arXiv subjects

Mengqi Zhang

Publications and source records attributed to Mengqi Zhang.

69 records · Page 4Linked to original sources

arXiv2022

Deep Contrastive Multiview Network Embedding

Multiview network embedding aims at projecting nodes in the network to low-dimensional vectors, while preserving their multiple relations and attribute information. Contrastive learning approaches have shown promising performance in this task. However, they neglect the semantic consistency between fused and view representations and have difficulty in modeling complementary information between different views. To deal with these deficiencies, this work presents a novel Contrastive leaRning framEwork for Multiview network Embedding (CREME). In our work, different views can be obtained based on the various relations among nodes. Then, we generate view embeddings via proper view encoders and utilize an attentive multiview aggregator to fuse these representations. Particularly, we design two collaborative contrastive objectives, view fusion InfoMax and inter-view InfoMin, to train the model in a self-supervised manner. The former objective distills information from embeddings generated from different views, while the latter captures complementary information among views to promote distinctive view embeddings. We also show that the two objectives can be unified into one objective for model training. Extensive experiments on three real-world datasets demonstrate that our proposed CREME is able to consistently outperform state-of-the-art methods.

cs.LG↗

arXiv2022

On the large-Weissenberg-number scaling laws in viscoelastic pipe flows

This work explains a scaling law of the first Landau coefficient of the derived Ginzburg-Landau equation (GLE) in the weakly nonlinear analysis of axisymmetric viscoelastic pipe flows in the large-Weissenberg-number ($Wi$) limit, recently reported in Wan et al. J. Fluid Mech. (2021), vol. 929, A16. Using an asymptotic method, we derive a reduced system, which captures the characteristics of the linear centre-mode instability near the critical condition in the large-$Wi$ limit. Based on the reduced system we then conduct a weakly nonlinear analysis using a multiple-scale expansion method, which readily explains the aforementioned scaling law of the Landau coefficient and some other scaling laws. Particularly, the equilibrium amplitude of disturbance near linear critical conditions is found to scale as $Wi^{-1/2}$, which may be of interest to experimentalists. The current analysis reduces the numbers of parameters and unknowns and exemplifies an approach to studying the viscoelastic flow at large $Wi$, which could shed new light on the understanding of its nonlinear dynamics.

physics.flu-dyn↗

arXiv2022

A three-dimensional dynamic mode decomposition analysis of wind farm flow aerodynamics

High-fidelity large-eddy simulations are suitable to obtain insight into the complex flow dynamics in extended wind farms. In order to better understand these flow dynamics, we use dynamic mode decomposition (DMD) to analyze and reconstruct the flow field in large-scale numerically simulated wind farms by large-eddy simulations (LES). Different wind farm layouts are considered, and we find that a combination of horizontal and vertical staggering leads to improved wind farm performance compared to traditional horizontal staggering. We analyze the wind farm flows using the amplitude selection (AP) and sparsity-promoting (SP method) DMD approach. We find that the AP method tends to select modes with a small length scale and a high frequency, while the SP method selects large coherent structures with low frequency. The latter are somewhat reminiscent of modes obtained using proper orthogonal decomposition (POD). We find that a relatively limited number of SP-DMD modes is sufficient to accurately reconstruct the flow field in the entire wind farm, whereas the AP-DMD method requires more modes to achieve an accurate reconstruction. Thus, the SP-DMD method has a smaller performance loss compared to the AP-DMD method in terms of the reconstruction of the flow field.

physics.flu-dyn↗

arXiv2022

Latent Structure Mining with Contrastive Modality Fusion for Multimedia Recommendation

Recent years have witnessed growing interests in multimedia recommendation, which aims to predict whether a user will interact with an item with multimodal contents. Previous studies focus on modeling user-item interactions with multimodal features included as side information. However, this scheme is not well-designed for multimedia recommendation. Firstly, only collaborative item-item relationships are implicitly modeled through high-order item-user-item co-occurrences. We argue that the latent semantic item-item structures underlying these multimodal contents could be beneficial for learning better item representations and assist the recommender models to comprehensively discover candidate items. Secondly, previous studies disregard the fine-grained multimodal fusion. Although having access to multiple modalities might allow us to capture rich information, we argue that the simple coarse-grained fusion by linear combination or concatenation in previous work is insufficient to fully understand content information and item relationships.To this end, we propose a latent structure MIning with ContRastive mOdality fusion method (MICRO for brevity). To be specific, we devise a novel modality-aware structure learning module, which learns item-item relationships for each modality. Based on the learned modality-aware latent item relationships, we perform graph convolutions that explicitly inject item affinities to modality-aware item representations. Then, we design a novel contrastive method to fuse multimodal features. These enriched item representations can be plugged into existing collaborative filtering methods to make more accurate recommendations. Extensive experiments on real-world datasets demonstrate the superiority of our method over state-of-the-art baselines.

