Liang He, Su Yi
We investigate thermodynamic properties of lattice Bose gases in optical cavities in the Mott-insulator limit. We find the system assumes anomalous thermodynamic behavior that can be traced back to the breaking of fundamental additivity by its infinite-long range interaction. Specifically, the system shows striking ensemble inequivalence between the canonical ensemble and the grand canonical one, sharply manifesting in the distinct anomalous structure of the thermodynamic phase diagram in the canonical ensemble. In particular, in the temperature regime around half of the on-site energy, the system manifests negative compressibility and anomalous reentrant phase transitions where the ordered charge density wave phase revives from the disordered homogenous phase upon increasing the temperature. Direct experimental observation of the anomalous behavior can be realized in the current experiments with well-controlled total particle number fluctuations.
Liang He, Xianhong Chen, Can Xu, Jia Liu
Most current state-of-the-art text-independent speaker verification systems take probabilistic linear discriminant analysis (PLDA) as their backend classifiers. The parameters of PLDA are often estimated by maximizing the objective function, which focuses on increasing the value of log-likelihood function, but ignoring the distinction between speakers. In order to better distinguish speakers, we propose a multi-objective optimization training for PLDA. Experiment results show that the proposed method has more than 10% relative performance improvement in both EER and MinDCF on the NIST SRE14 i-vector challenge dataset, and about 20% relative performance improvement in EER on the MCE18 dataset.
Liang He, Su Yi
We investigate the finite temperature spin density wave (SDW) and charge density wave (CDW) transition of two-component lattice spinor Bose gases in optical lattices in the Mott-insulator limit. At the temperature scale around half of the on-site interaction energy, we find a new critical regime emerges and features, in particular, a new bicritical line and two critical lines associated with the finite temperature SDW-CDW, homogeneous-SDW, and homogeneous-CDW transition, respectively. Direct calculation of the critical exponents for the scaling behavior and investigating on the effective theory in this critical regime show that they belong to the five-dimensional Ising universality class, clearly manifesting the long-range character of the system's interaction. Our prediction of the emergent criticality can be readily observed by current experimental setups operated at the intermediate temperature scale around half the on-site interaction energy.
Liang He, Terrence Edmonds, Zi-Wei Lin, Feng Liu, Denes Molnar, Fuqiang Wang
We trace the development of elliptic anisotropy ($v_2$) via parton-parton collision history in two transport models. The parton $v_2$ is studied as a function of the number of collisions of each parton in Au+Au and $d$+Au collisions at $\sqrt{s_{_{\rm NN}}}=200$ GeV. It is found that the majority of $v_2$ comes from the anisotropic escape probability of partons, with no fundamental difference at low and high transverse momenta. The contribution to $v_2$ from hydrodynamic-type collective flow is found to be small. Only when the parton-parton cross-section is set unrealistically large does this contribution start to take over. Our findings challenge the current paradigm emerged from hydrodynamic comparisons to anisotropy data.
Liang He, Lukas M. Sieberer, Ehud Altman, Sebastian Diehl
We numerically investigate the scaling properties of a one-dimensional driven-dissipative condensate described by a stochastic complex Ginzburg-Landau equation (SCGLE). We directly extract the static and dynamical scaling exponents from the dynamics of the condensate's phase field, and find that both coincide with the ones of the one-dimensional Kardar-Parisi-Zhang (KPZ) equation. We furthermore calculate the spatial and the temporal two-point correlation functions of the condensate field itself. The decay of the temporal two-point correlator assumes a stretched-exponential form, providing further quantitative evidence for an effective KPZ description. Moreover, we confirm the observability of this non-equilibrium scaling for typical current experimental setups with exciton-polariton systems, if cavities with a reduced $Q$ factor are used.
Liang He, Sebastian Diehl
We investigate the steady state phase diagram of two-component driven open condensates in one dimension. We identify a miscible-immiscible transition which is predominantly driven by gapped density fluctuations and occurs upon increasing the inter-component inelastic coupling. Below the transition in the miscible phase, we find the effective long wavelength dynamics to be described by a two-component Kardar-Parisi-Zhang (KPZ) equation that belongs to the nonequilibrium universality class of the one-dimensional single-component KPZ equation at generic choices of parameters. Our results are relevant for different experimental realizations for two-component driven open condensates in exciton-polariton systems.
