Atomization Energies Of Molecules Research Articles

Spherical vs non-spherical and symmetry-preserving vs symmetry-breaking densities of open-shell atoms in density functional theory.

The atomization energies of molecules from first-principles density functional approximations improve from the local spin-density approximation to the Perdew-Burke-Ernzerhof (PBE) generalized gradient approximation (GGA) to the strongly constrained and appropriately normed (SCAN) meta-GGA, and their sensitivities to non-spherical components of the density increase in the same order. Thus, these functional advances increase density sensitivity and imitate the exact constrained search over correlated wavefunctions better than that over ensembles. The diatomic molecules studied here, singlet C2 and F2 plus triplet B2 and O2, have cylindrically symmetric densities. Because the densities of the corresponding atoms are non-spherical, the approximate Kohn-Sham potentials for the atoms have a lower symmetry than that of the external (nuclear) potential so that the non-interacting wavefunctions are not eigenstates of the square of total orbital angular momentum, breaking a symmetry that yields a feature of the exact ground-state density. That spatial symmetry can be preserved by a non-self-consistent approach in which a self-consistent equilibrium-ensemble calculation is followed by integer re-occupation of the Kohn-Sham orbitals as the first of several steps. The symmetry-preserving approach is different from symmetry restoration based on projection. First-step space- (and space-spin-) symmetry preservation in atoms is shown to have a small effect on the atomization energies of molecules, quantifying earlier observations by Fertig and Kohn. Thus, the standard Kohn-Sham way of calculating atomization energies, with self-consistent symmetry breaking to minimize the energy, is justified at least for the common cases where the molecules cannot break symmetry. Unless symmetry breaking is allowed in the molecule, SCAN strongly underestimates the atomization energy of strongly correlated singlet C2.

A hybrid quantum regression model for the prediction of molecular atomization energies

Quantum machine learning is a relatively new research field that aims to combine the dramatic performance advantage offered by quantum computing and the ability of machine learning algorithms to learn complex distributions of high-dimensional data. The primary focus of this domain is the implementation of classical machine learning algorithms in the quantum mechanical domain and study of the speedup due to quantum parallelism, which could enable the development of novel techniques for solving problems such as quantum phase recognition and quantum error correction optimization. In this paper, we propose a hybrid quantum machine learning pipeline for predicting the atomization energies of various molecules using the nuclear charges and atomic positions of the constituent atoms. Firstly, we will be using a deep convolutional auto-encoder model for the feature extraction of data constructed from the eigenvalues and eigenvector centralities of the pairwise distance matrix calculated from atomic positions and the unrolled upper triangle of each Coulomb matrix calculated from nuclear charges, and we will then be using a quantum regression algorithm such as quantum linear regression, quantum radial basis function neural network and, a quantum neural network for estimating the atomization energy. The hybrid quantum neural network models do not seem to provide any speedup over their classical counterparts. Before implementing a quantum algorithm, we will also be using state-of-the-art classical machine learning and deep learning models such as XGBoost, multilayer perceptron, deep convolutional neural network, and a long short-term memory network to study the correlation between the extracted features and corresponding atomization energies of molecules.

Erratum: Simple self-interaction correction to random-phase-approximation-like correlation energies [Phys. Rev. A 100, 022515 (2019)