cs.IR↗

arXiv2022

Nonlinear spatiotemporal instabilities in two-dimensional electroconvective flows

This work studies the effects of a through-flow on two-dimensional electrohydrodynamic (EHD) flows of a dielectric liquid confined between two plane plates, as a model problem to further our understanding of the fluid mechanics in the presence of an electric field. The liquid is subjected to a strong unipolar charge injection from the bottom plate and a pressure gradient along the streamwise direction. Highly-accurate numerical simulations and weakly nonlinear stability analyses based on multiple-scale expansion and amplitude expansion methods are used to unravel the nonlinear spatiotemporal instability mechanisms in this combined flow. We found that the through-flow makes the hysteresis loop in the EHD flow narrower. In the numerical simulation of an impulse response, the leading and trailing edges of the wavepacket within the nonlinear regime are consistent with the linear ones, a result which we also verified against that in natural convection. In addition, as the bifurcation in EHD-Poiseuille flows is of a subcritical nature, nonlinear finite-amplitude solutions exist in the subcritical regime, and our calculation indicates that they are convectively unstable. The validity of the Ginzburg-Landau equation (GLE), derived from the weakly nonlinear expansion of Navier-Stokes equations and the Maxwell's equations in the quasi-electrostatic limit, serving as a physical reduced-order model for probing the spatiotemporal dynamics in this flow, has also been investigated. We found that the coefficients in the GLE calculated using amplitude expansion method can predict the absolute growth rates even when the parameters are away from the linear critical conditions, compared favourably with the local dispersion relation, whereas the validity range of the GLE derived from the multiple-scale expansion method is confined to the vicinity of the linear critical conditions.

physics.flu-dyn↗

arXiv2021

Reinforcement-learning-based control of confined cylinder wakes with stability analyses

This work studies the application of a reinforcement-learning-based (RL) flow control strategy to the flow past a cylinder confined between two walls in order to suppress vortex shedding. The control action is blowing and suction of two synthetic jets on the cylinder. The theme of this study is to investigate how to use and embed physical information of the flow in the RL-based control. First, global linear stability and sensitivity analyses based on the time-mean flow and the steady flow (which is a solution to the Navier-Stokes equations) are conducted in a range of blockage ratios and Reynolds numbers. It is found that the most sensitive region in the wake extends itself when either parameter increases in the parameter range we investigated here. Then, we utilise these physical results to help design RL-based control policies. We find that the controlled wake converges to the unstable steady base flow, where the vortex shedding can be successfully suppressed. A persistent oscillating control seems necessary to maintain this unstable state. The RL algorithm is able to outperform a gradient-based optimisation method (optimised in a certain period of time) in the long run. Furthermore, when the flow stability information is embedded in the reward function to penalise the instability, the controlled flow may become more stable. Finally, according to the sensitivity analyses, the control is most efficient when the probes are placed in the most sensitive region. The control can be successful even when few probes are properly placed in this manner.

physics.flu-dyn↗

arXiv2021

Onset of vortex shedding around a short cylinder

This paper presents results of three-dimensional direct numerical simulations (DNS) and global linear stability analyses (LSA) of a viscous incompressible flow past a finite-length cylinder with two free flat ends. The cylindrical axis is normal to the streamwise direction. The work focuses on the effects of aspect ratios (in the range of $0.5\leq \rm{\small AR} \leq2$, cylinder length over diameter) and Reynolds numbers ($Re\leq1000$ based on cylinder diameter and uniform incoming velocity) on the onset of vortex shedding in this flow. All important flow patterns have been identified and studied, especially as $\rm{\small AR}$ changes. The appearance of a steady wake pattern when $\rm{\small AR}\leq1.75$ has not been discussed earlier in the literature for this flow. LSA based on the time-mean flow has been applied to understand the Hopf bifurcation past which vortex shedding happens. The nonlinear DNS results indicate that there are two vortex shedding patterns at different $Re$, one is transient and the other is nonlinearly saturated. The vortex-shedding frequencies of these two flow patterns correspond to the eigenfrequencies of the two global modes in the stability analysis of the time-mean flow. Wherever possible, we compare the results of our analyses to those of the flows past other short-$\rm{\small AR}$ bluff bodies in order that our discussions bear more general meanings.