Liang He, Hongke Wang, Yongchang Cao, Zhen Wu, Jianbing Zhang, Xinyu Dai
Extracting relational facts from multimodal data is a crucial task in the field of multimedia and knowledge graphs that feeds into widespread real-world applications. The emphasis of recent studies centers on recognizing relational facts in which both entities are present in one modality and supplementary information is used from other modalities. However, such works disregard a substantial amount of multimodal relational facts that arise across different modalities, such as one entity seen in a text and another in an image. In this paper, we propose a new task, namely Multimodal Object-Entity Relation Extraction, which aims to extract "object-entity" relational facts from image and text data. To facilitate research on this task, we introduce MORE, a new dataset comprising 21 relation types and 20,264 multimodal relational facts annotated on 3,559 pairs of textual news titles and corresponding images. To show the challenges of Multimodal Object-Entity Relation Extraction, we evaluated recent state-of-the-art methods for multimodal relation extraction and conducted a comprehensive experimentation analysis on MORE. Our results demonstrate significant challenges for existing methods, underlining the need for further research on this task. Based on our experiments, we identify several promising directions for future research. The MORE dataset and code are available at https://github.com/NJUNLP/MORE.
Liang He, Peiran Jin, Yaosen Min, Shufang Xie, Lijun Wu, Tao Qin, Xiaozhuan Liang, Kaiyuan Gao, Yuliang Jiang, Tie-Yan Liu
Oct 31, 2024·q-bio.QM·PDF Proteins, essential to biological systems, perform functions intricately linked to their three-dimensional structures. Understanding the relationship between protein structures and their amino acid sequences remains a core challenge in protein modeling. While traditional protein foundation models benefit from pre-training on vast unlabeled datasets, they often struggle to capture critical co-evolutionary information, which evolutionary-based methods excel at. In this study, we introduce a novel pre-training strategy for protein foundation models that emphasizes the interactions among amino acid residues to enhance the extraction of both short-range and long-range co-evolutionary features from sequence data. Trained on a large-scale protein sequence dataset, our model demonstrates superior generalization ability, outperforming established baselines of similar size, including the ESM model, across diverse downstream tasks. Experimental results confirm the model's effectiveness in integrating co-evolutionary information, marking a significant step forward in protein sequence-based modeling.
Liang He, Su Yi
We investigate the finite temperature charge density wave (CDW) transition of lattice Bose gases within optical cavities in the deep Mott-insulator limit. We find a new critical regime emerges at a temperature around one-half of the on-site interaction energy, where the first order CDW transition at low temperatures terminates at a critical point and changes to a second order one. By directly calculating the critical exponents and constructing the effective theory in the corresponding critical regime, we find the emergent criticality belongs to the five-dimensional Ising universality class. Direct experimental observation of the emergent criticality can be readily performed by current experimental set-ups operated in the temperature regime around half the on-site interaction energy.
Liang He, Su Yi
We reveal a divergent issue associated with the mean-field theory for Bose gases in optical lattices constructed by the widely used straightforward mean-field decoupling of the hopping term, where the corresponding mean-field Hamiltonian generally assumes no lower energy bound once the spatial dependence of the mean-field superfluid order parameter is taken into account. Via a systematic functional integral approach, we solve this issue by establishing a general finite temperature mean-field theory that can treat any possible spatial dependence of the order parameter without causing the divergent issue. Interestingly, we find the theory generally assumes an intrinsic non-hermitian structure that originates from the indefiniteness of the hopping matrix of the system. Within this theory, we develop an efficient approach for investigating the physics of the system at finite temperature, where properties of the system can be calculated via straightforward investigation on the saddle points of an effective potential function for the order parameter. We illustrate our approach by investigating the finite temperature superfluid transition of Bose gases in optical lattices. Since the underlying finite temperature mean-field theory is quite general, this approach can be straightforwardly applied to investigate the finite temperature properties of related systems with phases possessing complex spatial structures.
