Search

We introduce DA-Code, a code generation benchmark specifically designed to assess LLMs on agent-based data science tasks. This benchmark features three core elements: First, the tasks within DA-Code are inherently challenging, setting them apart from traditional code generation tasks and demanding advanced coding skills in grounding and planning. Second, examples in DA-Code are all based on real and diverse data, covering a wide range of complex data wrangling and analytics tasks. Third, to solve the tasks, the models must utilize complex data science programming languages, to perform intricate data processing and derive the answers. We set up the benchmark in a controllable and executable environment that aligns with real-world data analysis scenarios and is scalable. The annotators meticulously design the evaluation suite to ensure the accuracy and robustness of the evaluation. We develop the DA-Agent baseline. Experiments show that although the baseline performs better than other existing frameworks, using the current best LLMs achieves only 30.5% accuracy, leaving ample room for improvement. We release our benchmark at https://da-code-bench.github.io., Comment: EMNLP 2024

Let \mu_{M,D} be the self-similar measure generated by the positive integer M=RN^q and the product-form digit set D=\{0,1,\dots,N-1\}\oplus N^{p_1}\{0,1,\dots,N-1\}\oplus \cdots \oplus N^{p_s}\{0,1,\dots,N-1\}, where R>1, N>1, q, p_i(1\leq i\leq s) are positive integers with gcd(R,N)=1 and p_1

Obesity is widely recognized as a critical and pervasive health concern. We strive to identify important genetic risk factors from hundreds of thousands of single nucleotide polymorphisms (SNPs) for obesity. We propose and apply a novel Quantile Regression with Insight Fusion (QRIF) approach that can integrate insights from established studies or domain knowledge to simultaneously select variables and modeling for ultra-high dimensional genetic data, focusing on high conditional quantiles of body mass index (BMI) that are of most interest. We discover interesting new SNPs and shed new light on a comprehensive view of the underlying genetic risk factors for different levels of BMI. This may potentially pave the way for more precise and targeted treatment strategies. The QRIF approach intends to balance the trade-off between the prior insights and the observed data while being robust to potential false information. We further establish the desirable asymptotic properties under the challenging non-differentiable check loss functions via Huber loss approximation and nonconvex SCAD penalty via local linear approximation. Finally, we develop an efficient algorithm for the QRIF approach. Our simulation studies further demonstrate its effectiveness.

Model-based methods in reinforcement learning offer a promising approach to enhance data efficiency by facilitating policy exploration within a dynamics model. However, accurately predicting sequential steps in the dynamics model remains a challenge due to the bootstrapping prediction, which attributes the next state to the prediction of the current state. This leads to accumulated errors during model roll-out. In this paper, we propose the Any-step Dynamics Model (ADM) to mitigate the compounding error by reducing bootstrapping prediction to direct prediction. ADM allows for the use of variable-length plans as inputs for predicting future states without frequent bootstrapping. We design two algorithms, ADMPO-ON and ADMPO-OFF, which apply ADM in online and offline model-based frameworks, respectively. In the online setting, ADMPO-ON demonstrates improved sample efficiency compared to previous state-of-the-art methods. In the offline setting, ADMPO-OFF not only demonstrates superior performance compared to recent state-of-the-art offline approaches but also offers better quantification of model uncertainty using only a single ADM.

Silk damping is well known in the study of cosmic microwave background (CMB) and accounts for suppression of the angular power spectrum of CMB on large angular multipoles. In this Letter, we study the effect of Silk damping on the scalar-induced gravitational waves (SIGWs). Resulting from the dissipation of cosmic fluid, the Silk damping notably suppresses the energy-density spectrum of SIGWs on scales comparable to a diffusion scale at the decoupling time of feebly-interacting particles. The effect offers a novel observable for probing the underlying particle interaction, especially for those mediated by heavy gauge bosons beyond the standard model of particles. We anticipate that pulsar timing arrays are sensitive to gauge bosons with mass $\sim10^{3}-10^{4}\,\mathrm{GeV}$, while space- and ground-based interferometers to those with mass $\sim10^7-10^{12}\,\mathrm{GeV}$, leading to essential complements to on-going and future experiments of high-energy physics., Comment: 7 pages, 2 figures, additional references added

Although supervised machine learning is popular for information extraction from clinical notes, creating large annotated datasets requires extensive domain expertise and is time-consuming. Meanwhile, large language models (LLMs) have demonstrated promising transfer learning capability. In this study, we explored whether recent LLMs can reduce the need for large-scale data annotations. We curated a manually-labeled dataset of 769 breast cancer pathology reports, labeled with 13 categories, to compare zero-shot classification capability of the GPT-4 model and the GPT-3.5 model with supervised classification performance of three model architectures: random forests classifier, long short-term memory networks with attention (LSTM-Att), and the UCSF-BERT model. Across all 13 tasks, the GPT-4 model performed either significantly better than or as well as the best supervised model, the LSTM-Att model (average macro F1 score of 0.83 vs. 0.75). On tasks with high imbalance between labels, the differences were more prominent. Frequent sources of GPT-4 errors included inferences from multiple samples and complex task design. On complex tasks where large annotated datasets cannot be easily collected, LLMs can reduce the burden of large-scale data labeling. However, if the use of LLMs is prohibitive, the use of simpler supervised models with large annotated datasets can provide comparable results. LLMs demonstrated the potential to speed up the execution of clinical NLP studies by reducing the need for curating large annotated datasets. This may result in an increase in the utilization of NLP-based variables and outcomes in observational clinical studies.