physics.flu-dyn↗

arXiv2021

Subcritical and supercritical bifurcations in axisymmetric viscoelastic pipe flows

Axisymmetric viscoelastic pipe flow of Oldroyd-B fluids has been recently found to be linearly unstable by Garg et al. Phys. Rev. Lett., 121.024502 (2018). From a nonlinear point of view, this means that the flow can transition to turbulence supercritically, in contrast to the subcritical Newtonian pipe flows. Experimental evidences of subcritical and supercritical bifurcations of viscoelastic pipe flows have been reported, but these nonlinear phenomena have not been examined theoretically. In this work, we study the weakly nonlinear stability of this flow by performing a multiple-scale expansion of the disturbance around linear critical conditions. The perturbed parameter is Reynolds number with the others being unperturbed. A third-order Ginzburg-Landau equation is derived with its coefficient indicating the bifurcation type of the flow. After exploring a large parameter space, we found that polymer concentration plays an important role: at high polymer concentrations (or small solvent-to-solution viscosity ratio $β\lessapprox0.785$), the nonlinearity stabilises the flow, indicating that the flow will bifurcate supercritically, while at low polymer concentrations ($β\gtrapprox 0.785$), the flow bifurcation is subcritical. The results agree qualitatively with experimental observations where critical $β\approx0.855$. The pipe flow of UCM fluids can be linearly unstable and its bifurcation type is also supercritical. At a fixed value of $β$, the Landau coefficient scales with the inverse of Weissenberg number ($Wi$) when $Wi$ is sufficiently large. The present analysis provides a theoretical understanding of the recent studies on the supercritical and subcritical routes to the elasto-inertial turbulence in viscoelastic pipe flows.

physics.flu-dyn↗

arXiv2021

Dynamic Graph Neural Networks for Sequential Recommendation

Modeling user preference from his historical sequences is one of the core problems of sequential recommendation. Existing methods in this field are widely distributed from conventional methods to deep learning methods. However, most of them only model users' interests within their own sequences and ignore the dynamic collaborative signals among different user sequences, making it insufficient to explore users' preferences. We take inspiration from dynamic graph neural networks to cope with this challenge, modeling the user sequence and dynamic collaborative signals into one framework. We propose a new method named Dynamic Graph Neural Network for Sequential Recommendation (DGSR), which connects different user sequences through a dynamic graph structure, exploring the interactive behavior of users and items with time and order information. Furthermore, we design a Dynamic Graph Recommendation Network to extract user's preferences from the dynamic graph. Consequently, the next-item prediction task in sequential recommendation is converted into a link prediction between the user node and the item node in a dynamic graph. Extensive experiments on three public benchmarks show that DGSR outperforms several state-of-the-art methods. Further studies demonstrate the rationality and effectiveness of modeling user sequences through a dynamic graph.

cs.IR↗

arXiv2021

Personalized Graph Neural Networks with Attention Mechanism for Session-Aware Recommendation

The problem of session-aware recommendation aims to predict users' next click based on their current session and historical sessions. Existing session-aware recommendation methods have defects in capturing complex item transition relationships. Other than that, most of them fail to explicitly distinguish the effects of different historical sessions on the current session. To this end, we propose a novel method, named Personalized Graph Neural Networks with Attention Mechanism (A-PGNN) for brevity. A-PGNN mainly consists of two components: one is Personalized Graph Neural Network (PGNN), which is used to extract the personalized structural information in each user behavior graph, compared with the traditional Graph Neural Network (GNN) model, which considers the role of the user when the node embeddding is updated. The other is Dot-Product Attention mechanism, which draws on the Transformer net to explicitly model the effect of historical sessions on the current session. Extensive experiments conducted on two real-world data sets show that A-PGNN evidently outperforms the state-of-the-art personalized session-aware recommendation methods.

