arXiv subjects

Yanxia Liu

Publications and source records attributed to Yanxia Liu.

At least 19 recordsLinked to original sources

arXiv2025

Hierarchy of localized many-body bound states in an interacting open lattice

We unveil the mechanism for the formation of puzzled boundary-localized bound states in a spinless fermionic open lattice with nearest-neighbor interactions. By solving the Bethe-ansatz equation analytically, we uncover asymmetrical string solutions corresponding to the boundary-localized bound states, which emerge in systems with at least three particles. The localized bound states can become bound states in continuum in a suitable parameter region. When the number of particles increases to five or more, additional bound states away from the edge are also observed. Through rigorous analysis, we derive recurrence relations of the quasi-momentum of the localized states as a function of the number of particles, predicting the presence of hierarchy of localized many-body bound states in interacting open lattices.

cond-mat.quant-gas↗

arXiv2024

CoNav: A Benchmark for Human-Centered Collaborative Navigation

Human-robot collaboration, in which the robot intelligently assists the human with the upcoming task, is an appealing objective. To achieve this goal, the agent needs to be equipped with a fundamental collaborative navigation ability, where the agent should reason human intention by observing human activities and then navigate to the human's intended destination in advance of the human. However, this vital ability has not been well studied in previous literature. To fill this gap, we propose a collaborative navigation (CoNav) benchmark. Our CoNav tackles the critical challenge of constructing a 3D navigation environment with realistic and diverse human activities. To achieve this, we design a novel LLM-based humanoid animation generation framework, which is conditioned on both text descriptions and environmental context. The generated humanoid trajectory obeys the environmental context and can be easily integrated into popular simulators. We empirically find that the existing navigation methods struggle in CoNav task since they neglect the perception of human intention. To solve this problem, we propose an intention-aware agent for reasoning both long-term and short-term human intention. The agent predicts navigation action based on the predicted intention and panoramic observation. The emergent agent behavior including observing humans, avoiding human collision, and navigation reveals the efficiency of the proposed datasets and agents.

cs.CV↗

arXiv2024

Fate of Two-Particle Bound States in the Continuum in non-Hermitian Systems

We unveil the existence of two-particle bound state in the continuum (BIC) in a one-dimensional interacting nonreciprocal lattice with a generalized boundary condition. By applying the Bethe-ansatz method, we can exactly solve the wavefunction and eigenvalue of the bound state in the continuum band, which enable us to precisely determine the phase diagrams of BIC. Our results demonstrate that the non-reciprocal hopping can delocalize the bound state and thus shrink the region of BIC. By analyzing the wavefunction, we identify the existence of two types of BICs with different spatial distributions and analytically derive the corresponding threshold values for the breakdown of BICs. The BIC with similar properties is also found to exist in another system with an impurity potential.

cond-mat.quant-gas↗

arXiv2024

Two mass-imbalanced atoms in a hard-wall trap: Deep learning integrability of many-body systems

The study of integrable systems has led to significant advancements in our understanding of many-body physics. We design a series of numerical experiments to analyze the integrability of a mass-imbalanced two-body system through energy level statistics and deep learning of wavefunctions. The level spacing distributions are fitted by a Brody distribution and the fitting parameter $\omega$ is found to separate the integrable and non-integrable mass ratios by a critical line $\omega=0$. The convolutional neural network built from the probability density images could identify the transition points between integrable and non-integrable systems with high accuracy, yet in a much shorter computation time. A brilliant example of the network's ability is to identify a new integrable mass ratio $1/3$ by learning from the known integrable case of equal mass, with a remarkable network confidence of $98.78\%$. The robustness of our neural networks is further enhanced by adversarial learning, where samples are generated by standard and quantum perturbations mixed in the probability density images and the wavefunctions, respectively.

cond-mat.quant-gas↗

arXiv2023

Semantic-Constraint Matching Transformer for Weakly Supervised Object Localization

Weakly supervised object localization (WSOL) strives to learn to localize objects with only image-level supervision. Due to the local receptive fields generated by convolution operations, previous CNN-based methods suffer from partial activation issues, concentrating on the object's discriminative part instead of the entire entity scope. Benefiting from the capability of the self-attention mechanism to acquire long-range feature dependencies, Vision Transformer has been recently applied to alleviate the local activation drawbacks. However, since the transformer lacks the inductive localization bias that are inherent in CNNs, it may cause a divergent activation problem resulting in an uncertain distinction between foreground and background. In this work, we proposed a novel Semantic-Constraint Matching Network (SCMN) via a transformer to converge on the divergent activation. Specifically, we first propose a local patch shuffle strategy to construct the image pairs, disrupting local patches while guaranteeing global consistency. The paired images that contain the common object in spatial are then fed into the Siamese network encoder. We further design a semantic-constraint matching module, which aims to mine the co-object part by matching the coarse class activation maps (CAMs) extracted from the pair images, thus implicitly guiding and calibrating the transformer network to alleviate the divergent activation. Extensive experimental results conducted on two challenging benchmarks, including CUB-200-2011 and ILSVRC datasets show that our method can achieve the new state-of-the-art performance and outperform the previous method by a large margin.

