arXiv subjects

Luke Shulenburger

Publications and source records attributed to Luke Shulenburger.

At least 19 recordsLinked to original sources

arXiv2020

Towards a Systematic Improvement of the Fixed-Node Approximation in Diffusion Monte Carlo for Solids -- A Case Study In Diamond

While Diffusion Monte Carlo (DMC) is in principle an exact stochastic method for \textit{ab initio} electronic structure calculations, in practice the fermionic sign problem necessitates the use of the fixed-node approximation and trial wavefunctions with approximate nodes (or zeros) must be used. This approximation introduces a variational error in the energy that potentially can be tested and systematically improved. Here, we present a computational method that produces trial wavefunctions with systematically improvable nodes for DMC calculations of periodic solids. These trial wavefunctions are efficiently generated with the configuration interaction using a perturbative selection made iteratively (CIPSI) method. A simple protocol in which both exact and approximate results for finite supercells are used to extrapolate to the thermodynamic limit is introduced.

physics.chem-ph↗

arXiv2020

Systematic Comparison and Cross-validation of Fixed-Node Diffusion Monte Carlo and Phaseless Auxiliary-Field Quantum Monte Carlo in Solids

Quantum Monte Carlo (QMC) methods are some of the most accurate methods for simulating correlated electronic systems. We investigate the compatibility, strengths and weaknesses of two such methods, namely, diffusion Monte Carlo (DMC) and auxiliary-field quantum Monte Carlo (AFQMC). The multi-determinant trial wave functions employed in both approaches are generated using the configuration interaction using a perturbative selection made iteratively (CIPSI) technique. Complete basis set full configuration interaction (CBS-FCI) energies estimated with CIPSI are used as a reference in this comparative study between DMC and AFQMC. By focusing on a set of canonical finite size solid state systems, we show that both QMC methods can be made to systematically converge towards the same energy once basis set effects and systematic biases have been removed. AFQMC shows a much smaller dependence on the trial wavefunction than DMC while simultaneously exhibiting a much larger basis set dependence. We outline some of the remaining challenges and opportunities for improving these approaches.

physics.comp-ph↗

arXiv2019

A new generation of effective core potentials from correlated calculations: 4s and 4p main group elements and first row additions

Recently, we developed a new method for generating effective core potentials (ECPs) using valence energy isospectrality with explicitly correlated all-electron (AE) excitations and norm-conservation criteria. We apply this methodology to the 3$^{rd}$-row main group elements, creating new correlation consistent effective core potentials (ccECPs) and also derive additional ECPs to complete the ccECP table for H-Kr. For K and Ca, we develop Ne-core ECPs and for the $4p$ main group elements, we construct [Ar]$3d^{10}$-core potentials. Scalar relativistic effects are included in their construction. Our ccECPs reproduce AE spectra with significantly better accuracy than many existing pseudopotentials and show better overall consistency across multiple properties. The transferability of ccECPs is tested on monohydride and monoxide molecules over a range of molecular geometries. For the constructed ccECPs we also provide optimized DZ - 6Z valence Gaussian basis sets.

cond-mat.mtrl-sci↗

arXiv2018

New generation of effective core potentials from correlated calculations: 2nd row elements

Very recently, we have introduced correlation consistent effective core potentials (ccECPs) derived from many-body approaches with the main target being its use in explicitly correlated methods but also in mainstream approaches. The ccECPs are based on reproducing excitation energies for a subset of valence states, i.e., achieving a near-isospectrality between the original and pseudo Hamiltonians. In addition, binding curves of dimer molecules have been used for refinement and overall improvement of transferability over a range of bond lengths. Here we apply similar ideas to the second row elements and study several aspects of the constructions in order to find the optimal (or nearly-optimal) solutions within the chosen ECP forms with $3s,3p$ valence space (Ne-core). New constructions exhibit accurate low-lying atomic excitations and equilibrium molecular bonds (on average within $\approx$ $0.03$ eV and $3$ mÅ), however, the errors for Al and Si oxide molecules at short bond lengths are notably larger for both ours and existing ECPs. Assuming this limitation, our ccECPs show a systematic balance between the criteria of atomic spectra accuracy and transferability for molecular bonds. In order to provide another option with much higher uniform accuracy, we also construct He-core ECPs for the whole row with typical discrepancies of $\approx$ 0.01 eV or smaller.

cond-mat.mtrl-sci↗

arXiv2018

Towards a systematically improvable many-body description of antiferromagnetic iron oxide