The random phase approximation (RPA) is exact for the exchange energy of a many-electron ground state, but RPA makes the correlation energy too negative by about 0.5 eV/electron. That large short-range error, which tends to cancel out of iso-electronic energy differences, is largely corrected by an exchange-correlation kernel, or (as in RPA+) by an additive local or semilocal correction. RPA+ is by construction exact for the homogeneous electron gas, and it is also accurate for the jellium surface. RPA+ often gives realistic total energies for atoms or solids in which spin-polarization corrections are absent or small. RPA and RPA+ also yield realistic singlet binding energy curves for H2 and N2, and thus RPA+ yields correct total energies even for spin-unpolarized atoms with fractional spins and strong correlation, as in stretched H2 or N2. However, RPA and RPA+ can be very wrong for spin-polarized one-electron systems (especially for stretched H2+), and also for the spin-polarization energies of atoms. The spin-polarization energy is often a small part of the total energy of an atom, but important for ionization energies, electron affinities, and the atomization energies of molecules. Here we propose a computationally efficient generalized RPA+ (gRPA+) that changes RPA+ only for spin-polarized systems by making gRPA+ exact for all one-electron densities, in the same simple semilocal way that the correlation energy densities of many meta-generalized gradent approximations are made self-correlation free. By construction, gRPA+ does not degrade the exact RPA+ description of jellium. gRPA+ is found to greatly improve upon RPA and RPA+ for the ionization energies and electron affinities of light atoms. Many versions of RPA with an approximate exchange-correlation kernel fail to be exact for all one-electron densities, and they can also be self-interaction corrected in this way.

Simple self-interaction correction to random-phase-approximation-like correlation energies

The random-phase approximation (RPA) is exact for the exchange energy of a many-electron ground state, but RPA makes the correlation energy too negative by about 0.5 eV/electron. That large short-range error, which tends to cancel out of isoelectronic energy differences, is largely corrected by an exchange-correlation kernel, or (as in RPA+) by an additive local or semilocal correction. RPA+ is by construction exact for the homogeneous electron gas, and it is also accurate for the jellium surface. RPA+ often gives realistic total energies for atoms or solids in which spin-polarization corrections are absent or small. RPA and RPA+ also yield realistic singlet binding energy curves for ${\mathrm{H}}_{2}$ and ${\mathrm{N}}_{2}$, and thus RPA+ yields correct total energies even for spin-unpolarized atoms with fractional spins and strong correlation, as in stretched ${\mathrm{H}}_{2}$ or ${\mathrm{N}}_{2}$. However, RPA and RPA+ can be very wrong for spin-polarized one-electron systems (especially for stretched ${{\mathrm{H}}_{2}}^{+}$), and also for the spin-polarization energies of atoms. The spin-polarization energy is often a small part of the total energy of an atom, but important for ionization energies, electron affinities, and the atomization energies of molecules. Here we propose a computationally efficient generalized RPA+ (gRPA+) that changes RPA+ only for spin-polarized systems by making gRPA+ exact for all one-electron densities, in the same simple semilocal way that the correlation energy densities of many metageneralized gradient approximations are made self-correlation free. By construction, gRPA+ does not degrade the exact RPA+ description of jellium. gRPA+ is found to greatly improve upon RPA and RPA+ for the ionization energies and electron affinities of light atoms. Many versions of RPA with an approximate exchange-correlation kernel fail to be exact for all one-electron densities, and they can also be self-interaction corrected in this way.

Effect of Complex-Valued Optimal Orbitals on Atomization Energies with the Perdew-Zunger Self-Interaction Correction to Density Functional Theory.

The spurious interaction of an electron with itself-self-interaction error-is one of the biggest problems in modern density-functional theory. Some of its most glaring effects, such as qualitatively incorrect charge separation upon dissociation, can be removed by an approximate self-interaction correction due to Perdew and Zunger (PZ) (Perdew, J.; Zunger, A. Phys. Rev. B 1981, 23, 5048). However, the correction introduces an explicit dependence on the occupied orbital densities, which makes proper minimization of the functional difficult. Previous work (Vydrov et al., J. Chem. Phys. 2006, 124, 094108) has suggested that the application of the PZ correction results in worse atomization energies than those obtained with the uncorrected parent functional. But, it has only recently been found that the correct minimization of the PZ energy functional requires complex-valued orbitals, which have not been used in previous work on atomization energies. Here, we study the effect of the proper use of complex-valued orbitals on the atomization energies of molecules in the W4-11 data set (Karton, A.; Daon, S.; Martin, J. M. Chem. Phys. Lett. 2001, 510, 165). We find that the correction has a tendency to weaken the binding of molecules. The correction using complex-valued orbitals is invariably found to yield better atomization energies than the correction with real-valued orbitals. The correction applied to the PBEsol exchange-correlation functional results in a functional that is more accurate than the (uncorrected) PBE functional.

