Searching for novel two-dimensional (2D) materials is crucial for the development of the next generation technologies such as electronics, optoelectronics, electrochemistry and biomedicine. In this work, we designed a series of 2D materials based on
endohedral fullerenes, and revealed that many of them integrate different functions in a single system, such as ferroelectricity with large electric dipole moments, multiple magnetic phases with both strong magnetic anisotropy and high Curie temperature, quantum spin Hall effect or quantum anomalous Hall effect with robust topologically protected edge states. We further proposed a new style topological field-effect transistor. These findings provide a strategy of using fullerenes as building blocks for the synthesis of novel 2D materials which can be easily controlled with a local electric field.
2D ferrovalley materials that exhibit spontaneous valley polarization are both fundamentally intriguing and practically appealing to be used in valleytronic devices. Usually, the research on 2D ferrovalley materials is mainly focused on inorganic sys
tems, severely suffering from in-plane magnetization. Here, we alternatively show by kp model analysis and high-throughput first-principles calculations that ideal spontaneous valley polarization is present in 2D organometallic lattice. We explore the design principle for organic 2D ferrovalley materials composed of (quasi-)planer molecules and transition-metal atoms in hexagonal lattice, and identify twelve promising candidates. These systems have a ferromagnetic or antiferromagnetic semiconducting state, and importantly they exhibit robust out-of-plane magnetization. The interplay between spin and valley, together with strong spin-orbit coupling of transition-metal atoms, guarantee the spontaneous valley polarization in these systems, facilitating the anomalous valley Hall effect. Our findings significantly broaden the scientific and technological impact of ferrovalley physics.
The convolutional neural network (CNN) is vulnerable to degraded images with even very small variations (e.g. corrupted and adversarial samples). One of the possible reasons is that CNN pays more attention to the most discriminative regions, but igno
res the auxiliary features when learning, leading to the lack of feature diversity for final judgment. In our method, we propose to dynamically suppress significant activation values of CNN by group-wise inhibition, but not fixedly or randomly handle them when training. The feature maps with different activation distribution are then processed separately to take the feature independence into account. CNN is finally guided to learn richer discriminative features hierarchically for robust classification according to the proposed regularization. Our method is comprehensively evaluated under multiple settings, including classification against corruptions, adversarial attacks and low data regime. Extensive experimental results show that the proposed method can achieve significant improvements in terms of both robustness and generalization performances, when compared with the state-of-the-art methods. Code is available at https://github.com/LinusWu/TENET_Training.
Human-Object Interaction (HOI) detection is an important problem to understand how humans interact with objects. In this paper, we explore interactiveness knowledge which indicates whether a human and an object interact with each other or not. We fou
nd that interactiveness knowledge can be learned across HOI datasets and bridge the gap between diverse HOI category settings. Our core idea is to exploit an interactiveness network to learn the general interactiveness knowledge from multiple HOI datasets and perform Non-Interaction Suppression (NIS) before HOI classification in inference. On account of the generalization ability of interactiveness, interactiveness network is a transferable knowledge learner and can be cooperated with any HOI detection models to achieve desirable results. We utilize the human instance and body part features together to learn the interactiveness in hierarchical paradigm, i.e., instance-level and body part-level interactivenesses. Thereafter, a consistency task is proposed to guide the learning and extract deeper interactive visual clues. We extensively evaluate the proposed method on HICO-DET, V-COCO, and a newly constructed PaStaNet-HOI dataset. With the learned interactiveness, our method outperforms state-of-the-art HOI detection methods, verifying its efficacy and flexibility. Code is available at https://github.com/DirtyHarryLYL/Transferable-Interactiveness-Network.
