Similarity-transformed Hamiltonian Research Articles

Extending the Perturbative Triples Correction in Coupled-Cluster Theory to Account for Inactive Orbitals.

Coupled-cluster singles and doubles calculations with perturbative triples [CCSD(T)] offer high accuracy with significant cost. The standard way to reduce the cost of a CCSD(T) calculation is by excluding molecular orbitals from the correlated calculation. While this speeds up the calculation, it also throws away the contribution from the inactive orbitals to the correlation energy. Here, we propose extending the standard CCSD(T) method to account for the effects of these inactive orbitals. This approach is based on a perturbation expansion of the similarity-transformed Hamiltonian, and the final method includes external singles and doubles corrections along with a semi-internal triples term. Compared to all-electron CCSD(T) calculations with the cc-pCVTZ basis set for a set of small molecules, we recovered, on average, 98% of the total correlation energy while using only 30% of the molecular orbitals. Using 72% of the molecular orbitals gave 99.5% of the correlation energy.

Growth Reduction of Similarity-Transformed Electronic Hamiltonians in Qubit Space.

Accurately solving the electronic structure problem through the variational quantum eigensolver (VQE) is hindered by the available quantum resources of current and near-term devices. One approach to relieving the circuit depth requirements for VQE is to "pre-process" the electronic Hamiltonian by a similarity transformation incorporating some degree of electronic correlation, with the remaining correlation left to be addressed by the circuit ansatz. This often comes at the price of a substantial increase in the number of terms to measure in the similarity-transformed Hamiltonian. In this work, we propose an efficient approach to sampling elements from the complete Pauli group for N qubits which minimizes the onset of new terms in the transformed Hamiltonian while facilitating substantial energy lowering. We benchmark the growth-mitigating generator selection technique for ground state energy estimations applied to models of the H4, N2, and H2O molecular systems. It is found that utilizing a selection procedure which obtains the growth-minimizing generator from the set of operators with the maximal energy gradient is the most competitive approach to reducing the onset of Hamiltonian terms while achieving systematic energy lowering of the reference state.

Fully self-consistent optimization of the Jastrow-Slater-type wave function using a similarity-transformed Hamiltonian

Fully self-consistent optimization of the Jastrow-Slater-type wave function using a similarity-transformed Hamiltonian

Time-dependent equation-of-motion coupled-cluster simulations with a defective Hamiltonian.

Simulations of laser-induced electron dynamics in a molecular system are performed using time-dependent (TD) equation-of-motion (EOM) coupled-cluster (CC) theory. The target system has been chosen to highlight potential shortcomings of truncated TD-EOM-CC methods [represented in this work by TD-EOM-CC with single and double excitations (TD-EOM-CCSD)], where unphysical spectroscopic features can emerge. Specifically, we explore driven resonant electronic excitations in magnesium fluoride in the proximity of an avoided crossing. Near the avoided crossing, the CCSD similarity-transformed Hamiltonian is defective, meaning that it has complex eigenvalues, and oscillator strengths may take on negative values. When an external field is applied to drive transitions to states exhibiting these traits, unphysical dynamics are observed. For example, the stationary states that make up the time-dependent state acquire populations that can be negative, exceed one, or even complex-valued.

Importance-sampling FCIQMC: Solving weak sign-problem systems.

We investigate the exact full configuration interaction quantum Monte Carlo algorithm (without the initiator approximation) applied to weak sign-problem fermionic systems, namely, systems in which the energy gap to the corresponding sign-free or "stoquastized" state is small. We show that the minimum number of walkers required to exactly overcome the signproblem can be significantly reduced via an importance-sampling similarity transformation even though the similarity-transformed Hamiltonian has the same stoquastic gap as the untransformed one. Furthermore, we show that in the off-half-filling Hubbard model at U/t = 8, the real-space (site) representation has a much weaker signproblem compared to the momentum space representation. By applying importance sampling using a Gutzwiller-like guiding wavefunction, we are able to substantially reduce the minimum number of walkers in the case of 2 × ℓ Hubbard ladders, enabling us to get exact energies for sizable ladders. With these results, we calculate the fundamental charge gap ΔEfund = E(N + 1) + E(N - 1) - 2E(N) for the ladder systems compared to strictly one-dimensional Hubbard chains and show that the ladder systems have a reduced fundamental gap compared to the 1D chains.