cs.CV↗

arXiv2023

Imbalance-Agnostic Source-Free Domain Adaptation via Avatar Prototype Alignment

Source-free Unsupervised Domain Adaptation (SF-UDA) aims to adapt a well-trained source model to an unlabeled target domain without access to the source data. One key challenge is the lack of source data during domain adaptation. To handle this, we propose to mine the hidden knowledge of the source model and exploit it to generate source avatar prototypes. To this end, we propose a Contrastive Prototype Generation and Adaptation (CPGA) method. CPGA consists of two stages: Prototype generation and Prototype adaptation. Extensive experiments on three UDA benchmark datasets demonstrate the superiority of CPGA. However, existing SF.UDA studies implicitly assume balanced class distributions for both the source and target domains, which hinders their real applications. To address this issue, we study a more practical SF-UDA task, termed imbalance-agnostic SF-UDA, where the class distributions of both the unseen source domain and unlabeled target domain are unknown and could be arbitrarily skewed. This task is much more challenging than vanilla SF-UDA due to the co-occurrence of covariate shifts and unidentified class distribution shifts between the source and target domains. To address this task, we extend CPGA and propose a new Target-aware Contrastive Prototype Generation and Adaptation (T-CPGA) method. Specifically, for better prototype adaptation in the imbalance-agnostic scenario, T-CPGA applies a new pseudo label generation strategy to identify unknown target class distribution and generate accurate pseudo labels, by utilizing the collective intelligence of the source model and an additional contrastive language-image pre-trained model. Meanwhile, we further devise a target label-distribution-aware classifier to adapt the model to the unknown target class distribution. We empirically show that T-CPGA significantly outperforms CPGA and other SF-UDA methods in imbalance-agnostic SF-UDA.

cs.CV↗

arXiv2023

General mapping of one-dimensional non-Hermitian mosaic models to non-mosaic counterparts: Mobility edges and Lyapunov exponents

We establish a general mapping from one-dimensional non-Hermitian mosaic models to their non-mosaic counterparts. This mapping can give rise to mobility edges and even Lyapunov exponents in the mosaic models if critical points of localization or Lyapunov exponents of localized states in the corresponding non-mosaic models have already been analytically solved. To demonstrate the validity of this mapping, we apply it to two non-Hermitian localization models: an Aubry-Andr\'e-like model with nonreciprocal hopping and complex quasiperiodic potentials, and the Ganeshan-Pixley-Das Sarma model with nonreciprocal hopping. We successfully obtain the mobility edges and Lyapunov exponents in their mosaic models. This general mapping may catalyze further studies on mobility edges, Lyapunov exponents, and other significant quantities pertaining to localization in non-Hermitian mosaic models.

cond-mat.dis-nn↗

arXiv2022

A general approach to the exact localized transition points of 1D mosaic disorder models

In this paper, we present a general correspondence between the mosaic and non-mosaic models, which can be used to obtain the exact solution for the mosaic ones. This relation holds not only for the quasicrystal models, but also for the Anderson models. Despite the different localization properties of the specific models, this relationship shares a unified form. Applying our method to the mosaic Anderson models, we find that there is a discrete set of extended states. At last, we also give the general analytical mobility edge for the mosaic slowly varying potential models and the mosaic Ganeshan-Pixley-Das Sarma models.

cond-mat.dis-nn↗

arXiv2022

Prototype-Guided Continual Adaptation for Class-Incremental Unsupervised Domain Adaptation

This paper studies a new, practical but challenging problem, called Class-Incremental Unsupervised Domain Adaptation (CI-UDA), where the labeled source domain contains all classes, but the classes in the unlabeled target domain increase sequentially. This problem is challenging due to two difficulties. First, source and target label sets are inconsistent at each time step, which makes it difficult to conduct accurate domain alignment. Second, previous target classes are unavailable in the current step, resulting in the forgetting of previous knowledge. To address this problem, we propose a novel Prototype-guided Continual Adaptation (ProCA) method, consisting of two solution strategies. 1) Label prototype identification: we identify target label prototypes by detecting shared classes with cumulative prediction probabilities of target samples. 2) Prototype-based alignment and replay: based on the identified label prototypes, we align both domains and enforce the model to retain previous knowledge. With these two strategies, ProCA is able to adapt the source model to a class-incremental unlabeled target domain effectively. Extensive experiments demonstrate the effectiveness and superiority of ProCA in resolving CI-UDA. The source code is available at https://github.com/Hongbin98/ProCA.git

