The only entanglement quantity is concurrence in a 2-qubit pure state. The maximum violation of Bells inequality is monotonically increasing for this quantity. Therefore, people expect that pure state entanglement is relevant to the quantum violation
. To understand the relation between violation and entanglement, we extend the study to three qubits. We consider all possible 3-qubit operators with a symmetric permutation. When only turning on one entanglement measure, the numerical result shows a contradiction to the expectation. The maximum violation does not have the same behavior as in 2-qubit pure states. Therefore, we conclude Violation$ eq$Quantum. In the end, we propose the generalized $R$-matrix or correlation matrix for the new diagnosis of Quantum Entanglement. We demonstrate the evidence by restoring the monotonically increasing result.
Optical Character Recognition (OCR) systems have been widely used in various of application scenarios. Designing an OCR system is still a challenging task. In previous work, we proposed a practical ultra lightweight OCR system (PP-OCR) to balance the
accuracy against the efficiency. In order to improve the accuracy of PP-OCR and keep high efficiency, in this paper, we propose a more robust OCR system, i.e. PP-OCRv2. We introduce bag of tricks to train a better text detector and a better text recognizer, which include Collaborative Mutual Learning (CML), CopyPaste, Lightweight CPUNetwork (LCNet), Unified-Deep Mutual Learning (U-DML) and Enhanced CTCLoss. Experiments on real data show that the precision of PP-OCRv2 is 7% higher than PP-OCR under the same inference cost. It is also comparable to the server models of the PP-OCR which uses ResNet series as backbones. All of the above mentioned models are open-sourced and the code is available in the GitHub repository PaddleOCR which is powered by PaddlePaddle.
Numerical reasoning skills are essential for complex question answering (CQA) over text. It requires opertaions including counting, comparison, addition and subtraction. A successful approach to CQA on text, Neural Module Networks (NMNs), follows the
programmer-interpreter paradigm and leverages specialised modules to perform compositional reasoning. However, the NMNs framework does not consider the relationship between numbers and entities in both questions and paragraphs. We propose effective techniques to improve NMNs numerical reasoning capabilities by making the interpreter question-aware and capturing the relationship between entities and numbers. On the same subset of the DROP dataset for CQA on text, experimental results show that our additions outperform the original NMNs by 3.0 points for the overall F1 score.
The scene graph generation (SGG) task aims to detect visual relationship triplets, i.e., subject, predicate, object, in an image, providing a structural vision layout for scene understanding. However, current models are stuck in common predicates, e.
g., on and at, rather than informative ones, e.g., standing on and looking at, resulting in the loss of precise information and overall performance. If a model only uses stone on road rather than blocking to describe an image, it is easy to misunderstand the scene. We argue that this phenomenon is caused by two key imbalances between informative predicates and common ones, i.e., semantic space level imbalance and training sample level imbalance. To tackle this problem, we propose BA-SGG, a simple yet effective SGG framework based on balance adjustment but not the conventional distribution fitting. It integrates two components: Semantic Adjustment (SA) and Balanced Predicate Learning (BPL), respectively for adjusting these imbalances. Benefited from the model-agnostic process, our method is easily applied to the state-of-the-art SGG models and significantly improves the SGG performance. Our method achieves 14.3%, 8.0%, and 6.1% higher Mean Recall (mR) than that of the Transformer model at three scene graph generation sub-tasks on Visual Genome, respectively. Codes are publicly available.
Quantifying genuine entanglement is a crucial task in quantum information theory. In this work, we give an approach of constituting genuine $m$-partite entanglement measure from any bipartite entanglement and any $k$-partite entanglement measure, $3l
eq k<m$.In addition, as a complement to the three-qubit concurrence triangle proposed in [Phys. Rev. Lett., 127, 040403], we show that the triangle relation is also valid for any other entanglement measure and system with any dimension. We also discuss the tetrahedron structure for the four-partite system via the triangle relation associated with tripartite and bipartite entanglement respectively. For multipartite system that contains more than four parties, there is no symmetric geometric structure as that of tri- and four-partite cases.
