G W+EDMFT method tackles both local and nonlocal quantum correlations

Researchers from the Flatiron Institute, alongside collaborators in Switzerland and France, have developed a new method called GW+EDMFT that simultaneously addresses both “local” and “nonlocal” quantum correlations within materials. This efficient, fully self-consistent implementation combines the GW method for nonlocal effects with extended dynamical mean-field theory (EDMFT) to treat strong local interactions, including a well-defined double-counting correction.

Applying this scheme to strontium manganese oxide and lanthanum nickel oxide, the team demonstrated stabilization of a Mott-insulating state and quasiparticle mass enhancement consistent with experimental results, establishing GW+EDMFT as a reliable framework for modeling strongly correlated quantum materials.

GW+EDMFT Addresses Local and Nonlocal Quantum Correlations

The computational cost of modeling strongly correlated quantum materials has been significantly reduced through a new implementation of the GW+EDMFT method, enabled by a technique for compressing two-particle correlation functions called interpolative separable density fitting (ISDF). This advancement allows for a simultaneous and self-consistent treatment of both “local” and “nonlocal” quantum correlations, a challenge that previously limited the efficiency of such calculations. Achieving self-consistency in both the Green’s function and the dynamically screened interaction is critical for accurately describing how electrons interact and move within these complex materials, and ISDF makes this possible.

The newly developed GW+EDMFT scheme includes a “well-defined double-counting correction” and a “self-consistently determined impurity interaction”, addressing common pitfalls in quantum material simulations. These features ensure a more rigorous and accurate approach by precisely accounting for electron-electron interactions and preventing overestimation of screening effects, a problem often encountered in simpler methods.

This level of detail is essential for accurately predicting material properties and understanding their behavior. The method’s efficiency stems from its ability to compress the vast amount of data required to describe two-particle correlations, making calculations feasible for realistic materials. Specifically, the self-consistent effective local interaction within GW+EDMFT corrects the “overscreening inherent to constrained-RPA approaches”, a significant improvement in predictive power. For SrMnO3, the method successfully stabilizes a Mott-insulating state, aligning with experimental observations, while for LaNiO3, it predicts a quasiparticle mass enhancement consistent with angle-resolved photoemission spectroscopy (ARPES) data.

These results confirm the reliability of GW+EDMFT as a first-principles framework for understanding strongly correlated quantum materials. The ability to accurately model SrMnO3 is particularly noteworthy, as the method’s predictions align “in quantitative agreement with experiment” and demonstrate the method’s capacity to capture the complex interplay of electron correlations that lead to insulating behavior in this material.

In LaNiO3, the local EDMFT corrections reveal enhanced local charge fluctuations, a subtle effect that is important for understanding the material’s metallic properties. The observation that these fluctuations are amplified by correlation effects highlights the importance of a fully self-consistent approach. Researchers at the Flatiron Institute, alongside collaborators at the University of Fribourg in Switzerland and Université Paris-Saclay in France, detailed that strong local correlations often exceed the capabilities of simpler, perturbative approaches.

These approaches tend to underestimate the strength of electron interactions in Mott insulators, leading to inaccurate predictions. The team found that the fully screened interaction in a Mott insulator, like SrMnO3, remains close to the local effective interaction due to strongly suppressed impurity polarization. This finding emphasises the importance of accurately capturing local correlations to understand the electronic structure of these materials.

The work also demonstrates that local vertex corrections reshape screening in a state-dependent manner, meaning that the way electrons shield each other varies depending on the material’s properties. This is particularly evident in the contrasting behavior of SrMnO3 and LaNiO3, where the method reveals distinct screening mechanisms. This observation provides valuable insight into the fundamental physics governing these materials. The development of fully self-consistent GW+EDMFT, facilitated by ISDF compression, establishes a “practical and reliable ab initio framework” for studying materials where nonlocal correlations beyond GW remain weak.

This opens up new avenues for investigating a wide range of quantum materials and understanding phenomena such as metal-insulator transitions and non-Fermi liquid behavior. The team’s success in accurately modeling both insulating and metallic materials demonstrates the versatility and robustness of this new approach. The method’s ability to accurately capture the interplay of screening, covalency, and strong correlations positions it as a powerful tool for exploring the complex world of strongly correlated quantum materials.

Ab Initio Quantum Embedding for Correlated Materials

The computational efficiency of modeling strongly correlated materials received a boost with the development of a method that simultaneously addresses both localized and delocalized quantum effects, a feat previously hindered by computational demands. Researchers successfully implemented a fully self-consistent GW+EDMFT scheme, using advances in two-particle correlation function compression to achieve a balance between accuracy and feasibility in describing complex electronic behavior. The newly refined approach tackles a long-standing challenge in condensed matter physics: accurately describing materials where electrons strongly interact with each other.

