arXiv subjects

Hao Sun

Publications and source records attributed to Hao Sun.

At least 469 records · Page 26Linked to original sources

arXiv2020

The Beauville-Narasimhan-Ramanan correspondence for twisted Higgs $V$-bundles and components of parabolic $\text{Sp}(2n,\mathbb{R})$-Higgs moduli Spaces

We generalize the classical Beauville-Narasimhan-Ramanan correspondence to the case of parabolic Higgs bundles with regular singularities and Higgs $V$-bundles. Using this correspondence along with Bott-Morse theoretic techniques we provide an exact component count for moduli spaces of maximal parabolic $\text{Sp}\left( 2n,\mathbb{R} \right)$-Higgs bundles with fixed parabolic structure.

math.AG↗

arXiv2020

MultiSpeech: Multi-Speaker Text to Speech with Transformer

Transformer-based text to speech (TTS) model (e.g., Transformer TTS~\cite{li2019neural}, FastSpeech~\cite{ren2019fastspeech}) has shown the advantages of training and inference efficiency over RNN-based model (e.g., Tacotron~\cite{shen2018natural}) due to its parallel computation in training and/or inference. However, the parallel computation increases the difficulty while learning the alignment between text and speech in Transformer, which is further magnified in the multi-speaker scenario with noisy data and diverse speakers, and hinders the applicability of Transformer for multi-speaker TTS. In this paper, we develop a robust and high-quality multi-speaker Transformer TTS system called MultiSpeech, with several specially designed components/techniques to improve text-to-speech alignment: 1) a diagonal constraint on the weight matrix of encoder-decoder attention in both training and inference; 2) layer normalization on phoneme embedding in encoder to better preserve position information; 3) a bottleneck in decoder pre-net to prevent copy between consecutive speech frames. Experiments on VCTK and LibriTTS multi-speaker datasets demonstrate the effectiveness of MultiSpeech: 1) it synthesizes more robust and better quality multi-speaker voice than naive Transformer based TTS; 2) with a MutiSpeech model as the teacher, we obtain a strong multi-speaker FastSpeech model with almost zero quality degradation while enjoying extremely fast inference speed.

eess.AS↗

arXiv2020

Incremental Bayesian tensor learning for structural monitoring data imputation and response forecasting

There has been increased interest in missing sensor data imputation, which is ubiquitous in the field of structural health monitoring (SHM) due to discontinuous sensing caused by sensor malfunction. To address this fundamental issue, this paper presents an incremental Bayesian tensor learning method for reconstruction of spatiotemporal missing data in SHM and forecasting of structural response. In particular, a spatiotemporal tensor is first constructed followed by Bayesian tensor factorization that extracts latent features for missing data imputation. To enable structural response forecasting based on incomplete sensing data, the tensor decomposition is further integrated with vector autoregression in an incremental learning scheme. The performance of the proposed approach is validated on continuous field-sensing data (including strain and temperature records) of a concrete bridge, based on the assumption that strain time histories are highly correlated to temperature recordings. The results indicate that the proposed probabilistic tensor learning approach is accurate and robust even in the presence of large rates of random missing, structured missing and their combination. The effect of rank selection on the imputation and prediction performance is also investigated. The results show that a better estimation accuracy can be achieved with a higher rank for random missing whereas a lower rank for structured missing.

cs.LG↗

arXiv2020

Non-local Policy Optimization via Diversity-regularized Collaborative Exploration

Conventional Reinforcement Learning (RL) algorithms usually have one single agent learning to solve the task independently. As a result, the agent can only explore a limited part of the state-action space while the learned behavior is highly correlated to the agent's previous experience, making the training prone to a local minimum. In this work, we empower RL with the capability of teamwork and propose a novel non-local policy optimization framework called Diversity-regularized Collaborative Exploration (DiCE). DiCE utilizes a group of heterogeneous agents to explore the environment simultaneously and share the collected experiences. A regularization mechanism is further designed to maintain the diversity of the team and modulate the exploration. We implement the framework in both on-policy and off-policy settings and the experimental results show that DiCE can achieve substantial improvement over the baselines in the MuJoCo locomotion tasks.

cs.LG↗

arXiv2020

Physics informed deep learning for computational elastodynamics without labeled data