We investigate the stellar kinematics of the Galactic disc in 7 $<$ $R$ $<$ 13,kpc using a sample of 118,945 red giant branch (RGB) stars from LAMOST and Gaia. We characterize the median, dispersion and skewness of the distributions of the 3D stellar
velocities, actions and orbital parameters across the age-metallicity and the disc $R$ -- $Z$ plane. Our results reveal abundant but clear stellar kinematic patterns and structures in the age -- metallicity and the disc $R$ -- $Z$ plane. The most prominent feature is the strong variations of the velocity, action, and orbital parameter distributions from the young, metal-rich thin disc to the old, metal-poor thick disc, a number of smaller-scale structures -- such as velocity streams, north-south asymmetries, and kinematic features of spiral arms -- are clearly revealed. Particularly, the skewness of $V_{phi}$ and $J_{phi}$ reveals a new substructure at $Rsimeq12$,kpc and $Zsimeq0$,kpc, possibly related to dynamical effects of spiral arms in the outer disc. We further study the stellar migration through analysing the stellar orbital parameters and stellar birth radii. The results suggest that the thick disc stars near the solar radii and beyond are mostly migrated from the inner disc of $Rsim4 - 6$,kpc due to their highly eccentrical orbits. Stellar migration due to dynamical processes with angular momentum transfer (churning) are prominent for both the old, metal-rich stars (outward migrators) and the young metal-poor stars (inward migrators). The spatial distribution in the $R$ -- $Z$ plane for the inward migrators born at a Galactocentric radius of $>$12,kpc show clear age stratifications, possibly an evidence that these inward migrators are consequences of splashes triggered by merger events of satellite galaxies that have been lasted in the past few giga years.
A new multifunctional 2D material is theoretically predicted based on systematic ab-initio calculations and model simulations for the honeycomb lattice of endohedral fullerene W@C28 molecules. It has structural bistability, ferroelectricity, multiple
magnetic phases, and excellent valley characters and can be easily functionalized by the proximity effect with magnetic isolators such as MnTiO3. Furthermore, we may also manipulate the valley Hall and spin transport properties by selectively switch a few W@C28 molecules to the metastable phase. These findings pave a new way in integrate different functions in a single 2D material for technological innovations.
Synthetic dimensions can be rendered in the physical space and this has been achieved with photonics and cold atomic gases, however, little to no work has been succeeded in acoustics because acoustic wave-guides cannot be weakly coupled in a continuo
us fashion. Here, we establish the theoretical principles and for the first time manufacture acoustic crystals composed of arrays of acoustic cavities strongly coupled through modulated channels to evidence one-dimensional (1D) and two-dimensional (2D) dynamic topological pumpings. In particular, the topological edge-bulkedge and corner-bulk-corner transport are physically illustrated in finite-sized acoustic structures. We delineate the generated 2D and four-dimensional (4D) quantum Hall effects by calculating first and second Chern numbers and demonstrating robustness against the geometrical imperfections. Synthetic dimensions could provide a powerful way for acoustic topological wave steering and open up a new platform to explore higher-order topological matter in dimensions four and higher.
Well-protected magnetization, tunable quantum states and long coherence time are desired for developing magnetic molecules as qubits quantum information processing and storage. Based on the first-principles calculations and dynamic simulations, we de
monstrate that endohedral fullerene molecule Ir@C28 has stable magnetization, huge magnetic anisotropy energy (> 30 meV per molecule) and bias-tunable structural phases. In particular, qubits based on Ir@C28 may have coherence times up to several mS at high temperature (~100K) after full consideration of spin-vibration couplings. These results suggest a new strategy of using endohedral fullerene as qubits for technological breakthroughs.
Finding new two-dimensional (2D) materials with novel quantum properties is highly desirable for technological innovations. In this work, we studied a series of metal-organic frameworks (MOFs) with different metal cores and discovered various attract
ive properties, such as room-temperature magnetic ordering, strong perpendicular magnetic anisotropy, huge topological band gap (>200meV), and excellent spin-filtering performance. As many MOFs have been successfully synthesized in experiments, our results suggest realistic new 2D functional materials for the design of spintronic nanodevices.
Human-Object Interaction (HOI) consists of human, object and implicit interaction/verb. Different from previous methods that directly map pixels to HOI semantics, we propose a novel perspective for HOI learning in an analytical manner. In analogy to
Harmonic Analysis, whose goal is to study how to represent the signals with the superposition of basic waves, we propose the HOI Analysis. We argue that coherent HOI can be decomposed into isolated human and object. Meanwhile, isolated human and object can also be integrated into coherent HOI again. Moreover, transformations between human-object pairs with the same HOI can also be easier approached with integration and decomposition. As a result, the implicit verb will be represented in the transformation function space. In light of this, we propose an Integration-Decomposition Network (IDN) to implement the above transformations and achieve state-of-the-art performance on widely-used HOI detection benchmarks. Code is available at https://github.com/DirtyHarryLYL/HAKE-Action-Torch/tree/IDN-(Integrating-Decomposing-Network).