Theory and implementation of a novel stochastic approach to coupled cluster.

We present a detailed discussion of our novel diagrammatic coupled cluster Monte Carlo (diagCCMC) [Scott et al. J. Phys. Chem. Lett. 10, 925 (2019)]. The diagCCMC algorithm performs an imaginary-time propagation of the similarity-transformed coupled cluster Schrödinger equation. Imaginary-time updates are computed by the stochastic sampling of the coupled cluster vector function: each term is evaluated as a randomly realized diagram in the connected expansion of the similarity-transformed Hamiltonian. We highlight similarities and differences between deterministic and stochastic linked coupled cluster theory when the latter is re-expressed as a sampling of the diagrammatic expansion and discuss details of our implementation that allow for a walker-less realization of the stochastic sampling. Finally, we demonstrate that in the presence of locality, our algorithm can obtain a fixed errorbar per electron while only requiring an asymptotic computational effort that scales quartically with system size, independent of the truncation level in coupled cluster theory. The algorithm only requires an asymptotic memory cost scaling linearly, as demonstrated previously. These scaling reductions require no ad hoc modifications to the approach.

Hermitian second-order methods for excited electronic states: Unitary coupled cluster in comparison with algebraic-diagrammatic construction schemes.

Employing an intermediate state representation (ISR) approach, Hermitian second-order methods for the calculation of electronic excitation energies are presented and compared in detail. These comprise the algebraic-diagrammatic construction scheme for the polarization propagator, a hybrid second-order ISR scheme based on traditional coupled-cluster theory as well as two similar approaches based on a unitary coupled-cluster (UCC) ansatz. Although in a strict perturbation-theoretical framework all prove to be identical, differences emerge when the corresponding converged cluster amplitudes are used and depending on how the similarity-transformed UCC Hamiltonian is evaluated. The resulting excitation energies, however, do not significantly differ for systems well described by means of perturbation theory.

Similarity transformation of the electronic Schrödinger equation via Jastrow factorization

By expressing the electronic wavefunction in an explicitly correlated (Jastrow-factorized) form, a similarity-transformed effective Hamiltonian can be derived. The effective Hamiltonian is non-Hermitian and contains three-body interactions. The resulting ground-state eigenvalue problem can be solved projectively using a stochastic configuration-interaction formalism. Our approach permits the use of highly flexible Jastrow functions, which we show to be effective in achieving extremely high accuracy, even with small basis sets. Results are presented for the total energies and ionization potentials of the first-row atoms, achieving accuracy within a mH of the basis-set limit, using modest basis sets and computational effort.

Diagrammatic Coupled Cluster Monte Carlo.

We propose a modified coupled cluster Monte Carlo algorithm that stochastically samples connected terms within the truncated Baker-Campbell-Hausdorff expansion of the similarity-transformed Hamiltonian by construction of coupled cluster diagrams on the fly. Our new approach-diagCCMC-allows propagation to be performed using only the connected components of the similarity-transformed Hamiltonian, greatly reducing the memory cost associated with the stochastic solution of the coupled cluster equations. We show that for perfectly local, noninteracting systems diagCCMC is able to represent the coupled cluster wavefunction with a memory cost that scales linearly with system size. The favorable memory cost is observed with the only assumption of fixed stochastic granularity and is valid for arbitrary levels of coupled cluster theory. Significant reduction in memory cost is also shown to smoothly appear with dissociation of a finite chain of helium atoms. This approach is also shown not to break down in the presence of strong correlation through the example of a stretched nitrogen molecule. Our novel methodology moves the theoretical basis of coupled cluster Monte Carlo closer to deterministic approaches.

Similarity-transformed equation-of-motion vibrational coupled-cluster theory.