We report variational and fixed-node diffusion quantum Monte Carlo (QMC) calculations of anti-ferromagnetic iron oxide (FeO) in the ground state B1 crystal structure. The goal of this study was a systematic investigation of the sensitivity of several ground state properties to a variety of QMC wave function generation techniques including advanced wave functions such as multi-determinant expansions and backflow transformations. We found that the predicted lattice distortion was largely controlled by the choice of single particle orbitals used to construct the wave function, rather than by subsequent wave function optimization techniques within QMC. However, the absolute magnetic moment was remarkably insensitive to the method of wave function construction. QMC estimates of total spin density indicate that in addition to strong electronic correlation of the Fe $3d$ states, charge transfer may be an important but challenging piece of physics to accurately capture within existing QMC methods. Finally, we highlight the need for advanced and systematically improvable many-body wave functions suitable for accurately describing challenging real systems.

cond-mat.mtrl-sci↗

arXiv2018

An efficient hybrid orbital representation for quantum Monte Carlo calculations

The scale and complexity of quantum system to which real-space quantum Monte Carlo (QMC) can be applied in part depends on the representation and memory usage of the trial wavefunction. B-splines, the computationally most efficient basis set, can have memory requirements exceeding the capacity of a single computational node. This situation has traditionally forced a difficult choice of either using slow internode communication or a potentially less accurate but smaller basis set such as Gaussians. Here, we introduce a hybrid representation of the single particle orbitals that combine a localized atomic basis set around atomic cores and B-splines in the interstitial regions to reduce the memory usage while retaining high speed of evaluation and either retaining or increasing overall accuracy. We present a benchmark calculation for NiO demonstrating a superior accuracy while using only one eighth the memory required for conventional B-splines. The hybrid orbital representation therefore expands the overall range of systems that can be practically studied with QMC.

cond-mat.mtrl-sci↗

arXiv2018

New generation of effective core potentials from correlated calculations: 3d transition metal series

Recently, we have introduced a new generation of effective core potentials (ECPs) designed for accurate correlated calculations but equally useful for a broad variety of approaches. The guiding principle has been the isospectrality of all-electron and ECP Hamiltonians for a subset of valence many-body states using correlated, nearly-exact calculations. Here we present such ECPs for the 3d transition series Sc to Zn with Ne-core, i.e, with semi-core 3s and 3p electrons in the valence space. Besides genuine many-body accuracy, the operators are simple, being represented by a few gaussians per symmetry channel with resulting potentials that are bounded everywhere. The transferability is checked on selected molecular systems over a range of geometries. The ECPs show a high overall accuracy with valence spectral discrepancies typically $\approx$ 0.01-0.02 eV or better. They also reproduce binding curves of hydride and oxide molecules typically within 0.02-0.03 eV deviations over the full non-dissociation range of interatomic distances.

cond-mat.mtrl-sci↗

arXiv2018

QMCPACK : An open source ab initio Quantum Monte Carlo package for the electronic structure of atoms, molecules, and solids

QMCPACK is an open source quantum Monte Carlo package for ab-initio electronic structure calculations. It supports calculations of metallic and insulating solids, molecules, atoms, and some model Hamiltonians. Implemented real space quantum Monte Carlo algorithms include variational, diffusion, and reptation Monte Carlo. QMCPACK uses Slater-Jastrow type trial wave functions in conjunction with a sophisticated optimizer capable of optimizing tens of thousands of parameters. The orbital space auxiliary field quantum Monte Carlo method is also implemented, enabling cross validation between different highly accurate methods. The code is specifically optimized for calculations with large numbers of electrons on the latest high performance computing architectures, including multicore central processing unit (CPU) and graphical processing unit (GPU) systems. We detail the program's capabilities, outline its structure, and give examples of its use in current research calculations. The package is available at http://www.qmcpack.org .

physics.comp-ph↗

arXiv2017

A New Generation of Effective Core Potentials for Correlated Calculations

We outline ideas on desired properties for a new generation of effective core potentials (ECPs) that will allow valence-only calculations to reach the full potential offered by recent advances in many-body wave function methods. The key improvements include consistent use of correlated methods throughout ECP constructions and improved transferability as required for an accurate description of molecular systems over a range of geometries. The guiding principle is the isospectrality of all-electron and ECP Hamiltonians for a subset of valence states. We illustrate these concepts on a few first- and second-row atoms (B, C, N, O, S) and we obtain higher accuracy in transferability than previous constructions while using a semi-local ECPs with a small number of parameters. In addition, the constructed ECPs enable many-body calculations of valence properties with higher (or same) accuracy than their all-electron counterparts with uncorrelated cores. This implies that the ECPs include also some of the impacts of core-core and core-valence correlations on valence properties. The results open further prospects for ECP improvements and refinements.