The multi-drifting subpulse behaviors in PSR J2007+0910 have been studied carefully with the high sensitivity observations of the Five-hundred-meter Aperture Spherical radio Telescope (FAST) at 1250 MHz. We found that there are at least six different single emission modes in PSR J2007+0910 are observed, four of which show significant subpulse drifting behaviors (modes A, B, C, and D), and the remaining two (modes $E_1$ and $E_2$) show stationary subpulse structures. The subpulse drifting periods of modes A, B, C, and D are $P_{3, A} = 8.7 \pm 1.6 P$, $P_{3, B} = 15.8 \pm 1.2 P$, $P_{3, C} = 21.6 \pm 1.3 P$ and $P_{3, D} = 32.3 \pm 0.9 P$, respectively, where $P$ represents the pulse period of this pulsar. The subpulse separation is almost the same for all modes $P_2 = 6.01 \pm 0.18 ^\circ$. Deep analysis suggests that the appearance and significant changes in the drifting period of multi-drifting subpulse emission modes for a pulsar may originate from the aliasing effect. The observed non-drifting modes may be caused by the spark point move with a period ~P_2. Our statistical analysis shows that the drift mode of this pulsar almost always switches from slower to faster drifts in the mode change. The interesting subpulse emission phenomenon of PSR J2007+0910 provides a unique opportunity to understand the switching mechanism of multi-drift mode., Comment: 10 pages, 10 figures, 2 tables, Accepted to MNRAS

We propose a new model to generate large anisotropies in the scalar-induced gravitational wave (SIGW) background via sound speed resonance in the inflaton-curvaton mixed scenario. Cosmological curvature perturbations are not only exponentially amplified at a resonant frequency, but also preserve significant non-Gaussianity of local type described by $f_{\mathrm{nl}}$. Besides a significant enhancement of energy-density fraction spectrum, large anisotropies in SIGWs can be generated, because of super-horizon modulations of the energy density due to existence of primordial non-Gaussianity. A reduced angular power spectrum $\tilde{C}_{\ell}$ could reach an amplitude of $[\ell(\ell+1)\tilde{C}_{\ell}]^{1/2} \sim 10^{-2}$, leading to potential measurements via planned gravitational-wave detectors such as DECIGO. The large anisotropies in SIGWs would serve as a powerful probe of the early universe, shedding new light on the inflationary dynamics, primordial non-Gaussianity, and primordial black hole dark matter., Comment: 10 pages, 2 figures, 2 columns, additional references added

CNN-based object detection methods have achieved significant progress in recent years. The classic structures of CNNs produce pyramid-like feature maps due to the pooling or other re-scale operations. The feature maps in different levels of the feature pyramid are used to detect objects with different scales. For more accurate object detection, the highest-level feature, which has the lowest resolution and contains the strongest semantics, is up-scaled and connected with the lower-level features to enhance the semantics in the lower-level features. However, the classic mode of feature connection combines the feature of lower-level with all the features above it, which may result in semantics degradation. In this paper, we propose a skipped connection to obtain stronger semantics at each level of the feature pyramid. In our method, the lower-level feature only connects with the feature at the highest level, making it more reasonable that each level is responsible for detecting objects with fixed scales. In addition, we simplify the generation of anchor for bounding box regression, which can further improve the accuracy of object detection. The experiments on the MS COCO and Wider Face demonstrate that our method outperforms the state-of-the-art methods.

Imitation learning (IL) enables agents to mimic expert behaviors. Most previous IL techniques focus on precisely imitating one policy through mass demonstrations. However, in many applications, what humans require is the ability to perform various tasks directly through a few demonstrations of corresponding tasks, where the agent would meet many unexpected changes when deployed. In this scenario, the agent is expected to not only imitate the demonstration but also adapt to unforeseen environmental changes. This motivates us to propose a new topic called imitator learning (ItorL), which aims to derive an imitator module that can on-the-fly reconstruct the imitation policies based on very limited expert demonstrations for different unseen tasks, without any extra adjustment. In this work, we focus on imitator learning based on only one expert demonstration. To solve ItorL, we propose Demo-Attention Actor-Critic (DAAC), which integrates IL into a reinforcement-learning paradigm that can regularize policies' behaviors in unexpected situations. Besides, for autonomous imitation policy building, we design a demonstration-based attention architecture for imitator policy that can effectively output imitated actions by adaptively tracing the suitable states in demonstrations. We develop a new navigation benchmark and a robot environment for \topic~and show that DAAC~outperforms previous imitation methods \textit{with large margins} both on seen and unseen tasks.

We demonstrate transition between the fine structure splitting of the ground state of triply ionized zirconium (Zr IV) is suitable for a terahertz (THz) atomic clock. Its transition frequency is about 37.52 THz and is mainly guided by the magnetic dipole (M1) transition and can be accessible by a readily available laser. We suggest to consider stable even isotopes of Zr and $M_J= \pm 1/2$ sublevels (i.e. $|4D_{3/2},M_J=\pm 1/2\rangle \rightarrow |4D_{5/2},M_J=\pm 1/2\rangle$ clock transition) for the experimental advantage. By performing necessary calculations, we have estimated possible systematics due to blackbody radiation, ac Stark, electric quadrupole and second-order Zeeman shifts along with shifts due to the second-order Doppler effects. The proposed THz atomic clock can be very useful in quantum thermometry and frequency metrology., Comment: 1 figure, 11 pages, accepted for publication in Physical Review A

Laser cutting of semiconductor wafers and transparent dielectrics has become a dominant process in manufacturing industries, encompassing a wide range of applications from flat display panels to microelectronic chips. Limited by the diffraction barrier imposed on the beam width and its longitudinal extend of laser focus, a trade-off must be made between cutting accuracy and aspect ratio in conventional laser processing, with accuracy typically approaching a micron and the aspect ratio on the order of $10^2$. Herein, we propose a method to circumvent this limitation. It is based on the laser modification induced by a back-scattering interference crawling mechanism, which creates a positive feedback for homogenizing longitudinal energy deposition and lateral sub-wavelength light confinement during laser-matter interaction. Consequently, cutting width on the scale of tens of nanometers and aspect ratio $10^3 \sim 10^4$ were simultaneously achieved. We refer to this technique as ``super stealth dicing'', which is validated through numerical simulations, ensuring its broad applicability. It can be applied to various transparent functional solids, such as glass, laser crystal, ferroelectric, and semiconductor, and is elevating the precision of future advanced laser dicing, patterning, and drilling into the nanometric era., Comment: 32 pages, 5 figures