Construction of an optimal GGA functional for molecules and solids

The specialized exchange-correlation functionals of Wu and Cohen [Phys. Rev. B 73, 235116 (2006)] (WC) and Hammer et al. [Phys. Rev. B 59, 7413 (1999)] [revised Perdew, Burke, and Ernzerhof (RPBE)] yield good results for either the lattice constants of solids or the atomization energies of molecules, respectively, but are rather poor for the opposite quantity. By combining the WC and RPBE functionals, we construct a functional that performs equally well for both molecules and solids. Our proposed functional, which is still in the form of the simple generalized gradient approximation, can thus be applied efficiently to systems that involve both finite and infinite systems, a case that is crucial, for example, in heterogeneous catalysis studies. Therefore, the chemisorption of CO on transition-metal surfaces was considered, and it is shown that our functional gives improved results.

Improved hybrid functional for solids: The HSEsol functional

We introduce the hybrid functional HSEsol. It is based on PBEsol, a revised Perdew-Burke-Ernzerhof functional, designed to yield accurate equilibrium properties for solids and their surfaces. We present lattice constants, bulk moduli, atomization energies, heats of formation, and band gaps for extended systems, as well as atomization energies for the molecular G2-1 test set. Compared to HSE, significant improvements are found for lattice constants and atomization energies of solids, but atomization energies of molecules are slightly worse than for HSE. Additionally, we present zero-point anharmonic expansion corrections to the lattice constants and bulk moduli, evaluated from ab initio phonon calculations.

Systematic investigation of a family of gradient-dependent functionals for solids

Eleven density functionals are compared with regard to their performance for the lattice constants of solids. We consider standard functionals, such as the local-density approximation and the Perdew-Burke-Ernzerhof (PBE) generalized-gradient approximation (GGA), as well as variations of PBE GGA, such as PBEsol and similar functionals, PBE-type functionals employing a tighter Lieb-Oxford bound, and combinations thereof. Several of these variations are proposed here for the first time. On a test set of 60 solids we perform a system-by-system analysis for selected functionals and a full statistical analysis for all of them. The impact of restoring the gradient expansion and of tightening the Lieb-Oxford bound is discussed, and confronted with previous results obtained from other codes, functionals or test sets. No functional is uniformly good for all investigated systems, but surprisingly, and pleasingly, the simplest possible modifications to PBE turn out to have the most beneficial effect on its performance. The atomization energy of molecules was also considered and on a testing set of six molecules, we found that the PBE functional is clearly the best, the others leading to strong overbinding.

The relativistic Dirac–Coulomb–Fock effect on atomization energies

We have used a second-order perturbation treatment of the Dirac–Coulomb–Hartree–Fock method to estimate relativistic contributions to energies in the G2/97 test set. The one-electron relativistic effect on atomization energies of molecules containing first- and second-row atoms nearly always reduces binding. When the relativistic corrections are included in G3 theory and assessed on the G2/97 test set, there is little change in overall performance.

A comparison of the accuracy of different functionals

The errors in the computed geometries, zero-point energies, and atomization energies of molecules containing only first and second row atoms are compared for several levels of theory, including Hartree-Fock, second-order Møller-Plesset perturbation theory (MP2), and density functional theory (DFT) using five different functionals, including two hybrid functionals. The 6-31G ∗ and 6-311 + G(3df, 2p) basis sets are used. Overall, the B3LYP hybrid functional yields the best results.

Aspects of density functional theory in ab initio quantum chemistry: external correlation for free

The restricted Hartree-Fock method is subject to a first-order perturbation theory regularization of the electron repulsion integrals representation, to mimic the electron-electron correlation hole effect. The efficiency of this regularization is demonstrated for electron affinities and ionization energies of atoms, and on atomization energies of molecules. The accuracy of the method is comparable to that of modern density functional approaches as it provides significant improvement on standard Hartree-Fock results. Conceptual similarities with density functional theory are discussed, as are the implications for ab initio calculations on large systems.