@inproceedings{14107, abstract = {Amodal perception requires inferring the full shape of an object that is partially occluded. This task is particularly challenging on two levels: (1) it requires more information than what is contained in the instant retina or imaging sensor, (2) it is difficult to obtain enough well-annotated amodal labels for supervision. To this end, this paper develops a new framework of Self-supervised amodal Video object segmentation (SaVos). Our method efficiently leverages the visual information of video temporal sequences to infer the amodal mask of objects. The key intuition is that the occluded part of an object can be explained away if that part is visible in other frames, possibly deformed as long as the deformation can be reasonably learned. Accordingly, we derive a novel self-supervised learning paradigm that efficiently utilizes the visible object parts as the supervision to guide the training on videos. In addition to learning type prior to complete masks for known types, SaVos also learns the spatiotemporal prior, which is also useful for the amodal task and could generalize to unseen types. The proposed framework achieves the state-of-the-art performance on the synthetic amodal segmentation benchmark FISHBOWL and the real world benchmark KINS-Video-Car. Further, it lends itself well to being transferred to novel distributions using test-time adaptation, outperforming existing models even after the transfer to a new distribution.}, author = {Yao, Jian and Hong, Yuxin and Wang, Chiyu and Xiao, Tianjun and He, Tong and Locatello, Francesco and Wipf, David and Fu, Yanwei and Zhang, Zheng}, booktitle = {36th Conference on Neural Information Processing Systems}, location = {New Orleans, LA, United States}, title = {{Self-supervised amodal video object segmentation}}, doi = {10.48550/arXiv.2210.12733}, year = {2022}, } @inproceedings{14114, abstract = {Algorithmic fairness is frequently motivated in terms of a trade-off in which overall performance is decreased so as to improve performance on disadvantaged groups where the algorithm would otherwise be less accurate. Contrary to this, we find that applying existing fairness approaches to computer vision improve fairness by degrading the performance of classifiers across all groups (with increased degradation on the best performing groups). Extending the bias-variance decomposition for classification to fairness, we theoretically explain why the majority of fairness methods designed for low capacity models should not be used in settings involving high-capacity models, a scenario common to computer vision. We corroborate this analysis with extensive experimental support that shows that many of the fairness heuristics used in computer vision also degrade performance on the most disadvantaged groups. Building on these insights, we propose an adaptive augmentation strategy that, uniquely, of all methods tested, improves performance for the disadvantaged groups.}, author = {Zietlow, Dominik and Lohaus, Michael and Balakrishnan, Guha and Kleindessner, Matthaus and Locatello, Francesco and Scholkopf, Bernhard and Russell, Chris}, booktitle = {2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition}, isbn = {9781665469470}, issn = {2575-7075}, location = {New Orleans, LA, United States}, pages = {10400--10411}, publisher = {Institute of Electrical and Electronics Engineers}, title = {{Leveling down in computer vision: Pareto inefficiencies in fair deep classifiers}}, doi = {10.1109/cvpr52688.2022.01016}, year = {2022}, } @inproceedings{14168, abstract = {Recent work has seen the development of general purpose neural architectures that can be trained to perform tasks across diverse data modalities. General purpose models typically make few assumptions about the underlying data-structure and are known to perform well in the large-data regime. At the same time, there has been growing interest in modular neural architectures that represent the data using sparsely interacting modules. These models can be more robust out-of-distribution, computationally efficient, and capable of sample-efficient adaptation to new data. However, they tend to make domain-specific assumptions about the data, and present challenges in how module behavior (i.e., parameterization) and connectivity (i.e., their layout) can be jointly learned. In this work, we introduce a general purpose, yet modular neural architecture called Neural Attentive Circuits (NACs) that jointly learns the parameterization and a sparse connectivity of neural modules without using domain knowledge. NACs are best understood as the combination of two systems that are jointly trained end-to-end: one that determines the module configuration and the other that executes it on an input. We demonstrate qualitatively that NACs learn diverse and meaningful module configurations on the NLVR2 dataset without additional supervision. Quantitatively, we show that by incorporating modularity in this way, NACs improve upon a strong non-modular baseline in terms of low-shot adaptation on CIFAR and CUBs dataset by about 10%, and OOD robustness on Tiny ImageNet-R by about 2.5%. Further, we find that NACs can achieve an 8x speedup at inference time while losing less than 3% performance. Finally, we find NACs to yield competitive results on diverse data modalities spanning point-cloud classification, symbolic processing and text-classification from ASCII bytes, thereby confirming its general purpose nature.