We examine energy level-crossings of fine-structure (FS) levels in the heavier highly charged ions (HCIs) with $d^6$ and $d^8$ configurations. From the analysis, we find that some of these HCIs are tailor-made for atomic clocks with quality factors ranging from $10^{16}$ to $10^{18}$ and fractional uncertainties below $10^{-19}$ level. Many of them are also realized to be highly sensitive to the observation of temporal variation of FS constant ($\alpha$) and testing violation of local Lorentz symmetry invariance (LLI). To probe variations of $\alpha$ and LLI, we have determined their corresponding sensitivity coefficients in the investigated HCIs. Similarly, we have estimated orders of magnitudes of the Zeeman, Stark, black-body radiation and electric quadrupole shifts of the clock transitions of the considered HCIs in order to demonstrate them as the potential candidates for atomic clocks., Comment: 5 pages, 1 figure

Robust regression has attracted a great amount of attention in the literature recently, particularly for taking asymmetricity into account simultaneously and for high-dimensional analysis. However, the majority of research on the topics falls in frequentist approaches, which are not capable of full probabilistic uncertainty quantification. This paper first proposes a new Huberised-type of asymmetric loss function and its corresponding probability distribution which is shown to have the scale-mixture of normals. Then we introduce a new Bayesian Huberised regularisation for robust regression. A by-product of the research is that a new Bayesian Huberised regularised quantile regression is also derived. We further present their theoretical posterior properties. The robustness and effectiveness of the proposed models are demonstrated in the simulation studies and the real data analysis.

We predict a scheme for the creation of isotropic three-dimensional droplets in Rydbeg-dressed Bose gases, which contain both repulsive contact interactions and attractive van der Waals interactions causing the quantum fluctuation effect non-negligible. We present detailed beyond mean-field calculations with Lee-Huang-Yang correction and demonstrate the existence of isotropic droplets under realistic experimental conditions. Stable droplets possess flat-top density distribution, and their chemical potentials decrease with the particle number expansion towarding a critical value. We distinguish droplets from bright solitons through peak density, width of condensate and quantum depletion calculations. We summarize a phase diagram of realizing droplets, and subsequently highlight the stability of droplets by real time evolution as well as collisions. Our work provides a novel platform for investigating excitation spectrum and superfluid nature of droplets.

Primordial gravitational waves are a crucial prediction of inflation theory, and their detection through their imprints on the cosmic microwave background is actively being pursued. However, these attempts have not yet been successful. In this paper, we propose a novel approach to detect primordial gravitational waves by searching for a signal of second-order tensor perturbations. These perturbations were produced due to nonlinear couplings between the linear tensor and scalar perturbations in the early universe. We anticipate a blue-tilted tensor spectral index, and suggest that the tensor-to-scalar ratio can potentially be measured with high precision using a detector network composed of the ground-based Einstein Telescope and the space-borne LISA project on a decade timescale., Comment: 20 pages, 2 figures, 1 table, 3 appendices

Soliton microcombs are helping to advance the miniaturization of a range of comb systems. These combs mode lock through the formation of short temporal pulses in anomalous dispersion resonators. Here, a new microcomb is demonstrated that mode locks through the formation of pulse pairs in normal-dispersion coupled-ring resonators. Unlike conventional microcombs, pulses in this system cannot exist alone, and instead must phase lock in pairs to form a bright soliton comb. Also, the pulses can form at recurring spectral windows and the pulses in each pair feature different optical spectra. This pairwise mode-locking modality extends to higher dimensions and we demonstrate 3-ring systems in which 3 pulses mode lock through alternating pairwise pulse coupling. The results are demonstrated using the new CMOS-foundry platform that has not previously produced bright solitons on account of its inherent normal dispersion. The ability to generate multi-color pulse pairs over multiple rings is an important new feature for microcombs. It can extend the concept of all-optical soliton buffers and memories to multiple storage rings that multiplex pulses with respect to soliton color and that are spatially addressable. The results also suggest a new platform for the study of quantum combs and topological photonics.

Normal group velocity dispersion (GVD) microcombs offer high comb line power and high pumping efficiency compared to bright pulse microcombs. The recent demonstration of normal GVD microcombs using CMOS-foundry-produced microresonators is an important step towards scalable production. However, the chromatic dispersion of CMOS devices is large and impairs generation of broadband microcombs. Here, we report the development of a microresonator in which GVD is reduced due to a couple-ring resonator configuration. Operating in the turnkey self-injection-locking mode, the resonator is hybridly integrated with a semiconductor laser pump to produce high-power-efficiency combs spanning a bandwidth of 9.9 nm (1.22 THz) centered at 1560 nm, corresponding to 62 comb lines. Fast, linear optical sampling of the comb waveform is used to observe the rich set of near-zero GVD comb behaviors, including soliton molecules, switching waves (platicons) and their hybrids. Tuning of the 20 GHz repetition rate by electrical actuation enables servo locking to a microwave reference, which simultaneously stabilizes the comb repetition rate, offset frequency and temporal waveform. This hybridly integrated system could be used in coherent communications or for ultra-stable microwave signal generation by two-point optical frequency division., Comment: 8 pages, 4 figures

A one-dimensional lattice model with mosaic quasiperiodic potential is found to exhibit interesting localization properties, e.g., clear mobility edges [Y. Wang et al., Phys. Rev. Lett. \textbf{125}, 196604 (2020)]. We generalize this mosaic quasiperiodic model to a two-dimensional version, and numerically investigate its localization properties: the phase diagram from the fractal dimension of the wavefunction, the statistical and scaling properties of the conductance. Compared with disordered systems, our model shares many common features but also exhibits some different characteristics in the same dimensionality and the same universality class. For example, the sharp peak at $g\sim 0$ of the critical distribution and the large $g$ limit of the universal scaling function $\beta$ resemble those behaviors of three-dimensional disordered systems., Comment: 9 pages, 7 figures