The symmetries of a crystal form the guiding principle to understand the topology of its band structure. They dictate the location and degrees of stable band crossings which lead to significant sources of Berry curvature. Here we show how non-crystal
line quasi-symmetries stabilize near-degeneracies of bands over extended regions in energy and in the Brillouin zone. Specifically, a quasi-symmetry is an exact symmetry of a $kcdot p$ Hamiltonian to lower-order that is broken by higher-order terms. Hence quasi-symmetric points are gapped, yet the gap is parametrically small and therefore does not influence the physical properties of the system. We demonstrate that in the eV-bandwidth semi-metal CoSi an internal quasi-symmetry stabilizes gaps in the 1-2 meV range over a large near-degenerate plane. This quasi-symmetry is key to explaining the surprising simplicity of the experimentally observed quantum oscillations of four interpenetrating Fermi surfaces around the R-point. Untethered from limitations of crystalline symmetry, quasi-symmetries can source large Berry curvature over wide ranges of energy and on low symmetry points - thereby impacting quasiparticle dynamics in unexpected places. Quasi-symmetries also lead to new types of Wigner-Von Neumann classifications.
Simulation of fermionic relativistic physics (such as Dirac and Weyl points) has led the dicovery of versatile and exotic phenomena in photonics, of which the optical-frequency realization is, however, still a challenging aim. Here we discover that t
he commonly-used woodpile photonic crystals for optical-frequency applications host novel fermionic relativistic degeneracies: a Dirac linenode and a topological quadratic degeneracy point, as {em guaranteed} by the nonsymmorphic crystalline symmetry. By reducing the space symmetry, type-II Dirac/Weyl points emerge as the descendants of the quadratic degeneracy point. These exotic optical waves mimicking the physics of unconventional fermionic relativistic waves and hosting anomalous optical properties in subwavelength, all-dielectric photonic crystals could open a new avenue for future optical science.
The goal of dialogue state tracking (DST) is to predict the current dialogue state given all previous dialogue contexts. Existing approaches generally predict the dialogue state at every turn from scratch. However, the overwhelming majority of the sl
ots in each turn should simply inherit the slot values from the previous turn. Therefore, the mechanism of treating slots equally in each turn not only is inefficient but also may lead to additional errors because of the redundant slot value generation. To address this problem, we devise the two-stage DSS-DST which consists of the Dual Slot Selector based on the current turn dialogue, and the Slot Value Generator based on the dialogue history. The Dual Slot Selector determines each slot whether to update slot value or to inherit the slot value from the previous turn from two aspects: (1) if there is a strong relationship between it and the current turn dialogue utterances; (2) if a slot value with high reliability can be obtained for it through the current turn dialogue. The slots selected to be updated are permitted to enter the Slot Value Generator to update values by a hybrid method, while the other slots directly inherit the values from the previous turn. Empirical results show that our method achieves 56.93%, 60.73%, and 58.04% joint accuracy on MultiWOZ 2.0, MultiWOZ 2.1, and MultiWOZ 2.2 datasets respectively and achieves a new state-of-the-art performance with significant improvements.
The quest to improve transparent conductors balances two key goals: increasing electrical conductivity and increasing optical transparency. To improve both simultaneously is hindered by the physical limitation that good metals with high electrical co
nductivity have large carrier densities that push the plasma edge into the ultra-violet range. Transparent conductors are compromises between electrical conductivity, requiring mobile electrons, and optical transparency based on immobile charges to avoid screening of visible light. Technological solutions reflect this trade-off, achieving the desired transparencies by reducing the conductor thickness or carrier density at the expense of a lower conductance. Here we demonstrate that highly anisotropic crystalline conductors offer an alternative solution, avoiding this compromise by separating the directions of conduction and transmission. Materials with a quasi-two-dimensional electronic structure have a plasma edge well below the range of visible light while maintaining excellent in-plane conductivity. We demonstrate that slabs of the layered oxides Sr$_2$RuO$_4$ and Tl$_2$Ba$_2$CuO$_{6+delta}$ are optically transparent even at macroscopic thicknesses >2$mu$m for c-axis polarized light. Underlying this observation is the fabrication of out-of-plane slabs by focused ion beam milling. This work provides a glimpse into future technologies, such as highly polarized and addressable optical screens, that advancements in a-axis thin film growth will enable.
Label Smoothing (LS) improves model generalization through penalizing models from generating overconfident output distributions. For each training sample the LS strategy smooths the one-hot encoded training signal by distributing its distribution mas
s over the non-ground truth classes. We extend this technique by considering example pairs, coined PLS. PLS first creates midpoint samples by averaging random sample pairs and then learns a smoothing distribution during training for each of these midpoint samples, resulting in midpoints with high uncertainty labels for training. We empirically show that PLS significantly outperforms LS, achieving up to 30% of relative classification error reduction. We also visualize that PLS produces very low winning softmax scores for both in and out of distribution samples.
