arXiv subjects

Patrick Rinke

Publications and source records attributed to Patrick Rinke.

At least 55 records · Page 3Linked to original sources

arXiv2020

Projective Preferential Bayesian Optimization

Bayesian optimization is an effective method for finding extrema of a black-box function. We propose a new type of Bayesian optimization for learning user preferences in high-dimensional spaces. The central assumption is that the underlying objective function cannot be evaluated directly, but instead a minimizer along a projection can be queried, which we call a projective preferential query. The form of the query allows for feedback that is natural for a human to give, and which enables interaction. This is demonstrated in a user experiment in which the user feedback comes in the form of optimal position and orientation of a molecule adsorbing to a surface. We demonstrate that our framework is able to find a global minimum of a high-dimensional black-box function, which is an infeasible task for existing preferential Bayesian optimization frameworks that are based on pairwise comparisons.

stat.ML↗

arXiv2020

Real-Time Time-Dependent Density Functional Theory Implementation of Electronic Circular Dichroism Applied to Nanoscale Metal-Organic Clusters

Electronic circular dichroism (ECD) is a powerful spectroscopical method for investigating chiral properties at the molecular level. ECD calculations with the commonly used linear-response time-dependent density functional theory (LR-TDDFT) framework can be prohibitively costly for large systems. To alleviate this problem, we present here an ECD implementation for the projector augmented-wave method in the real-time-propagation TDDFT (RT-TDDFT) framework in the open-source GPAW code. Our implementation supports both local atomic basis set and real-space finite-difference representations of wave functions. We benchmark our implementation against an existing LR-TDDFT implementation in GPAW for small chiral molecules. We then demonstrate the efficiency of our local atomic basis set implementation for a large hybrid nanocluster.

physics.chem-ph↗

arXiv2020

Boron doping in gallium oxide from first principles

We study the feasibility of boron doping in gallium oxide ($\text{Ga}_2\text{O}_3$) for neutron detection. $\text{Ga}_2\text{O}_3$ is a wide band-gap, radiation hard material which has potential for neutron detection if it can be doped with a neutron active element. We investigate the boron-10 isotope as a possible neutron active dopant. Intrinsic and boron induced defects in $\text{Ga}_2\text{O}_3$ are studied with semi-local and hybrid density-functional-theory calculations. We find that in growth conditions favourable for boron, boron substitutional defects are likely to form making boron doping of $\text{Ga}_2\text{O}_3$ feasible.

cond-mat.mtrl-sci↗

arXiv2020

Efficient Cysteine Conformer Search with Bayesian Optimization

Finding low-energy molecular conformers is challenging due to the high dimensionality of the search space and the computational cost of accurate quantum chemical methods for determining conformer structures and energies. Here, we combine active-learning Bayesian optimization (BO) algorithms with quantum chemistry methods to address this challenge. Using cysteine as an example, we show that our procedure is both efficient and accurate. After only one thousand single-point calculations and approximately thirty structure relaxations, which is less than 10% computational cost of the current fastest method, we have found the low-energy conformers in good agreement with experimental measurements and reference calculations.

physics.comp-ph↗

arXiv2020

A balanced treatment of static and dynamic correlation in free- and Mg-porphyrin

We present an ab-initio dynamical configuration interaction (DCI) study of free- and Mg-porphyrin. DCI is a recently developed active space theory based on the Löwdin downfolding technique. In the active space, static correlation is described exactly with full configuration interaction. In the high energy, dynamically correlated subspace, we treat correlation at the quasiparticle level in the $GW$ approximation of Green's function theory. The final theory combines wave function and Green's function methods to give a balanced description of static and dynamic correlation. The theory and algorithm give a multireference treatment of ground and excited states for low computational cost. The four orbital Gouterman model of porphyrin offers an ideal active space in a large, correlated system to test the cost and accuracy of the embedding for large systems. Our parameter free, fully ab-initio DCI calculations in the minimal four-level active space agree well with both experiment and more expensive benchmark theories for the $Q_x$ and $Q_y$ transitions of free- and Mg-porphyrin. Studying the convergence of the excitation energies suggests that DCI correctly captures size extensive correlation effects, making it a promising active space theory for large, strongly-correlated systems.