Attempts are made to unify gravity with the other three fundamental forces of nature. As suggested by higher dimensional models, this unification may require space and time variation of some dimensionless fundamental constants. In this scenario, probing temporal variation of the electromagnetic fine structure constant ($\alpha= \frac{e^2} {\hbar c}$) in low energy regimes at the cosmological time scale is of immense interest. Atomic clocks are ideal candidates for probing $\alpha$ variation because their transition frequencies are measured to ultra-high precision accuracy. Since atomic transition frequencies are functions of $\alpha$, measurements of clock frequencies at different temporal and spatial locations can yield signatures to ascertain such conjecture. Electrons in highly charged ions (HCIs) experience unusually enhanced relativistic effects. Hence level-crossings can be observed often in these ions compared to their isoelectronic neutral or singly charged atomic systems. Such a process features by their more significant relativistic sensitive coefficients ($q$) of atomic transitions. For unambiguous detection of subtle changes in the transition frequencies due to $\alpha$ variation, it would be judicious to contemplate transitions for which $q$ values are enormous. HCIs are considered one of the most suitable candidates for making atomic clocks as they are the least sensitive to external electromagnetic fields owing to their exceptionally contracted orbitals. The first HCI clock has been realized, but its accuracy is much less than the counter optical clocks based on neutral atoms and singly charged ions. The realization of HCI clocks can add an extra dimension to investigating fundamental physics. In this work, we survey HCIs suitable for clock candidates on the grounds of general features, including their potential to probe temporal variation of $\alpha$., Comment: 36 pages, 11 tables

Chip-integrated optical frequency combs are attractive optical sources in comb applications requiring high-repetition-rate, low power consumption, or compact size. Spontaneous soliton formation via Kerr parametric oscillation is a promising generation principle in these frequency combs, and has been demonstrated in several material platforms over the past decade. Of these materials, AlGaAs has one of the largest Kerr nonlinearity coefficients allowing low pump threshold comb generation. However, bright soliton generation using this material has only been possible at cryogenic temperature because of the large thermo-optic effect at room temperature, which hinders stable access to the soliton regime. Here, we report self-stabilized single soliton generation in AlGaAs microresonators at room temperature by utilizing a rising soliton step in large free-spectral-range resonators. With sub-milliWatt optical pump power, 1 THz repetition-rate soliton generation is demonstrated. Perfect soliton crystal formation and soliton breather states are also observed. Besides the advantages of large optical nonlinearity, the devices are natural candidates for integration with III-V pump lasers., Comment: 7 pages, 3 figure

Deep Learning (DL) based Compressed Sensing (CS) has been applied for better performance of image reconstruction than traditional CS methods. However, most existing DL methods utilize the block-by-block measurement and each measurement block is restored separately, which introduces harmful blocking effects for reconstruction. Furthermore, the neuronal receptive fields of those methods are designed to be the same size in each layer, which can only collect single-scale spatial information and has a negative impact on the reconstruction process. This paper proposes a novel framework named Multi-scale Dilated Convolution Neural Network (MsDCNN) for CS measurement and reconstruction. During the measurement period, we directly obtain all measurements from a trained measurement network, which employs fully convolutional structures and is jointly trained with the reconstruction network from the input image. It needn't be cut into blocks, which effectively avoids the block effect. During the reconstruction period, we propose the Multi-scale Feature Extraction (MFE) architecture to imitate the human visual system to capture multi-scale features from the same feature map, which enhances the image feature extraction ability of the framework and improves the performance of image reconstruction. In the MFE, there are multiple parallel convolution channels to obtain multi-scale feature information. Then the multi-scale features information is fused and the original image is reconstructed with high quality. Our experimental results show that the proposed method performs favorably against the state-of-the-art methods in terms of PSNR and SSIM., Comment: 28 pages, 8 figures, MsDCNN for CS

We theoretically investigate quantum transport properties of quantum anomalous Hall bilayers, with arbitrary ratio of lattice constants, i.e., with lattice mismatch. In the simplest case of ratio 1 (but with different model parameters in two layers), the inter-layer coupling results in resonant traversing between forward propagating waves in two layers. In the case of generic ratios, there is a quantized conductance plateau originated from two Chern numbers associated with two layers. However, the phase boundary of this quantization plateau consists of a fractal transitional region (instead of a clear transition line) of interpenetrating edge states (with quantized conductance) and bulk states (with unquantized conductance). We attribute these bulk states as mismatch induced in-gap bulk states. Different from in-gap localized states induced by random disorder, these in-gap bulk states are extended in the limit of vanishing random disorder. However, the detailed fine structure of this transitional region is sensitive to disorder, lattice structure, sample size, and even the configuration of leads connecting to it, due to the bulk and topologically trivial nature of these in-gap bulk states., Comment: 16 pages, 15 figures

The real space formalism of orbital magnetization (OM) is an average of the local OM over some appropriate region of the system. Previous studies prefer a bulk average (i.e., without including boundaries). Based on a bilayer model with an adjustable Chern number at half filling, we numerically investigate the effects from boundaries on the real space expressions of OM. The size convergence processes of its three constituent terms $M_{\mathrm{LC}}$, $M_{\mathrm{IC}}$, $M_{\mathrm{BC}}$ are analysed. The topological term $M_{\mathrm{BC}}$ makes a nonnegligible contribution from boundaries as a manifestation of edge states, especially in the case of nonzero Chern numbers. However, we show that the influence of the boundary on $M_{\mathrm{LC}}$ and $M_{\mathrm{IC}}$ exactly compensates that on $M_{\mathrm{BC}}$. This compensation effect leads to the conclusion that the whole sample average is also a correct algorithm in the thermodynamic limit, which gives the same value as those from the bulk average and the $k$ space formula. This clarification will be beneficial to further studies on orbitronics, as well as the orbital magnetoelectric effects in higher dimensions., Comment: 12 pages, 12 figures

Optical lattice clocks are the prospective devices that can probe many subtle physics including temporal variation of the fine structure constant ($\alpha_e$). These studies necessitate high-precision measurements of atomic clock frequency ratios to unprecedented accuracy. In contrast to the earlier claimed highest sensitive coefficient ($K$) clock transition to $\alpha_e$ in Yb [Phys. Rev. Lett. $\textbf{120}$, 173001 (2018)], we found the $4f^{14}6s6p\,(^3P_2) - 4f^{13}5d6s^{2}\,(^3P_2^*)$ transition of this atom can serve as the clock transition with the largest $K$ value (about $-$25(2)). Moreover, we demonstrate a scheme to attain simultaneous magic trapping conditions for this clock transition with the other two proposed clock transitions $4f^{14}6s^2\,(^1S_0) - 4f^{14}6s6p\,(^3P_2)$ and $4f^{14}6s^2\,(^1S_0) - 4f^{13}5d6s^{2}\,(^3P_2^*)$; also exhibiting large $K$ values. This magic condition can be realized by subjecting Yb atoms to a bias magnetic field at a particular polarization angle along the quantization axis in an experimental set up. Upon realization, it will serve as the most potential optical lattice clock to probe $\alpha_e$ variation.