Numerical methods such as finite element have been flourishing in the past decades for modeling solid mechanics problems via solving governing partial differential equations (PDEs). A salient aspect that distinguishes these numerical methods is how they approximate the physical fields of interest. Physics-informed deep learning is a novel approach recently developed for modeling PDE solutions and shows promise to solve computational mechanics problems without using any labeled data. The philosophy behind it is to approximate the quantity of interest (e.g., PDE solution variables) by a deep neural network (DNN) and embed the physical law to regularize the network. To this end, training the network is equivalent to minimization of a well-designed loss function that contains the PDE residuals and initial/boundary conditions (I/BCs). In this paper, we present a physics-informed neural network (PINN) with mixed-variable output to model elastodynamics problems without resort to labeled data, in which the I/BCs are hardly imposed. In particular, both the displacement and stress components are taken as the DNN output, inspired by the hybrid finite element analysis, which largely improves the accuracy and trainability of the network. Since the conventional PINN framework augments all the residual loss components in a "soft" manner with Lagrange multipliers, the weakly imposed I/BCs cannot not be well satisfied especially when complex I/BCs are present. To overcome this issue, a composite scheme of DNNs is established based on multiple single DNNs such that the I/BCs can be satisfied forcibly in a "hard" manner. The propose PINN framework is demonstrated on several numerical elasticity examples with different I/BCs, including both static and dynamic problems as well as wave propagation in truncated domains. Results show the promise of PINN in the context of computational mechanics applications.

math.NA↗

arXiv2020

Sparse representation for damage identification of structural systems

Identifying damage of structural systems is typically characterized as an inverse problem which might be ill-conditioned due to aleatory and epistemic uncertainties induced by measurement noise and modeling error. Sparse representation can be used to perform inverse analysis for the case of sparse damage. In this paper, we propose a novel two-stage sensitivity analysis-based framework for both model updating and sparse damage identification. Specifically, an $\ell_2$ Bayesian learning method is firstly developed for updating the intact model and uncertainty quantification so as to set forward a baseline for damage detection. A sparse representation pipeline built on a quasi-$\ell_0$ method, e.g., Sequential Threshold Least Squares (STLS) regression, is then presented for damage localization and quantification. Additionally, Bayesian optimization together with cross validation is developed to heuristically learn hyperparameters from data, which saves the computational cost of hyperparameter tuning and produces more reliable identification result. The proposed framework is verified by three examples, including a 10-story shear-type building, a complex truss structure, and a shake table test of an eight-story steel frame. Results show that the proposed approach is capable of both localizing and quantifying structural damage with high accuracy.

stat.AP↗

arXiv2020

Bridgeland stability conditions on some threefolds of general type

We prove the Bogomolov-Gieseker type inequality conjectured by Bayer, Macri and Toda for some products of three curves. This gives the first examples of Bridgeland stability conditions on some threefolds of general type. The key ingredients are the spreading out technique, Frobenius morphism and Bogomolov's inequality for product type varieties in positive characteristic, proved by the author recently.

math.AG↗

arXiv2020

T-odd generalized and quasi transverse momentum dependent parton distribution in a scalar spectator model

Generalized transverse momentum dependent parton distributions (GTMDs), as mother funtions of transverse momentum dependent parton distributions (TMDs) and generalized parton distributions (GPDs), encode the most general parton structure of hadrons. We calculate four twist-two time reversal odd GTMDs of pion in a scalar spectator model. We study the dependence of GTMDs on the longitudinal momentum fraction $x$ carried by the active quark and the transverse momentum $|\vec k_T|$ for different values of skewness $ξ$ defined as the longitudinal momentum transferred to the proton as well as the total momentum $|\vecΔ_T|$ transferred to the proton. In addition, the quasi-TMDs and quasi-GPDs of pion have also been investigated in this paper.

hep-ph↗

arXiv2020

Evolutionary Stochastic Policy Distillation

Solving the Goal-Conditioned Reward Sparse (GCRS) task is a challenging reinforcement learning problem due to the sparsity of reward signals. In this work, we propose a new formulation of GCRS tasks from the perspective of the drifted random walk on the state space, and design a novel method called Evolutionary Stochastic Policy Distillation (ESPD) to solve them based on the insight of reducing the First Hitting Time of the stochastic process. As a self-imitate approach, ESPD enables a target policy to learn from a series of its stochastic variants through the technique of policy distillation (PD). The learning mechanism of ESPD can be considered as an Evolution Strategy (ES) that applies perturbations upon policy directly on the action space, with a SELECT function to check the superiority of stochastic variants and then use PD to update the policy. The experiments based on the MuJoCo robotics control suite show the high learning efficiency of the proposed method.