}, author = {Rahaman, Nasim and Weiss, Martin and Locatello, Francesco and Pal, Chris and Bengio, Yoshua and Schölkopf, Bernhard and Li, Li Erran and Ballas, Nicolas}, booktitle = {36th Conference on Neural Information Processing Systems}, location = {New Orleans, United States}, title = {{Neural attentive circuits}}, volume = {35}, year = {2022}, } @inproceedings{14170, abstract = {The idea behind object-centric representation learning is that natural scenes can better be modeled as compositions of objects and their relations as opposed to distributed representations. This inductive bias can be injected into neural networks to potentially improve systematic generalization and performance of downstream tasks in scenes with multiple objects. In this paper, we train state-of-the-art unsupervised models on five common multi-object datasets and evaluate segmentation metrics and downstream object property prediction. In addition, we study generalization and robustness by investigating the settings where either a single object is out of distribution -- e.g., having an unseen color, texture, or shape -- or global properties of the scene are altered -- e.g., by occlusions, cropping, or increasing the number of objects. From our experimental study, we find object-centric representations to be useful for downstream tasks and generally robust to most distribution shifts affecting objects. However, when the distribution shift affects the input in a less structured manner, robustness in terms of segmentation and downstream task performance may vary significantly across models and distribution shifts. }, author = {Dittadi, Andrea and Papa, Samuele and Vita, Michele De and Schölkopf, Bernhard and Winther, Ole and Locatello, Francesco}, booktitle = {Proceedings of the 39th International Conference on Machine Learning}, location = {Baltimore, MD, United States}, pages = {5221--5285}, publisher = {ML Research Press}, title = {{Generalization and robustness implications in object-centric learning}}, volume = {2022}, year = {2022}, } @inproceedings{14171, abstract = {This paper demonstrates how to recover causal graphs from the score of the data distribution in non-linear additive (Gaussian) noise models. Using score matching algorithms as a building block, we show how to design a new generation of scalable causal discovery methods. To showcase our approach, we also propose a new efficient method for approximating the score's Jacobian, enabling to recover the causal graph. Empirically, we find that the new algorithm, called SCORE, is competitive with state-of-the-art causal discovery methods while being significantly faster.}, author = {Rolland, Paul and Cevher, Volkan and Kleindessner, Matthäus and Russel, Chris and Schölkopf, Bernhard and Janzing, Dominik and Locatello, Francesco}, booktitle = {Proceedings of the 39th International Conference on Machine Learning}, location = {Baltimore, MD, United States}, pages = {18741--18753}, publisher = {ML Research Press}, title = {{Score matching enables causal discovery of nonlinear additive noise models}}, volume = {162}, year = {2022}, } @inproceedings{14172, abstract = {An important component for generalization in machine learning is to uncover underlying latent factors of variation as well as the mechanism through which each factor acts in the world. In this paper, we test whether 17 unsupervised, weakly supervised, and fully supervised representation learning approaches correctly infer the generative factors of variation in simple datasets (dSprites, Shapes3D, MPI3D) from controlled environments, and on our contributed CelebGlow dataset. In contrast to prior robustness work that introduces novel factors of variation during test time, such as blur or other (un)structured noise, we here recompose, interpolate, or extrapolate only existing factors of variation from the training data set (e.g., small and medium-sized objects during training and large objects during testing). Models that learn the correct mechanism should be able to generalize to this benchmark. In total, we train and test 2000+ models and observe that all of them struggle to learn the underlying mechanism regardless of supervision signal and architectural bias. Moreover, the generalization capabilities of all tested models drop significantly as we move from artificial datasets towards more realistic real-world datasets. Despite their inability to identify the correct mechanism, the models are quite modular as their ability to infer other in-distribution factors remains fairly stable, providing only a single factoris out-of-distribution. These results point to an important yet understudied problem of learning mechanistic models of observations that can facilitate generalization.