The black-box nature of Deep Neural Networks (DNNs) severely hinders its performance improvement and application in specific scenes. In recent years, class activation mapping-based method has been widely used to interpret the internal decisions of models in computer vision tasks. However, when this method uses backpropagation to obtain gradients, it will cause noise in the saliency map, and even locate features that are irrelevant to decisions. In this paper, we propose an Absolute value Class Activation Mapping-based (Abs-CAM) method, which optimizes the gradients derived from the backpropagation and turns all of them into positive gradients to enhance the visual features of output neurons' activation, and improve the localization ability of the saliency map. The framework of Abs-CAM is divided into two phases: generating initial saliency map and generating final saliency map. The first phase improves the localization ability of the saliency map by optimizing the gradient, and the second phase linearly combines the initial saliency map with the original image to enhance the semantic information of the saliency map. We conduct qualitative and quantitative evaluation of the proposed method, including Deletion, Insertion, and Pointing Game. The experimental results show that the Abs-CAM can obviously eliminate the noise in the saliency map, and can better locate the features related to decisions, and is superior to the previous methods in recognition and localization tasks., Comment: Abs-CAM for Explanation of Convolutional Neural Networks

Recently there have been proposals to understand features of QCD such as confinement and chiral symmetry breaking by considering supersymmetric versions of the theory with various patterns of soft breaking. In this note we recall that with small soft breakings, SUSY QCD is suggestive of observed features of real QCD. But we outline some of the challenges to establishing these features of the theory with large soft breakings. It appears difficult to argue for confinement or chiral symmetry breaking; at best, one can say that {\it if} the non-supersymmetric theory does not confine and/or break chiral symmetry, phase transitions would appear inevitable as one increases the soft breakings. We also discuss, at large N, where confinement would imply chiral symmetry breaking, the challenges to establishing confinement., Comment: 21 pages Footnote and references added

We consider some aspects of the Transplanckian Censorship Conjecture, which states that for theories of quantum gravity which yield a quintessence universe, there is a limit on the lifetime of the state not too different than the current Hubble horizon. We consider some aspects of tunneling in the presence of time-varying fields, and note that lifetimes can be parameterically quite large. We consider what it might mean to find a quintessence potential in string theory, and argue that such states are likely to have very long lifetimes., Comment: 11 pages. References added and repaired

Soliton microcombs based on Kerr nonlinearity in microresonators have been a prominent miniaturized coherent light source. Here, for the first time, we demonstrate the existence of Kerr solitons in an optomechanical microresonator, for which a nonlinear model is built by incorporating a single mechanical mode and multiple optical modes. Interestingly, an exotic vibrational Kerr soliton state is found, which is modulated by a self-sustained mechanical oscillation. Besides, the soliton provides extra mechanical gain through the optical spring effect, and results in phonon lasing with a red-detuned pump. Various nonlinear dynamics is also observed, including limit cycle, higher periodicity, and transient chaos. This work provides a guidance for not only exploring many-body nonlinear interactions, but also promoting precision measurements by featuring superiority of both frequency combs and optomechanics.

In this paper, we develop two finite difference weighted essentially non-oscillatory (WENO) schemes with unequal-sized sub-stencils for solving the Degasperis-Procesi (DP) and $\mu$-Degasperis-Procesi ($\mu$DP) equations, which contain nonlinear high order derivatives, and possibly peakon solutions or shock waves. By introducing auxiliary variable(s), we rewrite the DP equation as a hyperbolic-elliptic system, and the \mdp equation as a first order system. Then we choose a linear finite difference scheme with suitable order of accuracy for the auxiliary variable(s), and two finite difference WENO schemes with unequal-sized sub-stencils for the primal variable. One WENO scheme uses one large stencil and several smaller stencils, and the other WENO scheme is based on the multi-resolution framework which uses a series of unequal-sized hierarchical central stencils. Comparing with the classical WENO scheme which uses several small stencils of the same size to make up a big stencil, both WENO schemes with unequal-sized sub-stencils are simple in the choice of the stencil and enjoy the freedom of arbitrary positive linear weights. Another advantage is that the final reconstructed polynomial on the target cell is a polynomial of the same degree as the polynomial over the big stencil, while the classical finite difference WENO reconstruction can only be obtained for specific points inside the target interval. Numerical tests are provided to demonstrate the high order accuracy and non-oscillatory properties of the proposed schemes.

An active optical clock based on highly charged Nd$^{9+}$ ion is proposed for the first time. The clock can offer ultra-narrow linewidth at the $\upmu$Hz-level which is more than two-order of magnitude below the currently recorded laser linewidth. Operating at 605(90) nm superradiation lasing transition between the $5p^2~4f$ ground state and one of the long-lived $5p~4f^2$ excited state, the proposed active clock is inherently immune against the cavity noise which provides high stability. The clock serves as a sensitive probe with high sensitivity to variation of the fine-structure constant with accuracies below 10$^{-19}$ level. Sophisticated relativistic many-body methods are employed to predict related atomic properties that have corroborated the above findings., Comment: 5 papges, 2 figures, 1 SM

Dissipative soliton Kerr frequency combs in microresonators have recently been demonstrated with the self-injection locking process. They have the advantage of turnkey deterministic comb generation and simplifying dark soliton generation in the normal dispersion regime. Here, the formation process of dark pulses triggered by self-injection locking is studied by regarding them as a pair of domain walls that connect domains having different intracavity powers. The self-injection locking mechanism allows the domain walls to self-regulate their position so that a wide range of dark comb states can be accessed, and the duty cycle is controlled by the feedback phase. Direct imaging of the dark pulse shape using the electro-optic sampling technique is used to verify the theory. The results provide new physical insights as well as a new operational modality for this important class of nonlinear waves., Comment: 20 pages, 10 figures

