arXiv subjects

Wei Zhu

Publications and source records attributed to Wei Zhu.

At least 217 records · Page 12Linked to original sources

arXiv2021

Measuring microlensing parallax via simultaneous observations from Chinese Space Station Telescope and Roman Telescope

Simultaneous observations from two spatially well-separated telescopes can lead to the measurements of the microlensing parallax parameter, an important quantity toward the determinations of the lens mass. The separation between Earth and Sun-Earth L2 point, $\sim0.01$ AU, is ideal for parallax measurements of short and ultra-short ($\sim$1\,hr to 10\,days) microlensing events, which are candidates of free-floating planet (FFP) events. In this work, we study the potential of doing so in the context of two proposed space-based missions, the Chinese Space Station Telescope (CSST) in a Leo orbit and the Nancy Grace Roman Space Telescope (\emph{Roman}) at L2. We show that the joint observations of the two can directly measure the microlensing parallax of nearly all FFP events with timescales $t_{\rm E}\lesssim$ 10\,days as well as planetary (and stellar binary) events that show caustic crossing features. The potential of using CSST alone in measuring microlensing parallax is also discussed.

astro-ph.EP↗

arXiv2021

Route Optimization via Environment-Aware Deep Network and Reinforcement Learning

Vehicle mobility optimization in urban areas is a long-standing problem in smart city and spatial data analysis. Given the complex urban scenario and unpredictable social events, our work focuses on developing a mobile sequential recommendation system to maximize the profitability of vehicle service providers (e.g., taxi drivers). In particular, we treat the dynamic route optimization problem as a long-term sequential decision-making task. A reinforcement-learning framework is proposed to tackle this problem, by integrating a self-check mechanism and a deep neural network for customer pick-up point monitoring. To account for unexpected situations (e.g., the COVID-19 outbreak), our method is designed to be capable of handling related environment changes with a self-adaptive parameter determination mechanism. Based on the yellow taxi data in New York City and vicinity before and after the COVID-19 outbreak, we have conducted comprehensive experiments to evaluate the effectiveness of our method. The results show consistently excellent performance, from hourly to weekly measures, to support the superiority of our method over the state-of-the-art methods (i.e., with more than 98% improvement in terms of the profitability for taxi drivers).

cs.LG↗

arXiv2021

OGLE-2019-BLG-0468Lb,c: two microlensing giant planets around a G-type star

With the aim of interpreting anomalous lensing events with no suggested models, we conducted a project of reinvestigating microlensing data in and before the 2019 season. In this work, we report a multi-planet system OGLE-2019-BLG-0468L found from the project. The light curve of the lensing event OGLE-2019-BLG-0468, which consists of three distinctive anomaly features, could not be explained by the usual binary-lens or binary-source interpretation. We find a solution explaining all anomaly features with a triple-lens interpretation, in which the lens is composed of two planets and their host, making the lens the fourth multi-planet system securely found by microlensing. The two planets have masses $\sim 3.4~M_{\rm J}$ and $\sim 10.2~M_{\rm J}$, and they are orbiting around a G-type star with a mass $\sim 0.9~M_\odot$ and a distance $\sim 4.4$ kpc. The host of the planets is most likely responsible for the light of the baseline object, although the possibility for the host to be a companion to the baseline object cannot be ruled out.

astro-ph.EP↗

arXiv2021

Systematic KMTNet Planetary Anomaly Search, Paper II: Six New $q<2\times 10^{-4}$ Mass-ratio Planets

We apply the automated AnomalyFinder algorithm of Paper I (Zang et al. 2021b) to 2018-2019 light curves from the $\simeq 13\,{\rm deg}^2$ covered by the six KMTNet prime fields, with cadences $Γ\geq 2\,{\rm hr}^{-1}$. We find a total of 11 planets with mass ratios $q<2\times 10^{-4}$, including six newly discovered planets, one planet that was reported in Paper I, and recovery of four previously discovered planets. One of the new planets, OGLE-2018-BLG-0977Lb, is in a planetary-caustic event, while the other five (OGLE-2018-BLG-0506Lb, OGLE-2018-BLG-0516Lb, OGLE-2019-BLG-1492Lb, KMT-2019-BLG-0253, and KMT-2019-BLG-0953) are revealed by a "dip" in the light curve as the source crosses the host-planet axis on the opposite side of the planet. These subtle signals were missed in previous by-eye searches. The planet-host separations (scaled to the Einstein radius), $s$, and planet-host mass ratios, $q$, are, respectively, $(s,q\times 10^5) = (0.88, 4.1)$, $(0.96\pm 0.10, 8.3)$, $(0.94\pm 0.07, 13)$, $(0.97\pm 0.07, 18)$, $(0.97\pm0.04,4.1)$, and $(0.74,18)$, where the "$\pm$" indicates a discrete degeneracy. The 11 planets are spread out over the range $-5<\log q < -3.7$. Together with the two planets previously reported with $q\sim 10^{-5}$ from the 2018-2019 non-prime KMT fields, this result suggests that planets toward the bottom of this mass-ratio range may be more common than previously believed.