}, author = {Schott, Lukas and Kügelgen, Julius von and Träuble, Frederik and Gehler, Peter and Russell, Chris and Bethge, Matthias and Schölkopf, Bernhard and Locatello, Francesco and Brendel, Wieland}, booktitle = {10th International Conference on Learning Representations}, location = {Virtual}, title = {{Visual representation learning does not generalize strongly within the same domain}}, year = {2022}, } @inproceedings{14173, abstract = {Since out-of-distribution generalization is a generally ill-posed problem, various proxy targets (e.g., calibration, adversarial robustness, algorithmic corruptions, invariance across shifts) were studied across different research programs resulting in different recommendations. While sharing the same aspirational goal, these approaches have never been tested under the same experimental conditions on real data. In this paper, we take a unified view of previous work, highlighting message discrepancies that we address empirically, and providing recommendations on how to measure the robustness of a model and how to improve it. To this end, we collect 172 publicly available dataset pairs for training and out-of-distribution evaluation of accuracy, calibration error, adversarial attacks, environment invariance, and synthetic corruptions. We fine-tune over 31k networks, from nine different architectures in the many- and few-shot setting. Our findings confirm that in- and out-of-distribution accuracies tend to increase jointly, but show that their relation is largely dataset-dependent, and in general more nuanced and more complex than posited by previous, smaller scale studies.}, author = {Wenzel, Florian and Dittadi, Andrea and Gehler, Peter Vincent and Carl-Johann Simon-Gabriel, Carl-Johann Simon-Gabriel and Horn, Max and Zietlow, Dominik and Kernert, David and Russell, Chris and Brox, Thomas and Schiele, Bernt and Schölkopf, Bernhard and Locatello, Francesco}, booktitle = {36th Conference on Neural Information Processing Systems}, isbn = {9781713871088}, location = {New Orleans, LA, United States}, pages = {7181--7198}, publisher = {Neural Information Processing Systems Foundation}, title = {{Assaying out-of-distribution generalization in transfer learning}}, volume = {35}, year = {2022}, } @inproceedings{14174, abstract = {Building sample-efficient agents that generalize out-of-distribution (OOD) in real-world settings remains a fundamental unsolved problem on the path towards achieving higher-level cognition. One particularly promising approach is to begin with low-dimensional, pretrained representations of our world, which should facilitate efficient downstream learning and generalization. By training 240 representations and over 10,000 reinforcement learning (RL) policies on a simulated robotic setup, we evaluate to what extent different properties of pretrained VAE-based representations affect the OOD generalization of downstream agents. We observe that many agents are surprisingly robust to realistic distribution shifts, including the challenging sim-to-real case. In addition, we find that the generalization performance of a simple downstream proxy task reliably predicts the generalization performance of our RL agents under a wide range of OOD settings. Such proxy tasks can thus be used to select pretrained representations that will lead to agents that generalize.}, author = {Dittadi, Andrea and Träuble, Frederik and Wüthrich, Manuel and Widmaier, Felix and Gehler, Peter and Winther, Ole and Locatello, Francesco and Bachem, Olivier and Schölkopf, Bernhard and Bauer, Stefan}, booktitle = {10th International Conference on Learning Representations}, location = {Virtual}, title = {{The role of pretrained representations for the OOD generalization of reinforcement learning agents}}, year = {2022}, } @inproceedings{14175, abstract = {Predicting the future trajectory of a moving agent can be easy when the past trajectory continues smoothly but is challenging when complex interactions with other agents are involved. Recent deep learning approaches for trajectory prediction show promising performance and partially attribute this to successful reasoning about agent-agent interactions. However, it remains unclear which features such black-box models actually learn to use for making predictions. This paper proposes a procedure that quantifies the contributions of different cues to model performance based on a variant of Shapley values. Applying this procedure to state-of-the-art trajectory prediction methods on standard benchmark datasets shows that they are, in fact, unable to reason about interactions. Instead, the past trajectory of the target is the only feature used for predicting its future. For a task with richer social interaction patterns, on the other hand, the tested models do pick up such interactions to a certain extent, as quantified by our feature attribution method. We discuss the limits of the proposed method and its links to causality.