cond-mat.mtrl-sci↗

arXiv2020

Relativistic correction scheme for core-level binding energies from $GW$

We present a relativistic correction scheme to improve the accuracy of 1s core-level binding energies calculated from Green's function theory in the $GW$ approximation, which does not add computational overhead. An element-specific corrective term is derived as the difference between the 1s eigenvalues obtained from the self-consistent solutions to the non- or scalar-relativistic Kohn-Sham equations and the four-component Dirac-Kohn-Sham equations for a free neutral atom. We examine the dependence of this corrective term on the molecular environment and on the amount of exact exchange in hybrid exchange-correlation functionals. This corrective term is then added as a perturbation to the quasiparticle energies from partially self-consistent and single-shot $GW$ calculations. We show that this element-specific relativistic correction, when applied to a previously reported benchmark set of 65 core-state excitations [J. Phys. Chem. Lett. 11, 1840 (2020)], reduces the mean absolute error (MAE) with respect to experiment from 0.55 to 0.30 eV and eliminates the species dependence of the MAE, which otherwise increases with the atomic number. The relativistic corrections also reduce the species dependence for the optimal amount of exact exchange in the hybrid functional used as starting point for the single-shot $G_0W_0$ calculations. Our correction scheme can be transferred to other methods, which we demonstrate for the Delta self-consistent field ($Δ$SCF) approach based on density functional theory.

physics.chem-ph↗

arXiv2020

Efficient hyperparameter tuning for kernel ridge regression with Bayesian optimization

Machine learning methods usually depend on internal parameters -- so called hyperparameters -- that need to be optimized for best performance. Such optimization poses a burden on machine learning practitioners, requiring expert knowledge, intuition or computationally demanding brute-force parameter searches. We here address the need for more efficient, automated hyperparameter selection with Bayesian optimization. We apply this technique to the kernel ridge regression machine learning method for two different descriptors for the atomic structure of organic molecules, one of which introduces its own set of hyperparameters to the method. We identify optimal hyperparameter configurations and infer entire prediction error landscapes in hyperparameter space, that serve as visual guides for the hyperparameter dependence. We further demonstrate that for an increasing number of hyperparameters, Bayesian optimization becomes significantly more efficient in computational time than an exhaustive grid search -- the current default standard hyperparameter search method -- while delivering an equivalent or even better accuracy.

physics.chem-ph↗

arXiv2020

Atomic structures and orbital energies of 61,489 crystal-forming organic molecules

Data science and machine learning in materials science require large datasets of technologically relevant molecules or materials. Currently, publicly available molecular datasets with realistic molecular geometries and spectral properties are rare. We here supply a diverse benchmark spectroscopy dataset of 61,489 molecules extracted from organic crystals in the Cambridge Structural Database (CSD), denoted OE62. Molecular equilibrium geometries are reported at the Perdew-Burke-Ernzerhof (PBE) level of density functional theory (DFT) including van der Waals corrections for all 62k molecules. For these geometries, OE62 supplies total energies and orbital eigenvalues at the PBE and the PBE hybrid (PBE0) functional level of DFT for all 62k molecules in vacuum as well as at the PBE0 level for a subset of 30,876 molecules in (implicit) water. For 5,239 molecules in vacuum, the dataset provides quasiparticle energies computed with many-body perturbation theory in the $G_0W_0$ approximation with a PBE0 starting point (denoted GW5000 in analogy to the GW100 benchmark set (M. van Setten et al. J. Chem. Theory Comput. 12, 5076 (2016))).

physics.comp-ph↗

arXiv2019

The GW compendium: A practical guide to theoretical photoemission spectroscopy

The GW approximation in electronic structure theory has become a widespread tool for predicting electronic excitations in chemical compounds and materials. In the realm of theoretical spectroscopy, the GW method provides access to charged excitations as measured in direct or inverse photoemission spectroscopy. The number of GW calculations in the past two decades has exploded with increased computing power and modern codes. The success of GW can be attributed to many factors: favorable scaling with respect to system size, a formal interpretation for charged excitation energies, the importance of dynamical screening in real systems, and its practical combination with other theories. In this review, we provide an overview of these formal and practical considerations. We expand, in detail, on the choices presented to the scientist performing GW calculations for the first time. We also give an introduction to the many-body theory behind GW, a review of modern applications like molecules and surfaces, and a perspective on methods which go beyond conventional GW calculations. This review addresses chemists, physicists and material scientists with an interest in theoretical spectroscopy. It is intended for newcomers to GW calculations but can also serve as an alternative perspective for experts and an up-to-date source of computational techniques.