A similarity-transformed equation-of-motion vibrational coupled-cluster (STEOM-XVCC) method is introduced as a one-mode theory with an effective vibrational Hamiltonian, which is similarity transformed twice so that its lower-order operators are dressed with higher-order anharmonic effects. The first transformation uses an exponential excitation operator, defining the equation-of-motion vibrational coupled-cluster (EOM-XVCC) method, and the second uses an exponential excitation-deexcitation operator. From diagonalization of this doubly similarity-transformed Hamiltonian in the small one-mode excitation space, the method simultaneously computes accurate anharmonic vibrational frequencies of all fundamentals, which have unique significance in vibrational analyses. We establish a diagrammatic method of deriving the working equations of STEOM-XVCC and prove their connectedness and thus size-consistency as well as the exact equality of its frequencies with the corresponding roots of EOM-XVCC. We furthermore elucidate the similarities and differences between electronic and vibrational STEOM methods and between STEOM-XVCC and vibrational many-body Green's function theory based on the Dyson equation, which is also an anharmonic one-mode theory. The latter comparison inspires three approximate STEOM-XVCC methods utilizing the common approximations made in the Dyson equation: the diagonal approximation, a perturbative expansion of the Dyson self-energy, and the frequency-independent approximation. The STEOM-XVCC method including up to the simultaneous four-mode excitation operator in a quartic force field and its three approximate variants are formulated and implemented in computer codes with the aid of computer algebra, and they are applied to small test cases with varied degrees of anharmonicity.

Benchmark Applications of Variations of Multireference Equation of Motion Coupled-Cluster Theory.

In this work, several variations of the multireference equation of motion (MR-EOM) methodology are investigated for the calculation of excitation spectra. These variants of MR-EOM are characterized by the following aspects: (1) the operators included in the sequence of similarity transformations of the molecular electronic Hamiltonian, (2) whether permutational symmetries (i.e., hermitization, vertex symmetry) are imposed on the final elements of the similarity-transformed Hamiltonian, (3) the size of the manifold over which the similarity-transformed Hamiltonian is diagonalized, (4) whether the two-body cumulant is included in the expressions defining the amplitudes and the elements of the transformed Hamiltonian. The MR-EOM methods are benchmarked for the calculation of the excitation energies of a test set of organic molecules. With the availability of reliable benchmark data for this test set, it is possible to gauge the relative accuracy of these approaches. We also further examine a subset of the MR-EOM methods for the calculation of the excitation energies of some transition-metal complexes. These systems prove to be particularly difficult for single-reference coupled-cluster methods.

Similarity-transformed perturbation theory on top of truncated local coupled cluster solutions: Theory and applications to intermolecular interactions.

Your correspondents develop and apply fully nonorthogonal, local-reference perturbation theories describing non-covalent interactions. Our formulations are based on a Löwdin partitioning of the similarity-transformed Hamiltonian into a zeroth-order intramonomer piece (taking local CCSD solutions as its zeroth-order eigenfunction) plus a first-order piece coupling the fragments. If considerations are limited to a single molecule, the proposed intermolecular similarity-transformed perturbation theory represents a frozen-orbital variant of the "(2)"-type theories shown to be competitive with CCSD(T) and of similar cost if all terms are retained. Different restrictions on the zeroth- and first-order amplitudes are explored in the context of large-computation tractability and elucidation of non-local effects in the space of singles and doubles. To accurately approximate CCSD intermolecular interaction energies, a quadratically growing number of variables must be included at zeroth-order.

Equation-of-Motion Coupled-Cluster Theory for Excitation Energies of Closed-Shell Systems with Spin–Orbit Coupling

Excitation energies of closed-shell systems based on the equation-of-motion (EOM) coupled-cluster theory at the singles and doubles (CCSD) level with spin-orbit coupling (SOC) included in the post-Hartree-Fock treatment are implemented in the present work. SOC can be included in both the CC and EOM steps (EOM-SOC-CCSD) or only in the EOM part (SOC-EOM-CCSD). The latter approach is an economical way to account for SOC effects, but excitation energies with this approach are not size-intensive. When the unlinked term in the latter approach is neglected (cSOC-EOM-CCSD), size-intensive excitation energies can be obtained. Time-reversal symmetry and spatial symmetry are exploited to reduce the computational effort. Imposing time-reversal symmetry results in a real matrix representation for the similarity-transformed Hamiltonian, which facilitates the requirement of time-reversal symmetry for new trial vectors in Davidson's algorithm. Results on some closed-shell atoms and molecules containing heavy elements show that EOM-SOC-CCSD can provide excitation energies and spin-orbit splittings with reasonable accuracy. On the other hand, the SOC-EOM-CCSD approach is able to afford accurate estimates of SOC effects for valence electrons of systems containing elements up to the fifth row, while cSOC-EOM-CCSD is less accurate for spin-orbit splittings of transitions involving p1/2 spinors, even for Kr.