cs.CV↗

arXiv2022

Dynamic quantum-enhanced sensing without entanglement in central spin systems

We propose a dynamic quantum sensing scheme by using a quantum many-spin system composed of a central spin interacting with many surrounding spins. Starting from a generalized Ising ring model, we investigate the error propagation formula of the central spin and it indicates that Heisenberg scaling can be reached while the probe state only needs to be a product state. Particularly, we derive an analytical form of the dynamic quantum Fisher information in a limit case, which explicitly exhibits the Heisenberg scaling. By comparing with numerical results, we demonstrate that the general case can be well approximated by the analytical result when the coupling strength among the surrounding spins is much weaker than the coupling strength between the central and surrounding spins. This analytic result guides us to find the appropriate probe state and the proper measurement time, to achieve the Heisenberg scaling in realistic situations. Furthermore, we investigate various effects which are important in practical quantum systems, including the central spin Zeeman term, the anisotropy of the hyperfine interaction and the inhomogeneity of the hyperfine coupling strength. Our result indicates that the dynamic quantum-enhanced sensing scheme seems feasible in realistic quantum central spin systems, like semiconductor quantum dots.

quant-ph↗

arXiv2021

Improved Knowledge Distillation via Adversarial Collaboration

Knowledge distillation has become an important approach to obtain a compact yet effective model. To achieve this goal, a small student model is trained to exploit the knowledge of a large well-trained teacher model. However, due to the capacity gap between the teacher and the student, the student's performance is hard to reach the level of the teacher. Regarding this issue, existing methods propose to reduce the difficulty of the teacher's knowledge via a proxy way. We argue that these proxy-based methods overlook the knowledge loss of the teacher, which may cause the student to encounter capacity bottlenecks. In this paper, we alleviate the capacity gap problem from a new perspective with the purpose of averting knowledge loss. Instead of sacrificing part of the teacher's knowledge, we propose to build a more powerful student via adversarial collaborative learning. To this end, we further propose an Adversarial Collaborative Knowledge Distillation (ACKD) method that effectively improves the performance of knowledge distillation. Specifically, we construct the student model with multiple auxiliary learners. Meanwhile, we devise an adversarial collaborative module (ACM) that introduces attention mechanism and adversarial learning to enhance the capacity of the student. Extensive experiments on four classification tasks show the superiority of the proposed ACKD.

cs.CV↗

arXiv2021

Semi-Online Knowledge Distillation

Knowledge distillation is an effective and stable method for model compression via knowledge transfer. Conventional knowledge distillation (KD) is to transfer knowledge from a large and well pre-trained teacher network to a small student network, which is a one-way process. Recently, deep mutual learning (DML) has been proposed to help student networks learn collaboratively and simultaneously. However, to the best of our knowledge, KD and DML have never been jointly explored in a unified framework to solve the knowledge distillation problem. In this paper, we investigate that the teacher model supports more trustworthy supervision signals in KD, while the student captures more similar behaviors from the teacher in DML. Based on these observations, we first propose to combine KD with DML in a unified framework. Furthermore, we propose a Semi-Online Knowledge Distillation (SOKD) method that effectively improves the performance of the student and the teacher. In this method, we introduce the peer-teaching training fashion in DML in order to alleviate the student's imitation difficulty, and also leverage the supervision signals provided by the well-trained teacher in KD. Besides, we also show our framework can be easily extended to feature-based distillation methods. Extensive experiments on CIFAR-100 and ImageNet datasets demonstrate the proposed method achieves state-of-the-art performance.

cs.CV↗

arXiv2021

Source-free Domain Adaptation via Avatar Prototype Generation and Adaptation

We study a practical domain adaptation task, called source-free unsupervised domain adaptation (UDA) problem, in which we cannot access source domain data due to data privacy issues but only a pre-trained source model and unlabeled target data are available. This task, however, is very difficult due to one key challenge: the lack of source data and target domain labels makes model adaptation very challenging. To address this, we propose to mine the hidden knowledge in the source model and exploit it to generate source avatar prototypes (i.e., representative features for each source class) as well as target pseudo labels for domain alignment. To this end, we propose a Contrastive Prototype Generation and Adaptation (CPGA) method. Specifically, CPGA consists of two stages: (1) prototype generation: by exploring the classification boundary information of the source model, we train a prototype generator to generate avatar prototypes via contrastive learning. (2) prototype adaptation: based on the generated source prototypes and target pseudo labels, we develop a new robust contrastive prototype adaptation strategy to align each pseudo-labeled target data to the corresponding source prototypes. Extensive experiments on three UDA benchmark datasets demonstrate the effectiveness and superiority of the proposed method.