cond-mat.mtrl-sci↗

arXiv2019

A quantum embedding theory in the screened Coulomb interaction: Combining configuration interaction with GW/BSE

We present a new quantum embedding theory called dynamical configuration interaction (DCI) that combines wave function and Green's function theories. DCI captures static correlation in a correlated subspace with configuration interaction and couples to high-energy, dynamic correlation outside the subspace with many-body perturbation theory based on Green's functions. In the correlated subspace, we use a wave function description to avoid embedding the two-particle vertex, which greatly simplifies the frequency structure of the embedding. DCI takes the strengths of both theories to balance static and dynamic correlation in a single, fully ab-initio embedding concept. We show that treating high-energy correlation up to the $GW$ and Bethe-Salpeter equation level is sufficient even for challenging multi-reference problems. Our theory treats ground and excited states on equal footing, and we compute the dissociation curve of N$_2$, vertical excitation energies of N$_2$ and C$_2$, and the ionization spectrum of benzene in excellent agreement with high level quantum chemistry methods and experiment.

physics.chem-ph↗

arXiv2019

Accurate absolute and relative core-level binding energies from $GW$

We present an accurate approach to compute X-ray photoelectron spectra based on the $GW$ Green's function method, that overcomes shortcomings of common density functional theory approaches. $GW$ has become a popular tool to compute valence excitations for a wide range of materials. However, core-level spectroscopy is thus far almost uncharted in $GW$. We show that single-shot perturbation calculations in the $G_0W_0$ approximation, which are routinely used for valence states, cannot be applied for core levels and suffer from an extreme, erroneous transfer of spectral weight to the satellite spectrum. The correct behavior can be restored by partial self-consistent $GW$ schemes or by using hybrid functionals with almost 50% of exact exchange as starting point for $G_0W_0$. We include also relativistic corrections and present a benchmark study for 65 molecular 1s excitations. Our absolute and relative $GW$ core-level binding energies agree within 0.3 and 0.2 eV with experiment, respectively.

physics.chem-ph↗

arXiv2019

Data-driven materials science: status, challenges and perspectives

Data-driven science is heralded as a new paradigm in materials science. In this field, data is the new resource, and knowledge is extracted from materials data sets that are too big or complex for traditional human reasoning - typically with the intent to discover new or improved materials or materials phenomena. Multiple factors, including the open science movement, national funding, and progress in information technology, have fueled its development. Such related tools as materials databases, machine learning, and high-throughput methods are now established as parts of the materials research toolset. However, there are a variety of challenges that impede progress in data-driven materials science: data veracity, integration of experimental and computational data, data longevity, standardization, and the gap between industrial interests and academic efforts. In this perspective article, we discuss the historical development and current state of data-driven materials science, building from the early evolution of open science to the rapid expansion of materials data infrastructures. We also review key successes and challenges so far, providing a perspective on the future development of the field.

physics.comp-ph↗

arXiv2019

Formation of Graphene atop a Si adlayer on the C-face of SiC

The structure of the SiC(000-1) surface, the C-face of the {0001} SiC surfaces, is studied as a function of temperature and of pressure in a gaseous environment of disilane (Si2H6). Various surface reconstructions are observed, both with and without the presence of an overlying graphene layer (which spontaneously forms at sufficiently high temperatures). Based on cross-sectional scanning transmission electron microscopy measurements, the interface structure that forms in the presence of the graphene is found to contain 1.4 - 1.7 monolayers (ML) of Si, a somewhat counter-intuitive result since, when the graphene forms, the system is actually under C-rich conditions. Using ab initio thermodynamics, it is demonstrated that there exists a class of Si-rich surfaces containing about 1.3 ML of Si that are stable on the surface (even under C-rich conditions) at temperatures above about 400 K. The structures that thus form consist of Si adatoms atop a Si adlayer on the C-face of SiC, with or without the presence of overlying graphene.

cond-mat.mes-hall↗

arXiv2019

DScribe: Library of Descriptors for Machine Learning in Materials Science

DScribe is a software package for machine learning that provides popular feature transformations ("descriptors") for atomistic materials simulations. DScribe accelerates the application of machine learning for atomistic property prediction by providing user-friendly, off-the-shelf descriptor implementations. The package currently contains implementations for Coulomb matrix, Ewald sum matrix, sine matrix, Many-body Tensor Representation (MBTR), Atom-centered Symmetry Function (ACSF) and Smooth Overlap of Atomic Positions (SOAP). Usage of the package is illustrated for two different applications: formation energy prediction for solids and ionic charge prediction for atoms in organic molecules. The package is freely available under the open-source Apache License 2.0.