cs.IR↗

arXiv2021

Dynamic Graph Collaborative Filtering

Dynamic recommendation is essential for modern recommender systems to provide real-time predictions based on sequential data. In real-world scenarios, the popularity of items and interests of users change over time. Based on this assumption, many previous works focus on interaction sequences and learn evolutionary embeddings of users and items. However, we argue that sequence-based models are not able to capture collaborative information among users and items directly. Here we propose Dynamic Graph Collaborative Filtering (DGCF), a novel framework leveraging dynamic graphs to capture collaborative and sequential relations of both items and users at the same time. We propose three update mechanisms: zero-order 'inheritance', first-order 'propagation', and second-order 'aggregation', to represent the impact on a user or item when a new interaction occurs. Based on them, we update related user and item embeddings simultaneously when interactions occur in turn, and then use the latest embeddings to make recommendations. Extensive experiments conducted on three public datasets show that DGCF significantly outperforms the state-of-the-art dynamic recommendation methods up to 30. Our approach achieves higher performance when the dataset contains less action repetition, indicating the effectiveness of integrating dynamic collaborative information.

cs.IR↗

arXiv2020

Price of Fairness in Budget Division for Egalitarian Social Welfare

We study a participatory budgeting problem of aggregating the preferences of agents and dividing a budget over the projects. A budget division solution is a probability distribution over the projects. The main purpose of our study concerns the comparison between the system optimum solution and a fair solution. We are interested in assessing the quality of fair solutions, i.e., in measuring the system efficiency loss under a fair allocation compared to the one that maximizes (egalitarian) social welfare. This indicator is called the price of fairness. We are also interested in the performance of several aggregation rules. Asymptotically tight bounds are provided both for the price of fairness and the efficiency guarantee of aggregation rules.

cs.GT↗

arXiv2020

Mechanism Design for Facility Location Games with Candidate Locations

We study the facility location games with candidate locations from a mechanism design perspective. Suppose there are n agents located in a metric space whose locations are their private information, and a group of candidate locations for building facilities. The authority plans to build some homogeneous facilities among these candidates to serve the agents, who bears a cost equal to the distance to the closest facility. The goal is to design mechanisms for minimizing the total/maximum cost among the agents. For the single-facility problem under the maximum-cost objective, we give a deterministic 3-approximation group strategy-proof mechanism, and prove that no deterministic (or randomized) strategy-proof mechanism can have an approximation ratio better than 3 (or 2). For the two-facility problem on a line, we give an anonymous deterministic group strategy-proof mechanism that is (2n-3)-approximation for the total-cost objective, and 3-approximation for the maximum-cost objective. We also provide (asymptotically) tight lower bounds on the approximation ratio.

cs.GT↗

arXiv2019

Critical Edge Identification: A K-Truss Based Model

In a social network, the strength of relationships between users can significantly affect the stability of the network. In this paper, we use the k-truss model to measure the stability of a social network. To identify critical connections, we propose a novel problem, named k-truss minimization. Given a social network G and a budget b, it aims to find b edges for deletion which can lead to the maximum number of edge breaks in the k-truss of G. We show that the problem is NP-hard. To accelerate the computation, novel pruning rules are developed to reduce the candidate size. In addition, we propose an upper bound based strategy to further reduce the searching space. Comprehensive experiments are conducted over real social networks to demonstrate the efficiency and effectiveness of the proposed techniques.

cs.SI↗

arXiv2015

Modal and nonmodal stability analysis of electrohydrodynamic flow with and without cross-flow

We report the results of a complete modal and nonmodal linear stability analysis of the electrohydrodynamic flow (EHD) for the problem of electroconvection in the strong injection region. Convective cells are formed by Coulomb force in an insulating liquid residing between two plane electrodes subject to unipolar injection. Besides pure electroconvection, we also consider the case where a cross-flow is present, generated by a streamwise pressure gradient, in the form of a laminar Poiseuille flow. The effect of charge diffusion, often neglected in previous linear stability analyses, is included in the present study and a transient growth analysis, rarely considered in EHD, is carried out. In the case without cross-flow, a non-zero charge diffusion leads to a lower linear stability threshold and thus to a more unstable low. The transient growth, though enhanced by increasing charge diffusion, remains small and hence cannot fully account for the discrepancy of the linear stability threshold between theoretical and experimental results. When a cross-flow is present, increasing the strength of the electric field in the high-$Re$ Poiseuille flow yields a more unstable flow in both modal and nonmodal stability analyses. Even though the energy analysis and the input-output analysis both indicate that the energy growth directly related to the electric field is small, the electric effect enhances the lift-up mechanism. The symmetry of channel flow with respect to the centerline is broken due to the additional electric field acting in the wall-normal direction. As a result, the centers of the streamwise rolls are shifted towards the injector electrode, and the optimal spanwise wavenumber achieving maximum transient energy growth increases with the strength of the electric field.

physics.flu-dyn↗