physics.chem-ph↗

arXiv2017

Embracing a new era of highly efficient and productive quantum Monte Carlo simulations

QMCPACK has enabled cutting-edge materials research on supercomputers for over a decade. It scales nearly ideally but has low single-node efficiency due to the physics-based abstractions using array-of-structures objects, causing inefficient vectorization. We present a systematic approach to transform QMCPACK to better exploit the new hardware features of modern CPUs in portable and maintainable ways. We develop miniapps for fast prototyping and optimizations. We implement new containers in structure-of-arrays data layout to facilitate vectorizations by the compilers. Further speedup and smaller memory-footprints are obtained by computing data on the fly with the vectorized routines and expanding single-precision use. All these are seamlessly incorporated in production QMCPACK. We demonstrate upto 4.5x speedups on recent Intel processors and IBM Blue Gene/Q for representative workloads. Energy consumption is reduced significantly commensurate to the speedup factor. Memory-footprints are reduced by up-to 3.8x, opening the possibility to solve much larger problems of future.

cs.DC↗

arXiv2016

Optimization and parallelization of B-spline based orbital evaluations in QMC on multi/many-core shared memory processors

B-spline based orbital representations are widely used in Quantum Monte Carlo (QMC) simulations of solids, historically taking as much as 50% of the total run time. Random accesses to a large four-dimensional array make it challenging to efficiently utilize caches and wide vector units of modern CPUs. We present node-level optimizations of B-spline evaluations on multi/many-core shared memory processors. To increase SIMD efficiency and bandwidth utilization, we first apply data layout transformation from array-of-structures to structure-of-arrays (SoA). Then by blocking SoA objects, we optimize cache reuse and get sustained throughput for a range of problem sizes. We implement efficient nested threading in B-spline orbital evaluation kernels, paving the way towards enabling strong scaling of QMC simulations. These optimizations are portable on four distinct cache-coherent architectures and result in up to 5.6x performance enhancements on Intel Xeon Phi processor 7250P (KNL), 5.7x on Intel Xeon Phi coprocessor 7120P, 10x on an Intel Xeon processor E5v4 CPU and 9.5x on BlueGene/Q processor. Our nested threading implementation shows nearly ideal parallel efficiency on KNL up to 16 threads. We employ roofline performance analysis to model the impacts of our optimizations. This work combined with our current efforts of optimizing other QMC kernels, result in greater than 4.5x speedup of miniQMC on KNL.

cs.DC↗

arXiv2016

Phase Stability of TiO$_2$ Polymorphs from Diffusion Quantum Monte Carlo

Titanium dioxide, TiO$_2$, has multiple applications in catalysis, energy conversion and memristive devices because of its electronic structure. Most of these applications utilize the naturally existing phases: rutile, anatase and brookite. Despite the simple form of TiO$_2$ and its wide uses, there is long-standing disagreement between theory and experiment on the energetic ordering of these phases that has never been resolved. We present the first analysis of phase stability at zero temperature using the highly accurate many-body fixed node diffusion Quantum Monte Carlo (QMC) method. We also include the effects of temperature by calculating the Helmholtz free energy including both internal energy and vibrational contributions from density functional perturbation theory based quasi harmonic phonon calculations. Our QMC calculations find that anatase is the most stable phase at zero temperature, consistent with many previous mean-field calculations. However, at elevated temperatures, rutile becomes the most stable phase. For all finite temperatures, brookite is always the least stable phase.

cond-mat.mtrl-sci↗

arXiv2016

Quantum Monte Carlo analysis of a charge ordered insulating antiferromagnet: the Ti$_4$O$_7$ Magnéli phase

The Magnéli phase Ti$_4$O$_7$ is an important transition metal oxide with a wide range of applications because of its interplay between charge, spin, and lattice degrees of freedom. At low temperatures, it has non-trivial magnetic states very close in energy, driven by electronic exchange and correlation interactions. We have examined three low-lying states, one ferromagnetic and two antiferromagnetic, and calculated their energies as well as Ti spin moment distributions using highly accurate Quantum Monte Carlo methods. We compare our results to those obtained from density functional theory-based methods that include approximate corrections for exchange and correlation. Our results confirm the nature of the states and their ordering in energy, as compared with density-functional theory methods. However, the energy differences and spin distributions differ. A detailed analysis suggests that non-local exchange-correlation functionals, in addition to other approximations such as LDA+U to account for correlations, are needed to simultaneously obtain better estimates for spin moments, distributions, energy differences and energy gaps.