cond-mat.mtrl-sci↗

arXiv2019

Chemical diversity in molecular orbital energy predictions with kernel ridge regression

Instant machine learning predictions of molecular properties are desirable for materials design, but the predictive power of the methodology is mainly tested on well-known benchmark datasets. Here, we investigate the performance of machine learning with kernel ridge regression (KRR) for the prediction of molecular orbital energies on three large datasets: the standard QM9 small organic molecules set, amino acid and dipeptide conformers, and organic crystal-forming molecules extracted from the Cambridge Structural Database. We focus on prediction of highest occupied molecular orbital (HOMO) energies, computed at density-functional level of theory. Two different representations that encode molecular structure are compared: the Coulomb matrix (CM) and the many-body tensor representation (MBTR). We find that KRR performance depends significantly on the chemistry of the underlying dataset and that the MBTR is superior to the CM, predicting HOMO energies with a mean absolute error as low as 0.09 eV. To demonstrate the power of our machine learning method, we apply our model to structures of 10k previously unseen molecules. We gain instant energy predictions that allow us to identify interesting molecules for future applications.

physics.chem-ph↗

arXiv2019

Efficient Bayesian Inference of Atomistic Structure in Complex Functional Materials

Tailoring the functional properties of advanced organic/inorganic heterogeonous devices to their intended technological applications requires knowledge and control of the microscopic structure inside the device. Atomistic quantum mechanical simulation methods deliver accurate energies and properties for individual configurations, however, finding the most favourable configurations remains computationally prohibitive. We propose a 'building block'-based Bayesian Optimisation Structure Search (BOSS) approach for addressing extended organic/inorganic interface problems and demonstrate its feasibility in a molecular surface adsorption study. In BOSS, a likelihood-free Bayesian scheme accelerates the identification of material energy landscapes with the number of sampled configurations during active learning, enabling structural inference with high chemical accuracy and featuring large simulation cells. This allowed us to identify several most favourable molecular adsorption configurations for $\mathrm{C}_{60}$ on the (101) surface of $\mathrm{TiO}_2$ anatase and clarify the key molecule-surface interactions governing structural assembly. Inferred structures were in good agreement with detailed experimental images of this surface adsorbate, demonstrating good predictive power of BOSS and opening the route towards large-scale surface adsorption studies of molecular aggregates and films.

cond-mat.mtrl-sci↗

arXiv2019

Database-driven high-throughput study for hybrid perovskite coating materials

We developed a high-throughput screening scheme to acquire candidate coating materials for hybrid perovskites. From more than 1.8 million entries of an inorganic compound database, we collected 93 binary and ternary materials with promising properties for protectively coating halide-perovskite photoabsorbers in perovskite solar cells. These candidates fulfill a series of criteria, including wide band gaps, abundant and non-toxic elements, water-insoluble, and small lattice mismatch with surface models of halide perovskites.

physics.app-ph↗

arXiv2018

Dynamical configuration interaction: Quantum embedding that combines wave functions and Green's functions

We present the concept, derivation, and implementation of dynamical configuration interaction, a quantum embedding theory that combines Green's function methodology with the many-body wave function. In a strongly-correlated active space, we use full configuration interaction (CI) to describe static correlation exactly. We add energy dependent corrections to the CI Hamiltonian which, in principle, include all remaining correlation derived from the bath space surrounding the active space. Next, we replace the exact Hamiltonian in the bath with one of excitations defined over a correlated ground state. This transformation is naturally suited to the methodology of many-body Green's functions. In this space, we use a modified $GW$/Bethe-Salpeter equation procedure to calculate excitation energies. Combined with an estimate of the ground state energy in the bath, we can efficiently compute the energy dependent corrections, which correlate the full set of orbitals, for very low computational cost. We present dimer dissociation curves for H$_2$ and N$_2$ in good agreement with exact results. Additionally, excited states of N$_2$ and C$_2$ are in excellent agreement with benchmark theory and experiment. By combining the strengths of two disciplines, we achieve a balanced description of static and dynamic correlation in a fully ab-initio, systematically improvable framework.

physics.chem-ph↗