cs.LG↗

arXiv2020

LightPAFF: A Two-Stage Distillation Framework for Pre-training and Fine-tuning

While pre-training and fine-tuning, e.g., BERT~\citep{devlin2018bert}, GPT-2~\citep{radford2019language}, have achieved great success in language understanding and generation tasks, the pre-trained models are usually too big for online deployment in terms of both memory cost and inference speed, which hinders them from practical online usage. In this paper, we propose LightPAFF, a Lightweight Pre-training And Fine-tuning Framework that leverages two-stage knowledge distillation to transfer knowledge from a big teacher model to a lightweight student model in both pre-training and fine-tuning stages. In this way the lightweight model can achieve similar accuracy as the big teacher model, but with much fewer parameters and thus faster online inference speed. LightPAFF can support different pre-training methods (such as BERT, GPT-2 and MASS~\citep{song2019mass}) and be applied to many downstream tasks. Experiments on three language understanding tasks, three language modeling tasks and three sequence to sequence generation tasks demonstrate that while achieving similar accuracy with the big BERT, GPT-2 and MASS models, LightPAFF reduces the model size by nearly 5x and improves online inference speed by 5x-7x.

cs.CL↗

arXiv2020

Inclusive diffractive $\rm η_{c}$ production in pp, pA and AA modes at the LHC

In this paper, the inclusive Pomeron-Pomeron, Reggeon-Reggeon, Pomeron-Reggeon as well as gluon-Pomeron(-Reggeon) and photon-Pomeron(-Reggeon) interactions for the $\rm η_{c}$ at the LHC energies have been examined in proton-proton, proton-nucleus and nucleus-nucleus collision modes. The cross section has been computed based on the NRQCD factorization and Regge theory formalism. The cross exchange of Pomeron-Reggeon contribution is important in $\rm pp$ and $\rm pA$ modes. The Pomeron-Pomeron contribution is significant in $\rm AA$ mode. The Pomeron contribution is considerable for $\rm AA$ and $\rm pA$ modes in single diffractive process where $\rm A$ undergoes the diffractive process. Reggeon contribution is sizable in $\rm pp$ and $\rm pA$ modes where $\rm p$ only undergoes the diffractive process. The Pomeron and Reggeon contributions in photon-Pomeron and-Reggeon process remain smaller than that of gluon-Pomeron and-Reggeon processes. Our results show that the experimental study of Reggeon, Pomeron and their cross exchange can be carried out in certain kinematic windows with the specific choice of the mode at the LHC. The investigation can be useful to better constrain the Reggeon and Pomeron parton content. The inclusive process serves as the background to related exclusive processes which should be predicted.

hep-ph↗

arXiv2020

Single spin asymmetry $A_{UL}^{\sin(2ϕ_h-2ϕ_R)}$ in dihadron semi-inclusive DIS

The single longitudinal spin asymmetry $A_{UL}^{\sin(2ϕ_h-2ϕ_R)}$ of dihadron production in semi-inclusive deep inelastic scattering (SIDIS) is examined through helicity-dependent dihadron fragmentation function (DiFF) $G_1^\perp$. The correlation of the longitudinal polarization of a fragmenting quark with the transverse momenta of the produced hadron pair is illustrated by this DiFF. The experimental investigation for this azimuthal asymmetry in dihadron SIDIS by the COMPASS Collaboration has lately yielded a very small signal. Here, the unknown T-odd dihadron fragmentation function $G_1^\perp$ utilizing a spectator model is computed. The model has been successfully used to describe the dihadron production in both the unpolarized and the single polarized processes to access the asymmetry and clarify why the signal is very small. The transverse momentum dependent factorization method, in which the transverse momentum of the final state hadron pair is left unintegrated, has been considered. The $\sin(2ϕ_h-2ϕ_R)$ asymmetry at the COMPASS kinematics is estimated and we compare it with the data. Besides, the predictions on the same asymmetry are also made at the HERMES and Electron Ion Collider.