Liang He, Walter Hofstetter
We study a system of ultra-cold fermionic polar molecules in a two-dimensional square lattice interacting via both the long-ranged dipole-dipole interaction and a short-ranged on-site attractive interaction. Singlet superfluid, charge density wave, and supersolid phases are found to exist in the system. We map out the zero temperature phase diagram and find that the supersolid phase is considerably stabilized by the dipole-dipole interaction and thus can exist over a large region of filling factors. We study the melting of the supersolid phase with increasing temperature, map out a finite temperature phase diagram of the system at fixed filling, and determine the parameter region where the supersolid phase can possibly be observed in experiments.
Liang He, Jia Pan, Dinesh Manocha
We present a novel approach for collision-free global navigation for continuous-time multi-agent systems with general linear dynamics. Our approach is general and can be used to perform collision-free navigation in 2D and 3D workspaces with narrow passages and crowded regions. As part of pre-computation, we compute multiple bridges in the narrow or tight regions in the workspace using kinodynamic RRT algorithms. Our bridge has certain geometric characteristics, that en- able us to calculate a collision-free trajectory for each agent using simple interpolation at runtime. Moreover, we combine interpolated bridge trajectories with local multi-agent navigation algorithms to compute global collision-free paths for each agent. The overall approach combines the performance benefits of coupled multi-agent algorithms with the pre- computed trajectories of the bridges to handle challenging scenarios. In practice, our approach can handle tens to hundreds of agents in real-time on a single CPU core in 2D and 3D workspaces.
Liang He, Yongqiang Li, Ehud Altman, Walter Hofstetter
We investigate the zero temperature quantum phases of a Bose-Bose mixture on a triangular lattice using Bosonic Dynamical Mean Field Theory (BDMFT). We consider the case of total filling one where geometric frustration arises for asymmetric hopping. We map out a rich ground state phase diagram including xy-ferromagnetic, spin-density wave, superfluid, and supersolid phases. In particular, we identify a stripe spin-density wave phase for highly asymmetric hopping. On top of the spin-density wave, we find that the system generically shows weak charge (particle) density wave order.
Liang He, Terrence Edmonds, Zi-Wei Lin, Feng Liu, Denes Molnar, Fuqiang Wang
We trace the development of elliptic anisotropy (v2) via parton-parton collision history in two transport models. The parton v2 is studied as a function of the number of collisions of each parton in Au+Au and d+Au collisions at sNN =200 GeV. It is found that the majority of v2 comes from the anisotropic escape probability of partons, with no fundamental difference at low and high transverse momenta. The contribution to v2 from hydrodynamic-type collective flow is found to be small. Only when the parton-parton cross-section is set unrealistically large does this contribution start to take over. Our findings challenge the current paradigm emerged from hydrodynamic comparisons to anisotropy data.
Liang He, Yu-xi Liu, S. Yi, C. P. Sun, Franco Nori
Mar 14, 2007·quant-ph·PDF We study the influence of a lossless material medium on the coherent storage and quantum state transfer of a quantized probe light in an ensemble of $Λ$-type atoms. The medium is modeled as uniformly distributed two-level atoms with same energy level spacing, coupling to a probe light. This coupled system can be simplified to a collection of two-mode polaritons which couple to one transition of the $Λ$-type atoms. We show that, when the other transition of $Λ$-type atoms is controlled by a classical light, the electromagnetically induced transparency can also occur for the polaritons. In this case the coherent storage and quantum transfer for photon states are achievable through the novel dark states with respect to the polaritons. By calculating the corresponding dispersion relation, we find the ensemble of the three-level atoms with $Λ$-type transitions may serve as quantum memory for it slows or even stops the light propagation through the mechanism of electromagnetically induced transparency. the corresponding dispersion relation, we find the ensemble of the three-level atoms with $Λ$-type transitions may serve as quantum memory for it slows or even stops the light propagation through the mechanism of electromagnetically induced transparency.
Liang He, Su Yi
We study the ground state properties of a spin-3 Cr condensate subject to an external magnetic field by numerically solving the Gross-Piteavskii equations. We show that the widely adopted single-mode approximation is invalid under a finite magnetic field. In particular, a phase separation like behavior may be induced by the magnetic field. We also point out the possible origin of the phase separation phenomenon.