WebFG 2020 is an international challenge hosted by Nanjing University of Science and Technology, University of Edinburgh, Nanjing University, The University of Adelaide, Waseda University, etc. This challenge mainly pays attention to the webly-supervised fine-grained recognition problem. In the literature, existing deep learning methods highly rely on large-scale and high-quality labeled training data, which poses a limitation to their practicability and scalability in real world applications. In particular, for fine-grained recognition, a visual task that requires professional knowledge for labeling, the cost of acquiring labeled training data is quite high. It causes extreme difficulties to obtain a large amount of high-quality training data. Therefore, utilizing free web data to train fine-grained recognition models has attracted increasing attentions from researchers in the fine-grained community. This challenge expects participants to develop webly-supervised fine-grained recognition methods, which leverages web images in training fine-grained recognition models to ease the extreme dependence of deep learning methods on large-scale manually labeled datasets and to enhance their practicability and scalability. In this technical report, we have pulled together the top WebFG 2020 solutions of total 54 competing teams, and discuss what methods worked best across the set of winning teams, and what surprisingly did not help., Comment: This is a technical report of the WebFG 2020 challenge (https://sites.google.com/view/webfg2020) associated with ACCV 2020

The emerging microresonator-based frequency combs revolutionize a broad range of applications from optical communications to astronomical calibration. Despite of their significant merits, low energy efficiency and the lack of all-optical dynamical control severely hinder the transfer of microcomb system to real-world applications. Here, by introducing active lasing medium into the soliton microcomb, for the first time, we experimentally achieve the chiral soliton with agile on-off switch and tunable dual-comb generation in a packaged microresonator. It is found that such a microresonator enables a soliton slingshot effect, the rapid soliton formation arising from the extra energy accumulation induced by inter-modal couplings. Moreover, tuning the erbium gain can generate versatile multi-soliton states, and extend the soliton operation window to a remarkable range over 18 GHz detuning. Finally, the gain-assisted chirality of counterpropagating soliton is demonstrated, which enables an unprecedented fast on-off switching of soliton microcombs. The non-trivial chiral soliton formation with active controllability inspires new paradigms of miniature optical frequency combs and brings the fast tunable soliton tools within reach., Comment: 11 pages, 8 figures

There have been many attempts to construct de Sitter space-times in string theory. While arguably there have been some successes, this has proven challenging, leading to the de Sitter swampland conjecture: quantum theories of gravity do not admit stable or metastable de Sitter space. Here we explain that, within controlled approximations, one lacks the tools to construct de Sitter space in string theory. Such approximations would require the existence of a set of (arbitrarily) small parameters, subject to severe constraints. But beyond this one also needs an understanding of big-bang and big-crunch singularities that is not currently accessible to standard approximations in string theory. The existence or non-existence of metastable de Sitter space in string theory remains a matter of conjecture., Comment: 20 pages. v2: references added; v3, v4: typos fixed

We demonstrate state-of-the-art technique of an active clock to provide a continuous superradiant lasing signal using an ensemble of trapped Cs atoms in the optical lattice. A magic wavelength of the proposed |7S1/2; F = 4, MF = 0> - |6P3/2; F = 3, MF = 0> clock transition in Cs atom is identified at 1181 nm for constructing the optical lattice. Pertinent optical lines are also found for pumping and repumping atoms from their ground states. A fractional uncertainty about 10-15 level to the clock frequency has been predicted by carrying out rigorous calculations of several atomic properties. The bad-cavity operational mode of the active clock is anticipated to improve its short-term stability remarkably by suppressing intrinsic thermal fluctuations. Thus, a composite clock system with better in both short-term and long-term stabilities can be built by combining the above proposed active clock with another high-accuracy passive optical clock.

The modern theory of orbital magnetization (OM) was developed by using Wannier function method, which has a formalism similar with the Berry phase. In this manuscript, we perform a numerical study on the fate of the OM under disorder, by using this method on the Haldane model in two dimensions, which can be tuned between a normal insulator or a Chern insulator at half filling. The effects of increasing disorder on OM for both cases are simulated. Energy renormalization shifts are observed in the weak disorder regime and topologically trivial case, which was predicted by a self-consistent T-matrix approximation. Besides this, two other phenomena can be seen. One is the localization trend of the band orbital magnetization. The other is the remarkable contribution from topological chiral states arising from nonzero Chern number or large value of integrated Berry curvature. If the fermi energy is fixed at the gap center of the clean system, there is an enhancement of |M| at the intermediate disorder, for both cases of normal and Chern insulators, which can be attributed to the disorder induced topological metal state before localization., Comment: 14 pages, 6 figures

Landauer-B\"uttiker formula describes the electronic quantum transports in nanostructures and molecules. It will be numerically demanding for simulations of complex or large size systems due to, for example, matrix inversion calculations. Recently, Chebyshev polynomial method has attracted intense interests in numerical simulations of quantum systems due to the high efficiency in parallelization, because the only matrix operation it involves is just the product of sparse matrices and vectors. Many progresses have been made on the Chebyshev polynomial representations of physical quantities for isolated or bulk quantum structures. Here we present the Chebyshev polynomial method to the typical electronic scattering problem, the Landauer-B\"uttiker formula for the conductance of quantum transports in nanostructures. We first describe the full algorithm based on the standard bath kernel polynomial method (KPM). Then, we present two simple butefficient improvements. One of them has a time consumption remarkably less than the direct matrix calculation without KPM. Some typical examples are also presented to illustrate the numerical effectiveness., Comment: 10 pages, 8 figures

We develop a method to extract the dissipation for a heavy moving impurity immersed in superfluid Fermi gases. The drag force is derived analytically. As a reward, we are able to extract the dynamical structure factor, from which density excitations of the system is carefully examined. We show that dissipations through drag force is associated with two types of excitations, one being single-particle and the other being collective. We map out the critical velocity for dissipation across the BEC-BCS crossover, consistent with existing experiments. For a magnetic impurity, we show that the dissipation is immune to collective excitations. Our study clearly manifests that dissipation and associated excitations can be controlled by coupling superfluid Fermi gases with a moving impurity, and paves the way for further exploring intriguing realm of nonequilibrium phenomena and dissipation dynamics., Comment: 6 pages, 5 figures