}, author = {Makansi, Osama and Kügelgen, Julius von and Locatello, Francesco and Gehler, Peter and Janzing, Dominik and Brox, Thomas and Schölkopf, Bernhard}, booktitle = {10th International Conference on Learning Representations}, location = {Virtual}, title = {{You mostly walk alone: Analyzing feature attribution in trajectory prediction}}, year = {2022}, } @inproceedings{14215, abstract = {Geospatial Information Systems are used by researchers and Humanitarian Assistance and Disaster Response (HADR) practitioners to support a wide variety of important applications. However, collaboration between these actors is difficult due to the heterogeneous nature of geospatial data modalities (e.g., multi-spectral images of various resolutions, timeseries, weather data) and diversity of tasks (e.g., regression of human activity indicators or detecting forest fires). In this work, we present a roadmap towards the construction of a general-purpose neural architecture (GPNA) with a geospatial inductive bias, pre-trained on large amounts of unlabelled earth observation data in a self-supervised manner. We envision how such a model may facilitate cooperation between members of the community. We show preliminary results on the first step of the roadmap, where we instantiate an architecture that can process a wide variety of geospatial data modalities and demonstrate that it can achieve competitive performance with domain-specific architectures on tasks relating to the U.N.'s Sustainable Development Goals.}, author = {Rahaman, Nasim and Weiss, Martin and Träuble, Frederik and Locatello, Francesco and Lacoste, Alexandre and Bengio, Yoshua and Pal, Chris and Li, Li Erran and Schölkopf, Bernhard}, booktitle = {36th Conference on Neural Information Processing Systems}, location = {New Orleans, LA, United States}, title = {{A general purpose neural architecture for geospatial systems}}, year = {2022}, } @unpublished{14220, abstract = {Although reinforcement learning has seen remarkable progress over the last years, solving robust dexterous object-manipulation tasks in multi-object settings remains a challenge. In this paper, we focus on models that can learn manipulation tasks in fixed multi-object settings and extrapolate this skill zero-shot without any drop in performance when the number of objects changes. We consider the generic task of bringing a specific cube out of a set to a goal position. We find that previous approaches, which primarily leverage attention and graph neural network-based architectures, do not generalize their skills when the number of input objects changes while scaling as K2. We propose an alternative plug-and-play module based on relational inductive biases to overcome these limitations. Besides exceeding performances in their training environment, we show that our approach, which scales linearly in K, allows agents to extrapolate and generalize zero-shot to any new object number.}, author = {Mambelli, Davide and Träuble, Frederik and Bauer, Stefan and Schölkopf, Bernhard and Locatello, Francesco}, booktitle = {arXiv}, title = {{Compositional multi-object reinforcement learning with linear relation networks}}, doi = {10.48550/arXiv.2201.13388}, year = {2022}, } @article{14381, abstract = {Expander graphs (sparse but highly connected graphs) have, since their inception, been the source of deep links between Mathematics and Computer Science as well as applications to other areas. In recent years, a fascinating theory of high-dimensional expanders has begun to emerge, which is still in a formative stage but has nonetheless already lead to a number of striking results. Unlike for graphs, in higher dimensions there is a rich array of non-equivalent notions of expansion (coboundary expansion, cosystolic expansion, topological expansion, spectral expansion, etc.), with differents strengths and applications. In this talk, we will survey this landscape of high-dimensional expansion, with a focus on two main results. First, we will present Gromov’s Topological Overlap Theorem, which asserts that coboundary expansion (a quantitative version of vanishing mod 2 cohomology) implies topological expansion (roughly, the property that for every map from a simplicial complex to a manifold of the same dimension, the images of a positive fraction of the simplices have a point in common). Second, we will outline a construction of bounded degree 2-dimensional topological expanders, due to Kaufman, Kazhdan, and Lubotzky.}, author = {Wagner, Uli}, issn = {2102-622X}, journal = {Bulletin de la Societe Mathematique de France}, pages = {281--294}, publisher = {Societe Mathematique de France}, title = {{High-dimensional expanders (after Gromov, Kaufman, Kazhdan, Lubotzky, and others)}}, doi = {10.24033/ast.1188}, volume = {438}, year = {2022}, } @article{14437, abstract = {Future LEDs could be based on lead halide perovskites. A breakthrough in preparing device-compatible solids composed of nanoscale perovskite crystals overcomes a long-standing hurdle in making blue perovskite LEDs.