Practical applications of fragment embedding and closely related local correlation methods critically depend on a judicious choice of a low-level theory to define the local embedding subspace and to capture long-range electrostatic and correlation effects outside the embedding region. Second-order Møller-Plesset perturbation theory (MP2) is by far the most widely used correlated low-level theory; however, its applicability becomes questionable in systems where MP2 is known to fail either quantitatively or qualitatively. In this work, we present the random phase approximation (RPA) as a promising alternative low-level theory to MP2 within the local natural orbital-based coupled-cluster (LNO-CC) framework. We demonstrate that RPA-based LNO-CC closely matches the performance of its MP2-based counterpart for systems with sizable energy gaps, while delivering significantly faster convergence toward the canonical coupled-cluster limit for metallic systems, particularly as the thermodynamic limit is approached. These results highlight the critical role of the low-level theory in fragment embedding and local correlation methods and identify RPA as a compelling alternative to the commonly used MP2.
We present in this contribution the first application of local correlation in the context of multicomponent methods. Multicomponent approaches allow for the targeted simulation of electrons together with other Fermions (most commonly protons) as quantum particles. These methods have become increasingly popular over the last years, particularly for the description of nuclear quantum effects (in strong hydrogen bonds, proton tunneling, and many more). However, most implementations are still based on canonical formulations of wave function theory, which we know for decades to be computationally inefficient for capturing dynamical correlation effects. Local correlation approaches, particularly with the use of pair natural orbitals (PNOs), enable asymptotically linear scaling of computational costs with very little impact on the overall accuracy. In this context, the efficient use of density fitting approximations in the integral calculation proves essential. We start by discussing our implementation of density-fitted NEO-MP2 and NEO-PNO-LMP2, upgrading the electronic correlation treatment up to PNO local coupled cluster level of theory. Several challenging examples are provided to benchmark the method in terms of accuracy as well as computational cost scaling. Following appropriate protocols, anharmonic corrections to localized X-H stretches can be applied routinely with little computational overhead.
In many cases, density-functional theory (DFT) with current standard approximate functionals offers a relatively accurate and computationally cheap description of the short-range dynamic electron correlation effects. However, in general, standard DFT does not treat the dispersion interaction effects adequately which, on the other hand, can be described by many-body perturbation theory (MBPT). It is therefore of interest to develop a hybrid model which combines the best of both the MBPT and DFT approaches. This can be achieved by splitting the two-electron interaction into long-range and short-range parts; the long-range part is then treated by MBPT and the short-range part by DFT. This work deals with the formulation of a general MBPT-DFT model (i.e., valid for any type of zeroth-order Hamiltonian) based on such a range separation. Applying the Rayleigh-Schr\"odinger formalism in this context, one finds that the generalized Bloch equation becomes self-consistent at each order of perturbation. This complication arises because the short-range part of the energy is a functional of the exact electron density, which is expanded in a perturbation series. In order to address this ``self-consistency problem'' and provide computable orbital-based expressions for any order of perturbation, a general one-electron reduced-density-matrix-based formalism is proposed. Two applications of our general formalism are presented: The derivation of a hybrid second-order M\o{}ller-Plesset-DFT model and the formulation of a range-separated optimized effective potential method based on a hybrid G\"orling-Levy-type MBPT-DFT model.
Treating electron correlation more accurately and efficiently is at the heart of the development of electronic structure methods. In the present work, we explore the use of stochastic approaches to evaluate high-order electron correlation energies, whose conventional computational scaling is unpleasantly steep, being O(Nn+3) with respect to the system size N and the perturbation order n for the Møller-Plesset (MP) series. To this end, starting from Goldstone’s time-dependent formulation of ab initio many-body perturbation theory (MBPT), we present a reformulation of MBPT, which naturally leads to a Monte Carlo scheme with O(nN2 + n2N + f(n)) scaling at each step, where f(n) is a function of n depending on the specific numerical scheme. Proof-of-concept calculations demonstrate that the proposed quantum Monte Carlo algorithm successfully extends the previous Monte Carlo approaches for MP2 and MP3 to higher orders by overcoming the factorial scaling problem. For the first time, Goldstone’s time-dependent formulation is made useful numerically for electron correlation energies, not only being purely as a theoretical tool.