Optimization of the Jastrow factor using the random-phase approximation and a similarity-transformed Hamiltonian: Application to band-structure calculation for some semiconductors and insulators

Based on the random-phase approximation and the transcorrelated (TC) method, we optimize the Jastrow factor together with one-electron orbitals in the Slater determinant in the correlated wave function with a new scheme for periodic systems. The TC method is one of the promising wave function theories for first-principles electronic structure calculation, where the many-body wave function is approximated as a product of a Slater determinant and a Jastrow factor, and the Hamiltonian is similarity-transformed by the Jastrow factor. Using this similarity-transformed Hamiltonian, we can optimize the one-electron orbitals without evaluating 3N-dimensional integrations for the N-electron system. In contrast, optimization of the Jastrow factor within the framework of the TC method is computationally much more expensive and has not been performed for solid-state calculations before. In this study, we also benefit from the similarity-transformation in optimizing the Jastrow factor. Our optimization scheme is tested in applications to some solids from narrow-gap semiconductors to wide-gap insulators, and it is verified that the band gap of a wide-gap insulator and the lattice constants of some solids are improved by this optimization with reasonable computational cost.

Optimization of the Jastrow factor in the correlated wave function of electrons using the first-principles transcorrelated method for solid-state calculations

The transcorrelated (TC) method is one of the wave-function-based methods used for first-principles electronic structure calculations, and in terms of the computational cost is applicable to solid-state calculation. In this method, a many-body wave function of electrons is approximated as a product of the Jastrow factor and the Slater determinant, and the first-principles Hamiltonian is similarity-transformed by the Jastrow factor. The Schrödinger equation is rewritten as an eigenvalue problem for this similarity-transformed Hamiltonian, from which one obtains a self-consistent field (SCF) equation for optimizing one-electron orbitals in the Slater determinant at low computational cost. In contrast, optimization of the Jastrow factor is computationally much more expensive and has not been performed for solid-state calculation of the TC method before. In this study, we develop a new method for optimizing the Jastrow factor at a reasonable computational cost using the random-phase approximation (RPA) and pseudo-variance minimization. We apply this method to some simple solids, and find that the band gap of a wide-band-gap insulator is much improved by RPA.

On the approximation of the similarity-transformed Hamiltonian in single-reference and multireference coupled cluster theory

We consider the recursive single commutator (RSC) approximation of the Baker–Campbell–Hausdorff expansion introduced by Yanai and Chan [T. Yanai, G.K.-L. Chan, J. Chem. Phys. 124 (2006) 194106] and apply it in order to approximate the similarity transformation of the Hamiltonian in both traditional and unitary coupled cluster theory. The equilibrium bond distance, harmonic vibrational frequency, and anharmonic constant of H2, HF, N2, CuH, and Cu2 were computed using the coupled cluster approach with single and double excitations (CCSD) and CCSD with the RSC approximation of the similarity-transformed Hamiltonian (CCSD-RSC). Our results demonstrate that the RSC approximation introduces substantial errors in the estimates of molecular properties. The leading pejorative effects of the RSC approximation can be traced back to the imbalanced description of diagrams arising from the term 12H^,T^2,T^2. Following this analysis we consider a modified RSC scheme correct to fourth-order in the energy, which is found to reproduce CCSD results more closely. The RSC scheme is also applied in conjunction with the state-specific multireference coupled cluster approach of Mukherjee and co-workers [U.S. Mahapatra, B. Datta, D. Mukherjee, J. Chem. Phys. 110 (1999) 6171] to compute the potential energy curve of the BeH2 model, the vibrational frequencies of ozone, and the singlet–triplet splitting of p-benzyne. These examples show that the deterioration of the results caused by the RSC scheme is analogous to the one observed in the single-reference case. Implications for the formulation of approximate internally contracted multireference theories are discussed.

A noniterative perturbative triples correction for the spin-flipping and spin-conserving equation-of-motion coupled-cluster methods with single and double substitutions

A noniterative N (7) triples correction for the equation-of-motion coupled-cluster method with single and double substitutions (CCSD) is presented. The correction is derived by second-order perturbation treatment of the similarity-transformed CCSD Hamiltonian. The spin-conserving variant of the correction is identical to the triples correction of Piecuch and co-workers [Mol. Phys. 104, 2149 (2006)] derived within method-of-moments framework and is not size intensive. The spin-flip variant of the correction is size intensive. The performance of the correction is demonstrated by calculations of electronic excitation energies in methylene, nitrenium ion, cyclobutadiene, ortho-, meta-, and para-benzynes, 1,2,3-tridehydrobenzene, as well as C-C bond breaking in ethane. In all cases except cyclobutadiene, the absolute values of the correction for energy differences were 0.1 eV or less. In cyclobutadiene, the absolute values of the correction were as large as 0.4 eV. In most cases, the correction reduced the errors against the benchmark values by about a factor of 2-3, the absolute errors being less than 0.04 eV.