astro-ph.EP↗

arXiv2021

Neural Networks Enforcing Physical Symmetries in Nonlinear Dynamical Lattices: The Case Example of the Ablowitz-Ladik Model

In this work we introduce symmetry-preserving, physics-informed neural networks (S-PINNs) motivated by symmetries that are ubiquitous to solutions of nonlinear dynamical lattices. Although the use of PINNs have recently attracted much attention in data-driven discovery of solutions chiefly to partial differential equations, we demonstrate that they fail at enforcing important physical laws including symmetries of solutions and conservation laws. Through the correlation of parity symmetries in both space and time of solutions to differential equations with their group equivariant representation, we construct group-equivariant NNs which respect spatio-temporal parity symmetry. Moreover, we adapt the proposed architecture to enforce different types of periodicity (or localization) of solutions to nonlinear dynamical lattices. We do so by applying S-PINNs to the completely integrable Ablowitz-Ladik model, and performing numerical experiments with a special focus on waveforms that are related to rogue structures. These include the Kuznetsov-Ma soliton, and Akhmediev breather as well as the Peregrine soliton. Our numerical results demonstrate the superiority and robustness of the proposed architecture over standard PINNs.

nlin.PS↗

arXiv2021

The current recorded signals of ultrahigh-energy $γ$-rays may come from EeVatrons in the galaxy

A hard $γ$-ray spectrum of supernova remnant G106.3+2.7 can be explained by using the hadronic model with the gluon condensation effect. This implies that not only PeVatrons but also EeVatrons generally exist in the universe including our galaxy, and they can accelerate protons to beyond "ankle" ($10^{19}~eV$). Although these proton beams are very weak in the galaxy and cannot be observed individually on the earth, the gluon condensation effect may greatly enhance the proton-proton cross section, which can compensate for the weak proton flux and produce the observed $γ$-rays. We also show that the gluon condensation effect in proton provides an efficient conversion mechanism for kinetic energy into $γ$-rays in the universe.

astro-ph.HE↗

arXiv2021

Learning Bias-Invariant Representation by Cross-Sample Mutual Information Minimization

Deep learning algorithms mine knowledge from the training data and thus would likely inherit the dataset's bias information. As a result, the obtained model would generalize poorly and even mislead the decision process in real-life applications. We propose to remove the bias information misused by the target task with a cross-sample adversarial debiasing (CSAD) method. CSAD explicitly extracts target and bias features disentangled from the latent representation generated by a feature extractor and then learns to discover and remove the correlation between the target and bias features. The correlation measurement plays a critical role in adversarial debiasing and is conducted by a cross-sample neural mutual information estimator. Moreover, we propose joint content and local structural representation learning to boost mutual information estimation for better performance. We conduct thorough experiments on publicly available datasets to validate the advantages of the proposed method over state-of-the-art approaches.

cs.LG↗

arXiv2021

Variational study of the Kitaev-Heisenberg-Gamma model

We compute the low-energy excitation spectrum and the dynamical spin structure factor of the Kitaev-Heisenberg-Gamma model through a variational approach based on the exact fractionalized excitations of the pure Kitaev honeycomb model. This novel approach reveals the physical reason for the asymmetric stability of the Kitaev spin liquid phases around the ferromagnetic and antiferromagnetic Kitaev limits. Moreover, we demonstrate that the fractionalized excitations form bound states in specific regions of each Kitaev spin liquid phase and that certain phase transitions induced by Heisenberg and Gamma interactions are driven by the condensation of such a bound state. Remarkably, this bound state appears as a sharp mode in the dynamical spin structure factor, while its condensation patterns at the appropriate phase transitions provide a simple explanation for the magnetically ordered phases surrounding each Kitaev spin liquid phase.

cond-mat.str-el↗

arXiv2021

AutoTrans: Automating Transformer Design via Reinforced Architecture Search

Though the transformer architectures have shown dominance in many natural language understanding tasks, there are still unsolved issues for the training of transformer models, especially the need for a principled way of warm-up which has shown importance for stable training of a transformer, as well as whether the task at hand prefer to scale the attention product or not. In this paper, we empirically explore automating the design choices in the transformer model, i.e., how to set layer-norm, whether to scale, number of layers, number of heads, activation function, etc, so that one can obtain a transformer architecture that better suits the tasks at hand. RL is employed to navigate along search space, and special parameter sharing strategies are designed to accelerate the search. It is shown that sampling a proportion of training data per epoch during search help to improve the search quality. Experiments on the CoNLL03, Multi-30k, IWSLT14 and WMT-14 shows that the searched transformer model can outperform the standard transformers. In particular, we show that our learned model can be trained more robustly with large learning rates without warm-up.