Liang He, Ruida Li, Mengqi Niu
Currently, most speaker recognition backends, such as cosine, linear discriminant analysis (LDA), or probabilistic linear discriminant analysis (PLDA), make decisions by calculating similarity or distance between enrollment and test embeddings which are already extracted from neural networks. However, for each embedding, the local structure of itself and its neighbor embeddings in the low-dimensional space is different, which may be helpful for the recognition but is often ignored. In order to take advantage of it, we propose a graph neural network (GNN) backend to mine latent relationships among embeddings for classification. We assume all the embeddings as nodes on a graph, and their edges are computed based on some similarity function, such as cosine, LDA+cosine, or LDA+PLDA. We study different graph settings and explore variants of GNN to find a better message passing and aggregation way to accomplish the recognition task. Experimental results on NIST SRE14 i-vector challenging, VoxCeleb1-O, VoxCeleb1-E, and VoxCeleb1-H datasets demonstrate that our proposed GNN backends significantly outperform current mainstream methods.
Liang He, Yougang Chu, Zhen Wu, Jianbing Zhang, Xinyu Dai, Jiajun Chen
Benchmarks are crucial for evaluating machine learning algorithm performance, facilitating comparison and identifying superior solutions. However, biases within datasets can lead models to learn shortcut patterns, resulting in inaccurate assessments and hindering real-world applicability. This paper addresses the issue of entity bias in relation extraction tasks, where models tend to rely on entity mentions rather than context. We propose a debiased relation extraction benchmark DREB that breaks the pseudo-correlation between entity mentions and relation types through entity replacement. DREB utilizes Bias Evaluator and PPL Evaluator to ensure low bias and high naturalness, providing a reliable and accurate assessment of model generalization in entity bias scenarios. To establish a new baseline on DREB, we introduce MixDebias, a debiasing method combining data-level and model training-level techniques. MixDebias effectively improves model performance on DREB while maintaining performance on the original dataset. Extensive experiments demonstrate the effectiveness and robustness of MixDebias compared to existing methods, highlighting its potential for improving the generalization ability of relation extraction models. We will release DREB and MixDebias publicly.
Liang He, Anchun Ji, Walter Hofstetter
We investigate the ground state properties of Bose-Bose mixtures with Rashba-type spin-orbit (SO) coupling in a square lattice. The system displays rich physics from the deep Mott-insulator (MI) all the way to the superfluid (SF) regime. In the deep MI regime, novel spin-ordered phases arise due to the effective Dzyaloshinskii-Moriya type super-exchange interactions. By employing the non-perturbative Bosonic Dynamical Mean-Field-Theory (BDMFT), we numerically study and establish the stability of these magnetic phases against increasing hopping amplitude. We show that as hopping is increased across the MI to SF transition, exotic superfluid phases with magnetic textures emerge. In particular, we identify a new spin-spiral magnetic texture with spatial period 3 in the superfluid close to the MI-SF transition.
Liang He, Xianhong Chen, Can Xu, Yi Liu, Jia Liu, Michael T Johnson
In this paper, we apply a latent class model (LCM) to the task of speaker diarization. LCM is similar to Patrick Kenny's variational Bayes (VB) method in that it uses soft information and avoids premature hard decisions in its iterations. In contrast to the VB method, which is based on a generative model, LCM provides a framework allowing both generative and discriminative models. The discriminative property is realized through the use of i-vector (Ivec), probabilistic linear discriminative analysis (PLDA), and a support vector machine (SVM) in this work. Systems denoted as LCM-Ivec-PLDA, LCM-Ivec-SVM, and LCM-Ivec-Hybrid are introduced. In addition, three further improvements are applied to enhance its performance. 1) Adding neighbor windows to extract more speaker information for each short segment. 2) Using a hidden Markov model to avoid frequent speaker change points. 3) Using an agglomerative hierarchical cluster to do initialization and present hard and soft priors, in order to overcome the problem of initial sensitivity. Experiments on the National Institute of Standards and Technology Rich Transcription 2009 speaker diarization database, under the condition of a single distant microphone, show that the diarization error rate (DER) of the proposed methods has substantial relative improvements compared with mainstream systems. Compared to the VB method, the relative improvements of LCM-Ivec-PLDA, LCM-Ivec-SVM, and LCM-Ivec-Hybrid systems are 23.5%, 27.1%, and 43.0%, respectively. Experiments on our collected database, CALLHOME97, CALLHOME00 and SRE08 short2-summed trial conditions also show that the proposed LCM-Ivec-Hybrid system has the best overall performance.