The evaporation of primordial black holes with a mass in the $1\ {\rm gram}\lesssim M_{\rm PBH}\lesssim$1000 kg range can lead to the production of dark matter particles of almost any mass in the range $0.1\ {\rm MeV}\lesssim m_{\rm DM}\lesssim 10^{18}$ GeV with the right relic density at very early times, $\tau\lesssim 10^{-10}$ s. We calculate, as a function of the primordial black holes mass and initial abundance, the combination of dark matter particle masses and number of effective dark degrees of freedom leading to the right abundance of dark matter today, whether or not evaporation stops around the Planck scale. In addition, since black hole evaporation can also lead to the production of a baryon asymmetry, we calculate where dark matter production and baryogenesis can concurrently happen, under a variety of assumptions: baryogenesis via grand unification boson decay, via leptogenesis, or via asymmetric co-genesis of dark matter and ordinary matter. Finally, we comment on possible ways to test this scenario., Comment: 29 pages, 4 figures, accepted for publication in JCAP

Quantum anomalous Hall (QAH) multilayers provide a platform of topological materials with high Chern numbers. We investigate the localization routes of bilayer QAH systems with Chern number C = 2 under strong disorder, by numerical simulations on their quantum transport properties and the Chern-Simons axion coupling. Compared to the single layer counterpart with C = 2, the localization trajectories present much richer behaviors, for example, the existence of the stable intermediate state with C = 1 can be tuned by model parameters. This state was always unstable in the single layer case. Furthermore, the two parameter scaling trajectories also exhibit multiple patterns, some of which were not captured by the standard Pruisken picture. During the process towards localization, the Chern-Simons axion coupling shows a surprisingly remarkable peak which is even higher and sharper in the large size limit. Therefore the disordered bilayer QAH system can be a good candidate for this nontrivial magnetoelectric coupling mediated by orbital motions., Comment: 11 pages, 11 figures

The production of the heavy quarkonium, i.e., $|(c\bar{b})[n]\rangle$ (or $|(b\bar{c})[n]\rangle$), $|(c\bar{c})[n]\rangle$, and $|(b\bar{b})[n]\rangle$- quarkonium [$|(Q\bar{Q'})[n]\rangle$-quarkonium for short], through Higgs $H^{0}$ boson semiexclusive decays is evaluated within the NRQCD framework, where $[n]$ stands for the production of the two color-singlet $S$-wave states, $|(Q\bar{Q'})[^1S_0]_{\textbf{1}} \rangle$ and $|(Q\bar{Q'})[^3S_1]_{\textbf{1}} \rangle$, the production of the four color-singlet $P$-wave states, i.e., $|(Q\bar{Q'})[^1P_0]_{\textbf{1}}\rangle$, $|(Q\bar{Q'})[^3P_J]_{\textbf{1}}\rangle$ (with $J =[0, 1, 2]$). Moreover, according to the velocity scaling rule of the NRQCD, the production of the two color-octet components, $|(Q\bar{Q'})g[^1S_0]_{\textbf{8}} \rangle$ and $|(Q\bar{Q'})g[^3S_1]_{\textbf{8}} \rangle$, are also taken into account. The "improved trace technology" to derive the simplified analytic expressions at the amplitude level is adopted, which shall be useful for dealing with these decay channels. If all higher heavy quarkonium states decay completely to the ground states, it should be obtained $\Gamma{(H^0\to |(c\bar{b})[^1S_0]_{\textbf{1}}\rangle)}=15.14$ KeV, $\Gamma{(H^0\to |(c\bar{c})[^1S_0]_{\textbf{1}}\rangle)}=1.547$ KeV, and $\Gamma{(H^0\to |(b\bar{b})[^1S_0]_{\textbf{1}}\rangle)}=1.311$ KeV. The production of $5.6\times10^{5}$ Bc meson, $4.7\times10^{4}$ charmonium meson, and $4.9\times10^{4}$ bottomonium meson per year in Higgs decays at the HE/HL-LHC can be obtained., Comment: 9 pages;5 figures. arXiv admin note: substantial text overlap with arXiv:1505.03275; text overlap with arXiv:0711.1898, arXiv:0805.4511, arXiv:1512.01327, arXiv:hep-ph/0502155, arXiv:1504.08236 by other authors

Three earlier relativistic coupled-cluster (RCC) calculations of dipole polarizability ($\alpha_d$) of the Cd atom are not in good agreement with the available experimental value of $49.65(1.65) \ e a_0^3$. Among these two are finite-field approaches in which the relativistic effects have been included approximately, while the other calculation uses a four component perturbed RCC method. However, another work adopting an approach similar to the latter perturbed RCC method gives a result very close to that of experiment. The major difference between these two perturbed RCC approaches lies in their implementation. To resolve this ambiguity, we have developed and employed the relativistic normal coupled-cluster (RNCC) theory to evaluate the $\alpha_d$ value of Cd. The distinct features of the RNCC method are that the expression for the expectation value in this approach terminates naturally and that it satisfies the Hellmann-Feynman theorem. In addition, we determine this quantity in the finite-field approach in the framework of A four-component relativistic coupled-cluster theory. Considering the results from both these approaches, we arrive at a reliable value of $\alpha_d=46.02(50) \ e a_0^3$. We also demonstrate that the contribution from the triples excitations in this atom is significant., Comment: 10 pages, 4 tables, 1 figure; Accepted in PRA