cs.CL↗

arXiv2021

CMV-BERT: Contrastive multi-vocab pretraining of BERT

In this work, we represent CMV-BERT, which improves the pretraining of a language model via two ingredients: (a) contrastive learning, which is well studied in the area of computer vision; (b) multiple vocabularies, one of which is fine-grained and the other is coarse-grained. The two methods both provide different views of an original sentence, and both are shown to be beneficial. Downstream tasks demonstrate our proposed CMV-BERT are effective in improving the pretrained language models.

cs.CL↗

arXiv2021

Heisenberg-Limited Waveform Estimation with Solid-State Spins in Diamond

The newly established Heisenberg limit in arbitrary waveform estimation is quite different with parameter estimation and shows a unique characteristic of a future quantum version of oscilloscope. However, it is still a non-trivial challenge to generate a large number of exotic quantum entangled states to achieve this quantum limit. Here, by employing the time-domain quantum difference detection method, we demonstrate Heisenberg-limited waveform quantum estimation with diamond spins under ambient condition in the experiment. Periodic dynamical decoupling is applied to enhance both the dynamic range and sensitivity by one order of magnitude. Using this quantum-enhanced estimation scheme, the estimation error of an unknown waveform is reduced by more than $5$ dB below the standard quantum limit with $N\sim{\text{2}} \times {\text{1}}{\text{0}^3}$ resources, where more than ${1 \times {\text{1}}{\text{0}^5}}$ resources would be required to achieve a similar error level using classical detection. This work provides an essential step towards realizing quantum-enhanced structure recognition in a continuous space and time.

quant-ph↗

arXiv2021

Lex-BERT: Enhancing BERT based NER with lexicons

In this work, we represent Lex-BERT, which incorporates the lexicon information into Chinese BERT for named entity recognition (NER) tasks in a natural manner. Instead of using word embeddings and a newly designed transformer layer as in FLAT, we identify the boundary of words in the sentences using special tokens, and the modified sentence will be encoded directly by BERT. Our model does not introduce any new parameters and are more efficient than FLAT. In addition, we do not require any word embeddings accompanying the lexicon collection. Experiments on Ontonotes and ZhCrossNER show that our model outperforms FLAT and other baselines.

cs.CL↗

arXiv2021

Constraining the Circumbinary Disk Tilt in the KH 15D system

KH 15D is a system which consists of a young, eccentric binary, and a circumbinary disk which obscures the binary as the disk precesses. We develop a self-consistent model that provides a reasonable fit to the photometric variability that was observed in the KH 15D system over the past 60 years. Our model suggests that the circumbinary disk has an inner edge $r_{\rm in}\lesssim 1 \ {\rm au}$, an outer edge $r_{\rm out} \sim {\rm a \ few \ au}$, and that the disk is misaligned relative to the stellar binary by $\sim$5-16 degrees, with the inner edge more inclined than the outer edge. The difference between the inclinations (warp) and longitude of ascending nodes (twist) at the inner and outer edges of the disk are of order $\sim$10 degrees and $\sim$15 degrees, respectively. We also provide constraints on other properties of the disk, such as the precession period and surface density profile. Our work demonstrates the power of photometric data in constraining the physical properties of planet-forming circumbinary disks.

astro-ph.EP↗

arXiv2021

Exoplanet Statistics and Theoretical Implications

In the last few years, significant advances have been made in understanding the distributions of exoplanet populations and the architecture of planetary systems. We review the recent progress of planet statistics, with a focus on the inner <~ 1 AU region of the planetary system that has been fairly thoroughly surveyed by the Kepler mission. We also discuss the theoretical implications of these statistical results for planet formation and dynamical evolution.

