstacles in modern quantum chemistry is the need to employ exceedingly large one-electron basis sets in electron correlation methods to obtain reliably accurate results. The family of explicitly correlated wave function methods circumvents this issue by introducing terms that explicitly depend on the interelectronic distance, thereby enforcing the correct electron–electron cusp behavior, albeit at the cost of additional complicated many-electron integrals when evaluated without further approximation. The main focus of the present work is the efficient, low-scaling evaluation of explicitly correlated approaches, examined here through the explicitly correlated correction to second-order Møller–Plesset perturbation theory (MP2-F12) as a representative example. While these methods are highly effective in reducing the basis set incompleteness error, the cost associated with the corresponding correction often far exceeds that of the underlying correlation method. To overcome these limitations, several techniques and strategies are introduced to significantly reduce computational demand and resolve the main bottlenecks that arise. A particularly straightforward approach is provided by the scaled MP2-F12 ansatz, which introduces a single empirically determined factor that scales the direct-type geminal–geminal terms to closely reproduce the MP2-F12 correction, thereby omitting the most expensive exchange-type contributions. Furthermore, the computation of Fock matrix elements spanning multiple orbital spaces is drastically accelerated by extending techniques originally developed for Hartree–Fock and density functional theory. The direct contributions are evaluated using a modified resolution-of-the-identity Coulomb (RI-J) approach employing the J-engine ansatz for integral construction, while the exchange contributions are treated via three-dimensional numerical quadrature in the form of the seminumerical sn-LinK method. These approaches reduce the computational cost by more than three orders of magnitude compared to the approximation-free evaluation while maintaining high accuracy, with additional speedups achieved by exploiting the massive parallelism of graphics processing units. The central focus of this work is the efficient decomposition of complicated exchange-type multi-electron integrals, where numerical quadrature plays a key role in improving performance. The complementary auxiliary basis set resolution of the identity (CABS-RI) reduces the most expensive exchange-type integral to a three-electron integral, which is then efficiently decomposed via numerical quadrature. In this context, a highly efficient batch-wise, distance-dependent integral screening exploiting the short-range behavior of F12 operators is introduced, significantly improving performance and enabling near-linear-scaling evaluations while retaining high accuracy. These strategies are extended to all exchange-type contributions in F12 theory, leading to an alternative formulation of integrals better suited for efficient factorization. A novel combination of numerical quadrature with density fitting is presented for the efficient computation of products of four-center two-electron integrals, alongside algorithms covering all exchange-type contributions, including advantageous batching strategies. In general, the formal computational cost with system size M is reduced from O(M⁵) to O(M⁴), with further improvements appearing feasible through the use of localized molecular orbitals combined with advanced integral screening techniques.
Using a second-order approximation to Random Phase Approximation renormalized (RPAr) many-body perturbation theory for the interacting density-density response function, we have developed a so-called higher-order terms (HOT) approximation for the correlation energy. In combination with the first-order RPAr correction, our new method faithfully captures the infinite-order correlation for a given exchange-correlation kernel, yielding errors of the total correlation energy on the order of 1% or less for most systems. For exchange-like kernels, our new method has the further benefit that the coupling-strength integration can be completely eliminated resulting in a modest reduction in computational cost compared to the traditional approach. When the correlation energy is accurately reproduced by the HOT approximation, structural properties and energy differences are also accurately reproduced, as we demonstrate for several periodic solids and some molecular systems. Energy differences involving fragmentation are challenging for the HOT method, however, due to errors that may not cancel between a composite system and its constituent pieces.
We derive a renormalized many-body perturbation theory (MBPT) starting from the random phase approximation (RPA). This RPA-renormalized perturbation theory extends the scope of single-reference MBPT methods to small-gap systems without significantly increasing the computational cost. The leading correction to RPA, termed the approximate exchange kernel (AXK), substantially improves upon RPA atomization energies and ionization potentials without affecting other properties such as barrier heights where RPA is already accurate. Thus, AXK is more balanced than second-order screened exchange [A. Grüneis et al., J. Chem. Phys. 131, 154115 (2009)], which tends to overcorrect RPA for systems with stronger static correlation. Similarly, AXK avoids the divergence of second-order Møller-Plesset (MP2) theory for small gap systems and delivers a much more consistent performance than MP2 across the periodic table at comparable cost. RPA+AXK thus is an accurate, non-empirical, and robust tool to assess and improve semi-local density functional theory for a wide range of systems previously inaccessible to first-principles electronic structure calculations.
The well-known and well-used Density Functional Theory (DFT) is a variational theory for the ground-state. There are no guarantees that it delivers good results for excited state properties such as quasiparticle energies or wave function. All that is guaranteed is that the total energy is exact, the chemical potential is correct, and that the eective electronic states (Kohn-Sham states) generate the right electron density at the variational minimum. Luttinger-Ward approaches to electronic structure are attractive as beyond-DFT schemes since, in principle, they oer a variational functional of the one-electron Green’s function that yields both the correct total energy and all one-particle properties at the variational extremum. However, compared to DFT, these methods face a number of additional challenges beyond the familiar problem of choosing an exchange-correlation functional: (a) they have a much greater computational cost compared to DFT, (b) one must choose some set of trial Green’s functions, and (c) one must locate an extremum (not necessarily a minimum). This talk aims both to be pedagogical and to provide a summary of our most recent work in this area. We focus mainly on the random-phase approximation (RPA) for the correlation functional which yields the GW self-energy for the Green’s function. We will describe ecient approaches for computing the RPA correlation energy; how various types of approximations to the correlation energy create a ladder of approximate self-energy operators and what type of physics is included in these approximations (exchange, screening, Hubbard U, etc.); and how the extremization is in fact highly problematic. We will end with some ideas for making practical progress with the extremization and their relation to the successful Quasiparticle Self-consistent GW (QSGW) scheme [1].