Roles of electron correlation effects in the determination of attachment energies, magnetic dipole hyperfine structure constants and electric dipole (E1) matrix elements of the low-lying states in the singly charged cadmium ion (Cd$^+$) have been analyzed. We employ the singles and doubles approximated relativistic coupled-cluster (RCC) method to calculate these properties. Intermediate results from the Dirac-Hartree-Fock approximation, second-order many-body perturbation theory and considering only the linear terms of the RCC method are given to demonstrate propagation of electron correlation effects in this ion. Contributions from important RCC terms are also given to highlight importance of various correlation effects in the evaluation of these properties. At the end, we also determine E1 polarizabilities ($\alpha^{E1}$) of the ground and $5p \ ^2P_{1/2;3/2}$ states of Cd$^+$ in the {\it ab initio} approach. We estimate them again by replacing some of the E1 matrix elements and energies from the measurements to reduce their uncertainties so that they can be used in the high precision experiments of this ion., Comment: 10 pages, 7 tables; Accepted in Phys. Rev. A

We investigate phase separation and hidden vortices in spin-orbit coupled ferromagnetic BoseEinstein condensates with rotation and Rabi coupling. The hidden vortices are invisible in density distribution but are visible in phase distribution, which can carry angular momentum like the ordinary quantized vortices. In the absence of the rotation, we observe the phase separation induced by the spin-orbit coupling and determine the entire phase diagram of the existence of phase separation. For the rotation case, in addition to the phase separation, we demonstrate particularly that the spin-orbit coupling can result in the hidden vortices and hidden vortex-antivortex pairs. The corresponding entire phase diagrams are determined, depending on the interplay of the spin-orbit coupling strength, the rotation frequency, and Rabi frequency, which reveals the critical condition of the occurrence of the hidden vortices and vortex-antivortex pairs. The hidden vortices here are proved to be long-lived in the time scale of experiment by the dynamic analysis. These findings not only provide a clear illustration of the phase separation in spin-orbit coupled spinor Bose-Einstein condensates, but also open a new direction for investigating the hidden vortices in high-spin quantum system., Comment: 8 pages, 10 figures

We propose here a few selective highly charged ions (HCIs), namely Ni$^{12+}$ and Cu$^{13+}$, Pd$^{12+}$ and Ag$^{13+}$, that not only promise to be very high accurate optical clocks below $10^{-19}$ uncertainties, but also offer quality factors larger than $10^{15}$ and yet possess simple atomic structures for the experimental set-up. Moreover, these ions have strong optical magnetic-dipole (M1) transitions than the previously proposed HCI clocks. They can be used for the cooling and detection techniques. To demonstrate the projected fractional uncertainties below $10^{-19}$ level, we have estimated the typical orders of magnitudes due to many of the conventional systematics manifested in an atomic clock experiment, such as Zeeman, Stark, black-body radiation, and electric quadrupole shifts, by performing calculations of the relevant atomic properties., Comment: 5 pages

The static and dynamic electric-dipole polarizabilities of the $6s^2\,^1S_0$ and $6s6p\,^3P_1^o$ states of Yb are calculated by using the relativistic ab initio method. Focusing on the red detuning region to the $6s^2\,^1S_0-6s6p\,^3P_1^o$ transition, we find two magic wavelengths at 1035.7(2) nm and 612.9(2) nm for the $6s^2\,^1S_0-6s6p\,^3P_1^o, M_J=0$ transition and three magic wavelengthes at 1517.68(6) nm, 1036.0(3) nm and 858(12) nm for the $6s^2\,^1S_0-6s6p\,^3P_1^o, M_J=\pm1$ transitions. Such magic wavelengths are of particular interest for attaining the state-insensitive cooling, trapping, and quantum manipulation of neutral Yb atom., Comment: 13 pages, 3 figures

We have investigated the in-gap bound states (IGBS) induced by a single nonmagnetic impurity in multiband superconductors with incipient bands. Contrary to the naive expectation, we found that even if the superconducting (SC) order parameter is sign-preserving s-wave on the Fermi surfaces, the incipient bands may still affect the appearance and locations of the IGBS, although the gap between the incipient bands and the Fermi level is much larger than the SC gap. Therefore in scanning tunneling microscopy experiments, the IGBS induced by a single nonmagnetic impurity are not the definitive evidences for the sign-changing order parameter on the Fermi surfaces. Our findings have special implications for the experimental determination of the pairing symmetry in the FeSe-based superconductors., Comment: To be published in Phys. Rev. B Rapid Commun

The microwave clock frequency of the $|5s~^2S_{1/2}, F=0,m_F=0 \rangle \leftrightarrow |5s~^2S_{1/2}, F=1,m_F=0 \rangle$ transition in the $^{113}$Cd$^+$ ion has been reported as 15199862855.0192(10) Hz [Opt. Lett. {\bf 40}, 4249 (2015)]. Fractional systematic due to the black-body radiation (BBR) shift ($\beta$) arising from the Stark effect in the above clock transition was used as $-1.1 \times 10^{-16}$ from our unpublished preliminary estimation. We present here a precise value as $\beta=-1.815(77) \times 10^{-16}$ by carrying out rigorous calculations of third-order polarizabilities of the hyperfine levels associated with the clock transition. This is determined by evaluating matrix elements of the magnetic dipole hyperfine interaction Hamiltonian, electric dipole operator and energies between many low-lying states of $^{113}$Cd$^+$. We employ all-order relativistic many-body theories in the frameworks of Fock-space coupled-cluster and relativistic multi-configuration Dirac-Fock methods., Comment: 6 pages

We introduce a new class of mean regression estimators -- penalized maximum tangent likelihood estimation -- for high-dimensional regression estimation and variable selection. We first explain the motivations for the key ingredient, maximum tangent likelihood estimation (MTE), and establish its asymptotic properties. We further propose a penalized MTE for variable selection and show that it is $\sqrt{n}$-consistent, enjoys the oracle property. The proposed class of estimators consists penalized $\ell_2$ distance, penalized exponential squared loss, penalized least trimmed square and penalized least square as special cases and can be regarded as a mixture of minimum Kullback-Leibler distance estimation and minimum $\ell_2$ distance estimation. Furthermore, we consider the proposed class of estimators under the high-dimensional setting when the number of variables $d$ can grow exponentially with the sample size $n$, and show that the entire class of estimators (including the aforementioned special cases) can achieve the optimal rate of convergence in the order of $\sqrt{\ln(d)/n}$. Finally, simulation studies and real data analysis demonstrate the advantages of the penalized MTE., Comment: 30 pages, 3 figures