astro-ph.EP↗

arXiv2021

An Earth-mass Planet in a Time of Covid-19: KMT-2020-BLG-0414Lb

We report the discovery of KMT-2020-BLG-0414Lb, with a planet-to-host mass ratio $q_2 = 0.9$--$1.2 \times 10^{-5} = 3$--$4~q_{\oplus}$ at $1σ$, which is the lowest mass-ratio microlensing planet to date. Together with two other recent discoveries ($4 \lesssim q/q_\oplus \lesssim 6$), it fills out the previous empty sector at the bottom of the triangular $(\log s, \log q)$ diagram, where $s$ is the planet-host separation in units of the angular Einstein radius $θ_{\rm E}$. Hence, these discoveries call into question the existence, or at least the strength, of the break in the mass-ratio function that was previously suggested to account for the paucity of very low-$q$ planets. Due to the extreme magnification of the event, $A_{\rm max}\sim 1450$ for the underlying single-lens event, its light curve revealed a second companion with $q_3 \sim 0.05$ and $|\log s_3| \sim 1$, i.e., a factor $\sim 10$ closer to or farther from the host in projection. The measurements of the microlens parallax $π_{\rm E}$ and the angular Einstein radius $θ_{\rm E}$ allow estimates of the host, planet, and second companion masses, $(M_1, M_2, M_3) \sim (0.3M_{\odot}, 1.0M_{\oplus}, 17M_{J})$, the planet and second companion projected separations, $(a_{\perp,2}, a_{\perp,3}) \sim (1.5, 0.15~{\rm or}~15)$~au, and system distance $D_{\rm L} \sim 1$ kpc. The lens could account for most or all of the blended light ($I \sim 19.3$) and so can be studied immediately with high-resolution photometric and spectroscopic observations that can further clarify the nature of the system. The planet was found as part of a new program of high-cadence follow-up observations of high-magnification events. The detection of this planet, despite the considerable difficulties imposed by Covid-19 (two KMT sites and OGLE were shut down), illustrates the potential utility of this program.

astro-ph.EP↗

arXiv2021

Temperature dependent coherence properties of NV ensemble in diamond up to 600K

Nitrogen-vacancy (NV) center in diamond is an ideal candidate for quantum sensors because of its excellent optical and coherence property. However, previous studies are usually conducted at low or room temperature. The lack of full knowledge of coherence properties of the NV center at high temperature limits NV's further applications. Here, we systematically explore the coherence properties of the NV center ensemble at temperatures from 300 K to 600 K. Coherence time $T_2$ decreases rapidly from $184 μs$ at 300 K to $30 μs$ at 600 K, which is attributed to the interaction with paramagnetic impurities. Single-quantum and double-quantum relaxation rates show an obvious temperature-dependent behavior as well, and both of them are dominated by the two phonon Raman process. While the inhomogeneous dephasing time $T_2^*$ and thermal echo decoherence time $T_{TE}$ remain almost unchanged as temperature rises. Since $T_{TE}$ changed slightly as temperature rises, a thermal-echo-based thermometer is demonstrated to have a sensitivity of $41 mK/\sqrt{Hz}$ at 450 K. These findings will help to pave the way toward NV-based high-temperature sensing, as well as to have a more comprehensive understanding of the origin of decoherence in the solid-state qubit.

quant-ph↗

arXiv2021

Experimental implementation of universal holonomic quantum computation on solid-state spins with optimal control

Experimental realization of a universal set of quantum logic gates with high-fidelity is critical to quantum information processing, which is always challenging by inevitable interaction between the quantum system and environment. Geometric quantum computation is noise immune, and thus offers a robust way to enhance the control fidelity. Here, we experimentally implement the recently proposed extensible nonadiabatic holonomic quantum computation with solid spins in diamond at room-temperature, which maintains both flexibility and resilience against decoherence and system control errors. Compared with previous geometric method, the fidelities of a universal set of holonomic single-qubit and two-qubit quantum logic gates are improved in experiment. Therefore, this work makes an important step towards fault-tolerant scalable geometric quantum computation in realistic systems.

quant-ph↗

arXiv2021

OGLE-2018-BLG-0567Lb and OGLE-2018-BLG-0962Lb: Two Microlensing Planets through Planetary-Caustic Channel

We present the analyses of two microlensing events, OGLE-2018-BLG-0567 and OGLE-2018-BLG-0962. In both events, the short-lasting anomalies were densely and continuously covered by two high-cadence surveys. The light-curve modeling indicates that the anomalies are generated by source crossings over the planetary caustics induced by planetary companions to the hosts. The estimated planet/host separation (scaled to the angular Einstein radius $θ_{\rm E}$) and mass ratio are $(s, q) = (1.81, 1.24\times10^{-3})$ and $(s, q) = (1.25, 2.38\times10^{-3})$, respectively. From Bayesian analyses, we estimate the host and planet masses as $(M_{\rm h}, M_{\rm p}) = (0.24_{-0.13}^{+0.16}\,M_{\odot}, 0.32_{-0.16}^{+0.34}\,M_{\rm J})$ and $(M_{\rm h}, M_{\rm p}) = (0.55_{-0.29}^{+0.32}\,M_{\odot}, 1.37_{-0.72}^{+0.80}\,M_{\rm J})$, respectively. These planetary systems are located at a distance of $7.07_{-1.15}^{+0.93}\,{\rm kpc}$ for OGLE-2018-BLG-0567 and $6.47_{-1.73}^{+1.04}\,{\rm kpc}$ for OGLE-2018-BLG-0962, suggesting that they are likely to be near the Galactic bulge. The two events prove the capability of current high-cadence surveys for finding planets through the planetary-caustic channel. We find that most published planetary-caustic planets are found in Hollywood events in which the source size strongly contributes to the anomaly cross section relative to the size of the caustic.

astro-ph.EP↗