cs.CV↗

arXiv2021

Exact solution of single impurity problem in non-reciprocal lattices: impurity induced size-dependent non-Hermitian skin effect

Non-Hermitian non-reciprocal systems are known to be extremely sensitive to boundary conditions, exhibiting diverse localizing behaviors and spectrum structures when translational invariance is locally broken, either by tuning the boundary coupling strength, or by introducing an effective boundary using impurities or defects. In this work, we consider the single impurity problem in the Hatano-Nelson model and the Su-Schreieffer-Heeger model, which can be exactly solved with the single impurity being treated as an effective boundary of the system. From our exact solutions for finite-size systems, we unveil that increasing the impurity strength can lead to a transition of the bulk states from non-skin states to skin states, accompanied by the change of the spectrum structure from an ellipse in the complex plane to a segment along the real axis. These exact results indicate that the critical value of impurity strength is size-dependent, and increases exponentially with the lattice size when the impurity is strong or the system is large enough. OUr exact solutions are also useful for determining the point-gap topological transition in the concerned models.

quant-ph↗

arXiv2021

Three-body bound states of two bosons and one impurity in one dimension

We investigate one-dimensional three-body systems composed of two identical bosons and one imbalanced atom (impurity) with two-body and three-body zero-range interactions. For the case in the absence of three-body interaction, we give a complete phase diagram of the number of three-body bound states in the whole region of mass ratio via the direct calculation of the Skornyakov-Ter-Martirosyan equations. We demonstrate that other low-lying three-body bound states emerge when the mass of the impurity particle is not equal to another two identical particles. We can obtain not only the binding energies but also the corresponding wave functions. When the mass of impurity atom is vary large, there are at most three three-body bound states. We then study the effect of three-body zero-range interaction and unveil that it can induces one more three-body bound state at a certain region of coupling strength ratio under a fixed mass ratio.

cond-mat.quant-gas↗

arXiv2021

Feynman diagram approach to dynamical Casimir effect in optimechanical cavity

In this paper we study an optomechnical system enclosed by an optical cavity with one mirror attached to a spring as a closed quantum system. We provide a different angle of studying the phenomenons related to the dynamical Casimir effect via Feynman diagram technique. Dressing effects of phonon, photon and coupling strength are discussed. The energy shift of ground state is obtained. The dynamical Casimir effect is modeled by the scattering processes converting phonons to photons and the corresponding scattering amplitudes are computed. The force-force correlation function of the radiation pressure is derived, whose non-Gaussian probability distribution is revealed.

quant-ph↗

arXiv2021

Exact solution of non-Hermitian systems with generalized boundary conditions: size-dependent boundary effect and fragility of skin effect

Systems with non-Hermitian skin effects are very sensitive to the imposed boundary conditions and lattice size, and thus an important question is whether non-Hermitian skin effects can survive when deviating from the open boundary condition. To unveil the origin of boundary sensitivity, we present exact solutions for one-dimensional non-Hermitian models with generalized boundary conditions and study rigorously the interplay effect of lattice size and boundary terms. Besides the open boundary condition, we identify the existence of non-Hermitian skin effect when one of the boundary hopping terms vanishes. Apart from this critical line on the boundary parameter space, we find that the skin effect is fragile under any tiny boundary perturbation in the thermodynamic limit, although it can survive in a finite size system. Moreover, we demonstrate that the non-Hermitian Su-Schreieffer-Heeger model exhibits a new phase diagram in the boundary critical line, which is different from either open or periodical boundary case.

quant-ph↗

arXiv2020

Exact non-Hermitian mobility edges in one-dimensional quasicrystal lattice with exponentially decaying hopping and its dual lattice

We analytically determine the non-Hermitian mobility edges of a one-dimensional quasiperiodic lattice model with exponential decaying hopping and complex potentials as well as its dual model, which is just a non-Hermitian generalization of the Ganeshan-Pixley-Das Sarma model with nonreciprocal nearest-neighboring hopping. The presence of non-Hermitian term destroys the self-duality symmetry and thus prevents us exploring the localization-delocalization point through looking for self-dual points. Nevertheless, by applying Avila's global theory, the Lyapunov exponent of the Ganeshan-Pixley-Das Sarma model can be exactly derived, which enables us to get an analytical expression of mobility edge of the non-Hermitian dual model. Consequently, the mobility edge of the original model is obtained by using the dual transformation, which creates exact mappings between the spectra and wavefunctions of these two models.

cond-mat.dis-nn↗