Coupled-cluster methods with perturbative inclusion of explicitly correlated terms: a preliminary investigation

We propose to account for the large basis-set error of a conventional coupled-cluster energy and wave function by a simple perturbative correction. The perturbation expansion is constructed by Löwdin partitioning of the similarity-transformed Hamiltonian in a space that includes explicitly correlated basis functions. To test this idea, we investigate the second-order explicitly correlated correction to the coupled-cluster singles and doubles (CCSD) energy, denoted here as the CCSD(2)(R12) method. The proposed perturbation expansion presents a systematic and easy-to-interpret picture of the "interference" between the basis-set and correlation hierarchies in the many-body electronic-structure theory. The leading-order term in the energy correction is the amplitude-independent R12 correction from the standard second-order Møller-Plesset R12 method. The cluster amplitudes appear in the higher-order terms and their effect is to decrease the basis-set correction, in accordance with the usual experience. In addition to the use of the standard R12 technology which simplifies all matrix elements to at most two-electron integrals, we propose several optional approximations to select only the most important terms in the energy correction. For a limited test set, the valence CCSD energies computed with the approximate method, termed , are on average precise to (1.9, 1.4, 0.5 and 0.1%) when computed with Dunning's aug-cc-pVXZ basis sets [X = (D, T, Q, 5)] accompanied by a single Slater-type correlation factor. This precision is a roughly an order of magnitude improvement over the standard CCSD method, whose respective average basis-set errors are (28.2, 10.6, 4.4 and 2.1%). Performance of the method is almost identical to that of the more complex iterative counterpart, CCSD(R12). The proposed approach to explicitly correlated coupled-cluster methods is technically appealing since no modification of the coupled-cluster equations is necessary and the standard Møller-Plesset R12 machinery can be reused.

Electronic Structure Calculations of Solids with a Similarity-Transformed Hamiltonian

A new wavefunction-based method of electronic band structure calculation of solids is proposed based on the transcorrelated (TC) method. In the TC method, a Jastrow–Slater-type correlated wavefunction is used as a trial function. Similarity transformation of the Hamiltonian with respect to the Jastrow correlation function is utilized to derive Hartree–Fock (HF)-like self-consistent-field (SCF) equations, which determine the one-body wavefunctions in the Slater determinant and their orbital energies. Since the electron correlation is taken into account, the method could be a practical tool for band structure calculation of solids where the HF method fails even qualitatively. The single-particle energy dispersion of the electron gas shows no singularity at the Fermi surface due to the screening effect incorporated in the TC method. It is also demonstrated that this method significantly reduces the band gaps of Si, C, and SiC when compared with their corresponding Hartree–Fock values.

Two new classes of non-iterative coupled-cluster methods derived from the method of moments of coupled-cluster equations

Two recently proposed classes of non-iterative coupled-cluster (CC) methods derived from the method of moments of CC equations (MMCC) are discussed. The first approach, termed MMCC/PT, combines the MMCC formalism with a simplified form of multi-reference perturbation theory. The second approach, which leads to completely renormalized (CR) CC methods employing the left eigenstates of the similarity-transformed Hamiltonian, such as CR-CCSD, exploits the recently developed biorthogonal formulation of the MMCC theory. Both approaches are capable of improving the results of standard CC and equation-of-motion CC (EOMCC) calculations for ground-state potential energy surfaces along bond breaking coordinates and excited states dominated by two-electron transitions with computer costs similar to those characterizing the popular (and failing) CCSD(T) approximation. The performance of the basic MMCC/PT and CR-CCSD approximations, in which non-iterative corrections due to triple excitations are added to the ground-state energies obtained with the CC singles and doubles (CCSD) approach, is illustrated by the results of calculations for the HF, F2 and H2O molecules. The improvements offered by these approaches in the excited-state calculations are illustrated by the results for the vertical excitation energies of the CH+ ion.