}, author = {Utzat, Hendrik and Ibáñez, Maria}, issn = {1476-4687}, journal = {Nature}, keywords = {Multidisciplinary}, number = {7941}, pages = {638--639}, publisher = {Springer Nature}, title = {{Molecular engineering enables bright blue LEDs}}, doi = {10.1038/d41586-022-04447-0}, volume = {612}, year = {2022}, } @misc{18291, author = {Katsaros, Georgios and Jirovec, Daniel}, publisher = {Institute of Science and Technology Austria}, title = {{Dynamics of Hole Singlet-Triplet Qubits with Large 𝑔-Factor Differences}}, doi = {10.15479/AT:ISTA:18291}, year = {2022}, } @article{18606, abstract = {Shear thickening is an intriguing rheological behaviour which consists in a brutal increase in the viscosity above a critical shear rate. It is famously encountered in suspensions of corn starch in water. Despite having been discovered in the early 1930's, its underlying mechanisms remained a mystery for a long time. In 2013–14, numerical and theoretical works [[1], [2], [3]] put forward a frictional transition scenario to explain this phenomenon. In this talk, I will present experimental work investigating this frictional transition scenario. In order to test the ideas of this model, one has to go further than standard rheological techniques, since they do not provide access to the frictional state of the measured suspension. I will thus focus on the techniques that we developed in order to evidence the frictional transition and link it to the presence of a shear-thickening behaviour.}, author = {Clavaud, Cécile}, issn = {2772-5693}, journal = {Science Talks}, publisher = {Elsevier}, title = {{Shear thickening in dense suspensions: an experimental study}}, doi = {10.1016/j.sctalk.2022.100038}, volume = {3}, year = {2022}, } @article{12261, abstract = {Dose–response relationships are a general concept for quantitatively describing biological systems across multiple scales, from the molecular to the whole-cell level. A clinically relevant example is the bacterial growth response to antibiotics, which is routinely characterized by dose–response curves. The shape of the dose–response curve varies drastically between antibiotics and plays a key role in treatment, drug interactions, and resistance evolution. However, the mechanisms shaping the dose–response curve remain largely unclear. Here, we show in Escherichia coli that the distinctively shallow dose–response curve of the antibiotic trimethoprim is caused by a negative growth-mediated feedback loop: Trimethoprim slows growth, which in turn weakens the effect of this antibiotic. At the molecular level, this feedback is caused by the upregulation of the drug target dihydrofolate reductase (FolA/DHFR). We show that this upregulation is not a specific response to trimethoprim but follows a universal trend line that depends primarily on the growth rate, irrespective of its cause. Rewiring the feedback loop alters the dose–response curve in a predictable manner, which we corroborate using a mathematical model of cellular resource allocation and growth. Our results indicate that growth-mediated feedback loops may shape drug responses more generally and could be exploited to design evolutionary traps that enable selection against drug resistance.}, author = {Angermayr, Andreas and Pang, Tin Yau and Chevereau, Guillaume and Mitosch, Karin and Lercher, Martin J and Bollenbach, Mark Tobias}, issn = {1744-4292}, journal = {Molecular Systems Biology}, keywords = {Applied Mathematics, Computational Theory and Mathematics, General Agricultural and Biological Sciences, General Immunology and Microbiology, General Biochemistry, Genetics and Molecular Biology, Information Systems}, number = {9}, publisher = {Embo Press}, title = {{Growth‐mediated negative feedback shapes quantitative antibiotic response}}, doi = {10.15252/msb.202110490}, volume = {18}, year = {2022}, } @article{12264, abstract = {Reproductive isolation (RI) is a core concept in evolutionary biology. It has been the central focus of speciation research since the modern synthesis and is the basis by which biological species are defined. Despite this, the term is used in seemingly different ways, and attempts to quantify RI have used very different approaches. After showing that the field lacks a clear definition of the term, we attempt to clarify key issues, including what RI is, how it can be quantified in principle, and how it can be measured in practice. Following other definitions with a genetic focus, we propose that RI is a quantitative measure of the effect that genetic differences between populations have on gene flow. Specifically, RI compares the flow of neutral alleles in the presence of these genetic differences to the flow without any such differences. RI is thus greater than zero when genetic differences between populations reduce the flow of neutral alleles between populations. We show how RI can be quantified in a range of scenarios. A key conclusion is that RI depends strongly on circumstances—including the spatial, temporal and genomic context—making it difficult to compare across systems. After reviewing methods for estimating RI from data, we conclude that it is difficult to measure in practice. We discuss our findings in light of the goals of speciation research and encourage the use of methods for estimating RI that integrate organismal and genetic approaches.