Traditional density functional theory often falls short in these scenarios, necessitating more sophisticated many-body formalisms. The team’s implementation ensures full self-consistency, meaning that the calculations account for how the different parts of the system influence each other throughout the process, a feature that enhances the reliability of the results.

GW+EDMFT Resolves Overscreening in SrMnO3 and LaNiO3

This approach overcomes limitations of previous methods by simultaneously addressing both localized and non-localized quantum interactions, a feat previously hampered by computational demands. Applying the scheme to the Mott insulator SrMnO3 and the correlated metal LaNiO3, the researchers show that the self-consistent effective local interaction resolves the overscreening inherent to constrained-RPA approaches.

A direct comparison revealed that simpler constrained-RPA approaches require the complete removal of screening within the nickel-d and oxygen-p orbitals to achieve similar interaction strengths, highlighting the GW+EDMFT method’s ability to accurately capture these effects without such drastic simplification. The success of this method stems from its ability to address a critical issue in modeling correlated materials: overscreening.

Traditional constrained-RPA methods often overestimate the screening of electron interactions, leading to inaccurate predictions, particularly for insulating materials. This is particularly evident in the calculated bosonic propagator for strontium manganese oxide, where the GW+EDMFT screened interaction diverges significantly from perturbative results, approaching an effective local interaction substantially larger than predicted by constrained-RPA. Analysis of the screened interaction in lanthanum nickel oxide revealed a smooth interpolation between high-energy behavior predicted by scGW and low-energy behavior from G0W0, with a crossover occurring between 5 and 10 electron volts.

Self-consistency weakens the frequency dependence of the screened interaction, a trend more pronounced in the metallic lanthanum nickel oxide. The researchers demonstrated that in the Mott insulating system, non-perturbative vertex corrections strongly suppress charge fluctuations, leading to a significantly larger and less structured effective local interaction than predicted by constrained-RPA.

Dynamical Screening and Double-Counting in GW+EDMFT

This method, built upon the GW approximation and extended dynamical mean-field theory (EDMFT), relies on a specific formulation within the Almbladh free-energy functional, where many-body effects are encoded in the correlation functional, distinguishing it from simpler approaches. The GW+EDMFT approach differs from standard calculations by incorporating diagrams across the entire B space, encompassing long-range screening and non-local correlations, while also accounting for interactions within the C space.

This distinction is important because it allows for a more accurate description of screening processes, particularly in materials where traditional methods overestimate the strength of interactions. Achieving full self-consistency corrects this overscreening, resulting in a more realistic depiction of spectral properties in Mott insulators. For lanthanum nickel oxide (LaNiO3), GW+EDMFT yielded a U(0) of 9-10 electron volts, slightly higher than those previously used in DFT+DMFT studies, but with a comparable J value of 0.7-1 electron volt.

The comparison with constrained-RPA calculations demonstrated that a closer agreement with U(0) is only achieved when screening within the nickel-d and oxygen-p manifold is removed, highlighting the importance of accurately capturing the complex interplay of electronic correlations. The work demonstrates that perturbative approaches systematically underestimate the screened interaction in Mott insulators, as strong local correlations extend beyond their reach. This framework, with its well-defined double-counting correction and self-consistently determined impurity interaction, represents a step forward in the field of materials science.

👉 More information
🗞 Electronic correlations and dynamical screening with ab initio quantum embedding
✍️ Chia-Nan Yeh, Francesco Petocchi, Alexander Hampel, Philipp Werner, Olivier Parcollet, Antoine Georges and Miguel A. Morales
🧠 DOI: http://link.aps.org/doi/10.1103/bp46-97l5

Stay current

See today’s quantum computing news on Quantum Zeitgeist for the latest breakthroughs in qubits, hardware, algorithms, and industry deals.

Avatar of Ivy Delaney

Ivy Delaney

Ivy Delaney has been working with neural networks and machine learning since the mid-nineties, back when a couple of hidden layers and a long afternoon of training counted as ambitious. She has watched the field go from academic curiosity to the thing quietly running underneath everything, and she brings that long view to quantum computing. For Quantum Zeitgeist she covers the ground where the two fields meet. That means quantum machine learning and the variational algorithms it leans on, and it also means the less glamorous but more interesting story of classical machine learning already doing real work inside quantum machines, decoding error-correcting codes, calibrating noisy hardware and learning the error models that simulators depend on. She writes about the hardware those algorithms have to run on too, and about the post-quantum cryptography scramble that the same hardware has set off. Her stories typically start with the paper, whether that is peer-reviewed work, conference proceedings or an arXiv preprint, with the source linked so you can hold a claim up against the research it came from. She is unimpressed by benchmarks that will not say what they beat, and by demonstrations that only work in the press release.

Latest Posts by Ivy Delaney: