arXiv subjects

Xiaoguang Wang

Publications and source records attributed to Xiaoguang Wang.

At least 73 records · Page 4Linked to original sources

arXiv2019

Photon-assisted entanglement and squeezing generation and decoherence suppression via a quadratic optomechanical coupling

Entanglement and quantum squeezing have wide applications in quantum technologies due to their non-classical characteristics. Here we study entanglement and quantum squeezing in an open spin-optomechanical system, in which a Rabi model (a spin coupled to the mechanical oscillator) is coupled to an ancillary cavity field via a quadratic optomechanical coupling. We find that their performances can be significantly modulated via the photon of the ancillary cavity, which comes from photon-dependent spin-oscillator coupling and detuning. Specifically, a fully switchable spin-oscillator entanglement can be achieved, meanwhile a strong mechanical squeezing is also realized. Moreover, we study the environment-induced decoherence and dissipation, and find that they can be mitigated by increasing the number of photons. This work provides an effective way to manipulate entanglement and quantum squeezing and to suppress decoherence in the cavity quantum electrodynamics with a quadratic optomechanics.

quant-ph↗

arXiv2019

CUTIE: Learning to Understand Documents with Convolutional Universal Text Information Extractor

Extracting key information from documents, such as receipts or invoices, and preserving the interested texts to structured data is crucial in the document-intensive streamline processes of office automation in areas that includes but not limited to accounting, financial, and taxation areas. To avoid designing expert rules for each specific type of document, some published works attempt to tackle the problem by learning a model to explore the semantic context in text sequences based on the Named Entity Recognition (NER) method in the NLP field. In this paper, we propose to harness the effective information from both semantic meaning and spatial distribution of texts in documents. Specifically, our proposed model, Convolutional Universal Text Information Extractor (CUTIE), applies convolutional neural networks on gridded texts where texts are embedded as features with semantical connotations. We further explore the effect of employing different structures of convolutional neural network and propose a fast and portable structure. We demonstrate the effectiveness of the proposed method on a dataset with up to $4,484$ labelled receipts, without any pre-training or post-processing, achieving state of the art performance that is much better than the NER based methods in terms of either speed and accuracy. Experimental results also demonstrate that the proposed CUTIE model being able to achieve good performance with a much smaller amount of training data.

cs.CV↗

arXiv2019

Powerful harmonic charging in quantum batteries

We consider a classical harmonic driving field as the energy charger for the quantum batteries, which consist of an ensemble of two-level atoms. The maximum stored energy and the final state are derived analytically with the optimal driving frequency. At the end of charging procedure, each of atoms is in the upper state and the batteries are charging completely, which exhibits a substantial improvement over the square-wave charger. Involving the interatomic correlations, we find that the repulsive couplings show an advantage in achieving fully charging with shorter charging period. However, the attractive interactions induce a negative effects on the charging, since the ground state undergoes a quantum phase transition from a separable state to a doubly degenerate state. Approaching to the phase transition regime, the maximum stored energy drops sharply from the fully-charging value. The phase transition favors to suppress the charging of the battery and prevents the final state to be a separable state due to quantum fluctuations in our quantum batteries.

quant-ph↗

arXiv2019

Multiqubit matter-wave interferometry under decoherence and the Heisenberg scaling recovery

Most matter-wave interferometry (MWI) schemes for quantum sensing are so far evaluated in ideal situations without noises. In this work, we provide assessments of generic multiqubit MWI schemes under Markovian dephasing noises. We find that for certain classes of the MWI schemes with scale factors that are nonlinearly dependent on the interrogation time, the optimal precision of maximally entangled probes \emph{decreases} with increasing the particle number $N$, for both independent and collective dephasing situations. This result challenges the conventional wisdom found in dephasing Ramsey-type interferometers. We initiate the analyses by investigating the optimal precision of multiqubit Sagnac atom interferometry for rotation sensing. And we show that due to the competition between the unconventional interrogation-time quadratic phase accumulation and the exponential dephasing processes, the Greenberger-Horne-Zeilinger (GHZ) state, which is the optimal input state in noiseless scenarios, leads to vanishing quantum Fisher information in the large-$N$ regime. Then our assessments are further extended to generic MWI schemes for quantum sensing with entangled states and under decoherence. Finally, a quantum error-correction logical GHZ state is tentatively analyzed, which could have the potential to recover the Heisenberg scaling and improve the sensitivity.

quant-ph↗

arXiv2018

Improving the Interpretability of Deep Neural Networks with Knowledge Distillation

Deep Neural Networks have achieved huge success at a wide spectrum of applications from language modeling, computer vision to speech recognition. However, nowadays, good performance alone is not sufficient to satisfy the needs of practical deployment where interpretability is demanded for cases involving ethics and mission critical applications. The complex models of Deep Neural Networks make it hard to understand and reason the predictions, which hinders its further progress. To tackle this problem, we apply the Knowledge Distillation technique to distill Deep Neural Networks into decision trees in order to attain good performance and interpretability simultaneously. We formulate the problem at hand as a multi-output regression problem and the experiments demonstrate that the student model achieves significantly better accuracy performance (about 1\% to 5\%) than vanilla decision trees at the same level of tree depth. The experiments are implemented on the TensorFlow platform to make it scalable to big datasets. To the best of our knowledge, we are the first to distill Deep Neural Networks into vanilla decision trees on multi-class datasets.

cs.LG↗

arXiv2018

Dynamics of Newton maps

In this paper, we study the dynamics of Newton maps for arbitrary polynomials. Let $p$ be an arbitrary polynomial with at least three distinct roots, and $f$ be its Newton map. It is shown that the boundary $\partial B$ of any immediate root basin $B$ of $f$ is locally connected. Moreover, $\partial B$ is a Jordan curve if and only if ${\rm deg}(f|_B)=2$. This implies that the boundaries of all components of root basins, for all polynomials' Newton maps, from the viewpoint of topology, are tame.

math.DS↗

arXiv2018

Enhancing the sensitivity of rotation in a multi-atom Sagnac interferometer

We investigate quantum sensing of rotation with a multi-atom Sagnac interferometer and present multi-partite entangled states to enhance the sensitivity of rotation frequency. For studying the sensitivity, we first present a Hermitian generator with respect to the rotation frequency. The generator, which contains the Sagnac phase, is a linear superposition of a z component of the collective spin and a quadrature operator of collective bosons depicting the trapping modes, which enables us to conveniently study the quantum Fisher information (QFI) for any initial states. With the generator, we derive the general QFI which can be of square dependence on the particle number, leading to Heisenberg limit. And we further find that the QFI may be of biquadratic dependence on the radius of the ring which confines atoms, indicating that larger QFI is achieved by enlarging the radius. In order to obtain the square and biquadratic dependence, we propose to use partially and globally entangled states as inputs to enhance the sensitivity of rotation.

quant-ph↗

arXiv2018

Phase-space geometric Sagnac interferometer for rotation sensing

Quantum information processing with geometric features of quantum states may provide promising noise-resilient schemes for quantum metrology. In this work, we theoretically explore phase-space geometric Sagnac interferometers with trapped atomic clocks for rotation sensing, which could be intrinsically robust to certain decoherence noises and reach high precision. With the wave guide provided by sweeping ring-traps, we give criteria under which the well-known Sagnac phase is a pure or unconventional geometric phase with respect to the phase space. Furthermore, corresponding schemes for geometric Sagnac interferometers with designed sweeping angular velocity and interrogation time are presented, and the experimental feasibility is also discussed. Such geometric Sagnac interferometers are capable of saturating the ultimate precision limit given by the quantum Cramér-Rao bound.

quant-ph↗

arXiv2018

Interpretable Deep Convolutional Neural Networks via Meta-learning

Model interpretability is a requirement in many applications in which crucial decisions are made by users relying on a model's outputs. The recent movement for "algorithmic fairness" also stipulates explainability, and therefore interpretability of learning models. And yet the most successful contemporary Machine Learning approaches, the Deep Neural Networks, produce models that are highly non-interpretable. We attempt to address this challenge by proposing a technique called CNN-INTE to interpret deep Convolutional Neural Networks (CNN) via meta-learning. In this work, we interpret a specific hidden layer of the deep CNN model on the MNIST image dataset. We use a clustering algorithm in a two-level structure to find the meta-level training data and Random Forest as base learning algorithms to generate the meta-level test data. The interpretation results are displayed visually via diagrams, which clearly indicates how a specific test instance is classified. Our method achieves global interpretation for all the test instances without sacrificing the accuracy obtained by the original deep CNN model. This means our model is faithful to the deep CNN model, which leads to reliable interpretations.

cs.LG↗

arXiv2018

Homogenization of a Biot-Stokes system modeling deformable vuggy porous media

Vugs are small to medium-sized cavities inside rock, which have significant effects on the fluid flow in rock. Moreover, the presence of vugs may have non-trivial impacts on the geomechanical behavior of rock. How to quantify and analyze such effects is still an opening problem. To this end, we derive a macroscopic poroelastic model for a single-phase viscous fluid flow through a deformable vuggy porous medium. At first, a vuggy porous medium is divided into two parts: the porous matrix and vugs. Then, we model the hydro-mechanical coupling process on the fine scale using Biot's equations within porous matrix, Stokes equations within the vugs, and an extended Beavers-Joseph-Saffman boundary condition on the porous-fluid interface. Next, based on the homogenization theory, we obtain a macroscopic Biot's equations governing the hydro-mechanical coupling behavior of vuggy porous media on larger scale. Subsequently, the macroscopic poroelastic coefficients, such as the effective Darcy permeability, effective Young's modulus and effective Biot coefficient, can be computed from three cell problems. Finally, several numerical examples are designed to demonstrate the computational procedure of evaluating the geomechanical behavior of vuggy porous media.

physics.comp-ph↗

arXiv2017

Hyperbolic components and cubic polynomials

In the space of cubic polynomials, Milnor defined a notable curve $\mathcal S_p$, consisting of cubic polynomials with a periodic critical point, whose period is exactly $p$. In this paper, we show that for any integer $p\geq 1$, any bounded hyperbolic component on $\mathcal{S}_p$ is a Jordan disk.

math.DS↗

arXiv2016

Optimal Conventional Measurements for Quantum-Enhanced Interferometry

A major obstacle to attain the fundamental precision limit of the phase estimation in an interferometry is the identification and implementation of the optimal measurement. Here we demonstrate that this can be accomplished by the use of three conventional measurements among interferometers with Bayesian estimation techniques. Conditions that hold for the precision limit to be attained with these measurements are obtained by explicitly calculating the Fisher information. Remarkably, these conditions are naturally satisfied in most interferometric experiments. We apply our results to an experiment of atomic spectroscopy and examine robustness of phase sensitivity for the two-axis counter-twisted state suffering from detection noise.

quant-ph↗

arXiv2016

Asymptotic properties of functionals of increments of a continuous semimartingale with stochastic sampling times

This paper is concerned with asymptotic behavior of a variety of functionals of increments of continuous semimartingales. Sampling times are assumed to follow a rather general discretization scheme. If an underlying semimartingale is thought of as a financial asset price process, a general sampling scheme like the one employed in this paper is capable of reflecting what happens whenever the financial trading data are recorded in a tick-by-tick fashion. A law of large numbers and a central limit theorem are proved after an appropriate normalization. One application of our result is an extension of the realized kernel estimator of integrated volatility to the case of random sampling times.

math.PR↗

arXiv2016

Quantum Fisher information and symmetric logarithmic derivative via anti-commutators

Symmetric logarithmic derivative (SLD) is a key quantity to obtain quantum Fisher information (QFI) and to construct the corresponding optimal measurements. Here we develop a method to calculate the SLD and QFI via anti-commutators. This method is originated from the Lyapunov representation and would be very useful for cases that the anti-commutators among the state and its partial derivative exhibits periodic properties. As an application, we discuss a class of states, whose squares linearly depend on the states themselves, and give the corresponding analytical expressions of SLD and QFI. A noisy scenario of this class of states is also considered and discussed. Finally, we readily apply the method to the block-diagonal states and the multi-parameter estimation problems.

quant-ph↗

arXiv2016

Moduli space of cubic Newton maps

In this article, we study the topology and bifurcations of the moduli space $\mathcal{M}_3$ of cubic Newton maps. It's a subspace of the moduli space of cubic rational maps, carrying the Riemann orbifold structure $(\mathbb{\widehat{C}}, (2,3,\infty))$. We prove two results: (1). The boundary of the unique unbounded hyperbolic component is a Jordan arc and the boundaries of all other hyperbolic components are Jordan curves. (2).The Head's angle map is surjective and monotone. The fibers of this map are characterized completely. The first result is a moduli space analogue of the first author's dynamical regularity theorem \cite{Ro08}. The second result confirms a conjecture of Tan Lei.

math.DS↗

arXiv2016

Global topology of hyperbolic components I: Cantor circle case

The hyperbolic components in the moduli space ${M}_d$ of degree $d\geq2$ rational maps are mysterious and fundamental topological objects. For those in the connectedness locus, they are known to be the finite quotients of the Euclidean space $\mathbb{R}^{4d-4}$. In this paper, we study the hyperbolic components in the disconnectedness locus and with minimal complexity: those in the Cantor circle locus. We show that each of them is a finite quotient of the space $\mathbb{R}^{4d-4-n}\times\mathbb{T}^{n}$, where $n$ is determined by the dynamics. The proof relates Riemann surface theory (Abel's Theorem), dynamical system and algebraic topology.

math.DS↗

arXiv2016

Quantum multiparameter metrology with generalized entangled coherent state

We propose a generalized form of entangled coherent states (ECS) and apply them in a multi-arm optical interferometer to estimate multiple phase shifts. We obtain the quantum Cramer-Rao bounds for both the linear and nonlinear parameterization protocols. Through the analysis, we find that, utilizing the simultaneous estimation, this generalized form of ECS gives a better precision than the generalized NOON states [Phys. Rev. Lett. \textbf{111}, 070403 (2013)]. Moreover, comparing with the independent estimation, both the linear and nonlinear protocols have the same advantage in the relation to the number of the parameters.

quant-ph↗

arXiv2016

Coherent-State Approach for Majorana representation

By representing a quantum state and its evolution with the majorana stars on the Bloch sphere, the Majorana representation (MR) provide us an intuitive way to study a physical system with SU(2) symmetry. In this work, based on coherent states, we propose a method to establish generalization of MR for a general symmetry. By choosing a generalized coherent state as a reference state, we give a more general MR for both finite and infinite systems and the corresponding star equations are given. Using this method, we study the squeezed vacuum states for three different symmetries, Heisenberg-Weyl, SU(2) and SU(1,1), and express the effect of squeezing parameter on the distribution of stars. Furthermore, we also study the dynamical evolution of stars for an initial coherent state driven by a nonlinear Hamiltonian, and find that at a special time point, the stars are distributed on two orthogonal large circles.

quant-ph↗