}, author = {Westram, Anja M and Stankowski, Sean and Surendranadh, Parvathy and Barton, Nicholas H}, issn = {1420-9101}, journal = {Journal of Evolutionary Biology}, keywords = {Ecology, Evolution, Behavior and Systematics}, number = {9}, pages = {1143--1164}, publisher = {Wiley}, title = {{What is reproductive isolation?}}, doi = {10.1111/jeb.14005}, volume = {35}, year = {2022}, } @article{12265, author = {Westram, Anja M and Stankowski, Sean and Surendranadh, Parvathy and Barton, Nicholas H}, issn = {1420-9101}, journal = {Journal of Evolutionary Biology}, keywords = {Ecology, Evolution, Behavior and Systematics}, number = {9}, pages = {1200--1205}, publisher = {Wiley}, title = {{Reproductive isolation, speciation, and the value of disagreement: A reply to the commentaries on ‘What is reproductive isolation?’}}, doi = {10.1111/jeb.14082}, volume = {35}, year = {2022}, } @article{12268, abstract = {The complexity of the microenvironment effects on cell response, show accumulating evidence that glioblastoma (GBM) migration and invasiveness are influenced by the mechanical rigidity of their surroundings. The epithelial–mesenchymal transition (EMT) is a well-recognized driving force of the invasive behavior of cancer. However, the primary mechanisms of EMT initiation and progression remain unclear. We have previously showed that certain substrate stiffness can selectively stimulate human GBM U251-MG and GL15 glioblastoma cell lines motility. The present study unifies several known EMT mediators to uncover the reason of the regulation and response to these stiffnesses. Our results revealed that changing the rigidity of the mechanical environment tuned the response of both cell lines through change in morphological features, epithelial-mesenchymal markers (E-, N-Cadherin), EGFR and ROS expressions in an interrelated manner. Specifically, a stiffer microenvironment induced a mesenchymal cell shape, a more fragmented morphology, higher intracellular cytosolic ROS expression and lower mitochondrial ROS. Finally, we observed that cells more motile showed a more depolarized mitochondrial membrane potential. Unravelling the process that regulates GBM cells’ infiltrative behavior could provide new opportunities for identification of new targets and less invasive approaches for treatment.}, author = {Basilico, Bernadette and Palamà, Ilaria Elena and D’Amone, Stefania and Lauro, Clotilde and Rosito, Maria and Grieco, Maddalena and Ratano, Patrizia and Cordella, Federica and Sanchini, Caterina and Di Angelantonio, Silvia and Ragozzino, Davide and Cascione, Mariafrancesca and Gigli, Giuseppe and Cortese, Barbara}, issn = {2234-943X}, journal = {Frontiers in Oncology}, keywords = {Cancer Research, Oncology}, publisher = {Frontiers Media}, title = {{Substrate stiffness effect on molecular crosstalk of epithelial-mesenchymal transition mediators of human glioblastoma cells}}, doi = {10.3389/fonc.2022.983507}, volume = {12}, year = {2022}, } @article{12269, abstract = {We study the thermalization of a small XX chain coupled to long, gapped XXZ leads at either side by observing the relaxation dynamics of the whole system. Using extensive tensor network simulations, we show that such systems, although not integrable, appear to show either extremely slow thermalization or even lack thereof since the two cannot be distinguished within the accuracy of our numerics. We show that the persistent oscillations observed in the spin current in the middle of the XX chain are related to eigenstates of the entire system located within the gap of the boundary chains. We find from exact diagonalization that some of these states remain strictly localized within the XX chain and do not hybridize with the rest of the system. The frequencies of the persistent oscillations determined by numerical simulations of dynamics match the energy differences between these states exactly. This has important implications for open systems, where the strongly interacting leads are often assumed to thermalize the central system. Our results suggest that, if we employ gapped systems for the leads, this assumption does not hold.}, author = {Ljubotina, Marko and Roy, Dibyendu and Prosen, Tomaž}, issn = {2469-9969}, journal = {Physical Review B}, number = {5}, publisher = {American Physical Society}, title = {{Absence of thermalization of free systems coupled to gapped interacting reservoirs}}, doi = {10.1103/physrevb.106.054314}, volume = {106}, year = {2022}, }