cond-mat.mtrl-sci↗

arXiv2016

Benchmarking Adiabatic Quantum Optimization for Complex Network Analysis

We lay the foundation for a benchmarking methodology for assessing current and future quantum computers. We pose and begin addressing fundamental questions about how to fairly compare computational devices at vastly different stages of technological maturity. We critically evaluate and offer our own contributions to current quantum benchmarking efforts, in particular those involving adiabatic quantum computation and the Adiabatic Quantum Optimizers produced by D-Wave Systems, Inc. We find that the performance of D-Wave's Adiabatic Quantum Optimizers scales roughly on par with classical approaches for some hard combinatorial optimization problems; however, architectural limitations of D-Wave devices present a significant hurdle in evaluating real-world applications. In addition to identifying and isolating such limitations, we develop algorithmic tools for circumventing these limitations on future D-Wave devices, assuming they continue to grow and mature at an exponential rate for the next several years.

quant-ph↗

arXiv2016

X-Ray Thomson scattering without the Chihara decomposition

X-Ray Thomson Scattering (XRTS) is an important experimental technique used to measure the temperature, ionization state, structure, and density of warm dense matter (WDM). The fundamental property probed in these experiments is the electronic dynamic structure factor (DSF). In most models, this is decomposed into three terms [Chihara, J. Phys. F: Metal Phys. {\bf 17}, 295 (1987)] representing the response of tightly bound, loosely bound, and free electrons. Accompanying this decomposition is the classification of electrons as either bound or free, which is useful for gapped and cold systems but becomes increasingly questionable as temperatures and pressures increase into the WDM regime. In this work we provide unambiguous first principles calculations of the dynamic structure factor of warm dense beryllium, independent of the Chihara form, by treating bound and free states under a single formalism. The computational approach is real-time finite-temperature time-dependent density functional theory (TDDFT) being applied here for the first time to WDM. We compare results from TDDFT to Chihara-based calculations for experimentally relevant conditions in shock-compressed beryllium.

cond-mat.mtrl-sci↗

arXiv2015

The Nature of the Interlayer Interaction in Bulk and Few-Layer Phosphorus

An outstanding challenge of theoretical electronic structure is the description of van der Waals (vdW) interactions in molecules and solids. Renewed interest in resolving this is in part motivated by the technological promise of layered systems including graphite, transition metal dichalcogenides, and more recently, black phosphorus, in which the interlayer interaction is widely believed to be dominated by these types of forces. We report a series of quantum Monte Carlo (QMC) calculations for bulk black phosphorus and related few-layer phosphorene, which elucidate the nature of the forces that bind these systems and provide benchmark data for the energetics of these systems. We find a significant charge redistribution due to the interaction between electrons on adjacent layers. Comparison to density functional theory (DFT) calculations indicate not only wide variability even among different vdW corrected functionals, but the failure of these functionals to capture the trend of reorganization predicted by QMC. The delicate interplay of steric and dispersive forces between layers indicate that few-layer phosphorene presents an unexpected challenge for the development of vdW corrected DFT.

cond-mat.mtrl-sci↗

arXiv2015

Shock Response and Phase Transitions of MgO at Planetary Impact Conditions

The moon-forming impact and the subsequent evolution of the proto-Earth is strongly dependent on the properties of materials at the extreme conditions generated by this violent collision. We examine the high pressure behavior of MgO, one of the dominant constituents in the earth's mantle, using high-precision, plate impact shock compression experiments performed on Sandia National Laboratories Z-Machine and extensive quantum simulations using Density Functional Theory (DFT) and quantum Monte Carlo (QMC). The combined data span from ambient conditions to 1.2 TPa and 42,000 K, showing solid-solid and solid-liquid phase boundaries. Furthermore our results indicate under impact that the solid and liquid phases coexist for more than 100 GPa, pushing complete melting to pressures in excess of 600 GPa. The high pressure required for complete shock melting places a lower bound on the relative velocities required for the moon forming impact.

cond-mat.mtrl-sci↗

arXiv2015

Beyond chemical accuracy: The pseudopotential approximation in diffusion Monte Carlo calculations of the HCP to BCC phase transition in beryllium

Motivated by the disagreement between recent diffusion Monte Carlo calculations and experiments on the phase transition pressure between the ambient and beta-Sn phases of silicon, we present a study of the HCP to BCC phase transition in beryllium. This lighter element provides an oppor- tunity for directly testing many of the approximations required for calculations on silicon and may suggest a path towards increasing the practical accuracy of diffusion Monte Carlo calculations of solids in general. We demonstrate that the single largest approximation in these calculations is the pseudopotential approximation. After removing this we find excellent agreement with experiment for the ambient HCP phase and results similar to careful calculations using density functional theory for the phase transition pressure.

physics.chem-ph↗