Perhaps the most important approximations to the electronic structure problem in quantum chemistry are those based on coupled cluster and density functional theories. Coupled cluster theory has been called the ``gold standard'' of quantum chemistry due to the high accuracy that it achieves for weakly correlated systems. Kohn--Sham density functionals based on semilocal approximations are, without a doubt, the most widely used methods in chemistry and material science because of their high accuracy/cost ratio. The root of the success of coupled cluster and density functionals is their ability to efficiently describe the dynamic part of the electron correlation. However, both traditional coupled cluster and density functional approximations may fail catastrophically when substantial static correlation is present. This severely limits the applicability of these methods to a plethora of important chemical and physical problems such as, e.g., the description of bond breaking, transition states, transition metal-, lanthanide- and actinide-containing compounds, and superconductivity. In an attempt to tackle this problem, nonstandard (single-reference) coupled cluster-based techniques that aim to describe static correlation have been recently developed: pair coupled cluster doubles (pCCD) and singlet-paired coupled cluster doubles (CCD0). The ability to describe static correlation in pCCD and CCD0 comes, however, at the expense of important amounts of dynamic correlation so that the high accuracy of standard coupled cluster becomes unattainable. Thus, the reliable and efficient description of static and dynamic correlation in a simultaneous manner remains an open problem for quantum chemistry and many-body theory in general. In this thesis, different ways to combine pCCD and CCD0 with density functionals in order to describe static and dynamic correlation simultaneously (and efficiently) are explored. The combination of wavefunction and density functional methods has a long history in quantum chemistry (practical implementations have appeared in the literature since the 1970s). However, this kind of techniques have not achieved widespread use due to problems such as double counting of correlation and the symmetry dilemma---the fact that wavefunction methods respect the symmetries of Hamiltonian, while modern functionals are designed to work with broken symmetry densities. Here, particular mathematical features of pCCD and CCD0 are exploited to avoid these problems in an efficient manner. The two resulting families of approximations, denoted as pCCD+DFT and CCD0+DFT, are shown to be able to describe static and dynamic correlation in standard benchmark calculations. Furthermore, it is also shown that CCD0+DFT lends itself to combination with correlation from the direct random phase approximation (dRPA). Inclusion of dRPA in the long-range via the technique of range-separation allows for the description of dispersion correlation, the remaining part of the correlation. Thus, when combined with the dRPA, CCD0+DFT can account for all three-types of electron correlation that are necessary to accurately describe molecular systems. Lastly, applications of CCD0+DFT to actinide chemistry are considered in this work. The accuracy of CCD0+DFT for predicting equilibrium geometries and vibrational frequencies of actinide molecules and ions is assessed and compared to that of well-established quantum chemical methods. For this purpose, the
f0 actinyl series (UO
22+, NpO
23+, PuO
24+), the isoelectronic NUN, and Thorium (ThO, ThO
2+) and Nobelium (NoO, NoO
2) oxides are studied. It is shown that the CCD0+DFT description of these species agrees with available experimental data and is comparable with the results given by the highest-level calculations that are possible for such heavy compounds while being, at least, an order of magnitude lower in computational cost.
We present a formulation of the explicitly correlated second-order Møller-Plesset (MP2-F12) energy in which all nontrivial post-mean-field steps are formulated with linear computational complexity in system size. The two key ideas are the use of pair-natural orbitals for compact representation of wave function amplitudes and the use of domain approximation to impose the block sparsity. This development utilizes the concepts for sparse representation of tensors described in the context of the domain based local pair-natural orbital-MP2 (DLPNO-MP2) method by us recently [Pinski et al., J. Chem. Phys. 143, 034108 (2015)]. Novel developments reported here include the use of domains not only for the projected atomic orbitals, but also for the complementary auxiliary basis set (CABS) used to approximate the three- and four-electron integrals of the F12 theory, and a simplification of the standard B intermediate of the F12 theory that avoids computation of four-index two-electron integrals that involve two CABS indices. For quasi-1-dimensional systems (n-alkanes), the ON DLPNO-MP2-F12 method becomes less expensive than the conventional ON(5) MP2-F12 for n between 10 and 15, for double- and triple-zeta basis sets; for the largest alkane, C200H402, in def2-TZVP basis, the observed computational complexity is N(∼1.6), largely due to the cubic cost of computing the mean-field operators. The method reproduces the canonical MP2-F12 energy with high precision: 99.9% of the canonical correlation energy is recovered with the default truncation parameters. Although its cost is significantly higher than that of DLPNO-MP2 method, the cost increase is compensated by the great reduction of the basis set error due to explicit correlation.