hep-ph↗

arXiv2020

Physics-informed deep learning for incompressible laminar flows

Physics-informed deep learning has drawn tremendous interest in recent years to solve computational physics problems, whose basic concept is to embed physical laws to constrain/inform neural networks, with the need of less data for training a reliable model. This can be achieved by incorporating the residual of physics equations into the loss function. Through minimizing the loss function, the network could approximate the solution. In this paper, we propose a mixed-variable scheme of physics-informed neural network (PINN) for fluid dynamics and apply it to simulate steady and transient laminar flows at low Reynolds numbers. A parametric study indicates that the mixed-variable scheme can improve the PINN trainability and the solution accuracy. The predicted velocity and pressure fields by the proposed PINN approach are also compared with the reference numerical solutions. Simulation results demonstrate great potential of the proposed PINN for fluid flow simulation with a high accuracy.

physics.flu-dyn↗

arXiv2020

Double spin asymmetry in dihadron production in SIDIS off the longitudinally polarized nucleon target

In this paper we study the double longitudinal spin asymmetry of dihadron production in semi-inclusive deep inelastic scattering (SIDIS). We calculate a unknown twist-3 dihadron fragmentation function $\widetilde{D}^\sphericalangle$ within a spectator model which has been used successfully in describing the dihadron production in both the unpolarized and the single polarized processes. The collinear picture, in which the transverse momentum of the final state hadron pair is integrated out, has been considered. The $\cosϕ_R$ azimuthal asymmetry arises from the coupling $e_L H_1^\sphericalangle$ and the coupling $g_1 \widetilde{D}^\sphericalangle$ is studied. We estimate the $\cosϕ_R$ asymmetry at the kinematics of COMPASS and compare with the data. The prediction at the future Electron Ion Collider (EIC) has also been presented.

hep-ph↗

arXiv2020

Transverse single spin asymmetry $A_{UT}^{\sin(ϕ_h-ϕ_S)}$ for single hadron production in SIDIS

In this paper we study the single spin asymmetry $A_{UT}^{\sin(ϕ_h-ϕ_S)}$ of a single hadron production in semi-inclusive deep inelastic scattering (SIDIS) within the framework of transverse momentum dependent (TMD) factorization up to next-to-leading logarithmic (NLL) order of QCD. The asymmetry is contributed by the convolution of the Sivers function and the unpolarized fragmentation function. Specifically, the Sivers function in the coordinate space and perturbative region can be represented as the convolution of the $C$-coefficients and the corresponding collinear correlation functions, among which the Qiu-Sterman function is the most relevant one. We perform a detailed phenomenological analysis of the Sivers asymmetry at the kinematics of the HERMES and the COMPASS measurements. It is found that the obtained $x_B$-, $z_h$- and $P_{h\perp}$-dependent asymmetries are basically consistent with the HERMES and the COMPASS measurements.

hep-ph↗

arXiv2020

Reciprocal Learning Networks for Human Trajectory Prediction

We observe that the human trajectory is not only forward predictable, but also backward predictable. Both forward and backward trajectories follow the same social norms and obey the same physical constraints with the only difference in their time directions. Based on this unique property, we develop a new approach, called reciprocal learning, for human trajectory prediction. Two networks, forward and backward prediction networks, are tightly coupled, satisfying the reciprocal constraint, which allows them to be jointly learned. Based on this constraint, we borrow the concept of adversarial attacks of deep neural networks, which iteratively modifies the input of the network to match the given or forced network output, and develop a new method for network prediction, called reciprocal attack for matched prediction. It further improves the prediction accuracy. Our experimental results on benchmark datasets demonstrate that our new method outperforms the state-of-the-art methods for human trajectory prediction.

cs.CV↗

arXiv2020

Hard diffractive $\rm η_{c,b}$ hadroproduction at the LHC

In this paper, we investigate the inclusive diffractive hadroproduction for $\rm η_{c}$ and $\rm η_{b}$ at the LHC energies. Based on the NRQCD factorization formalism and the resolved-Pomeron model for the quarkonium production mechanism, we estimate the rapidity, momentum fraction loss dependence of the cross section. We give prediction ratios for single and central diffractive processes with respect to non diffractive process. These inclusive processes are sensitive to gluon content of Pomeron for small-$x$ and Reggeon for large-$x$, useful to study small and large-$x$ physics and good to test different mechanism for $\rm η_{c}$ and $\rm η_{b}$ production at the LHC. They also serve as the background to related exclusive processes thus should be predicted. Our results demonstrate that the Reggeon contribution of diffractive processes can be sizable, even sometimes dominant over Pomeron, and that its study can be useful to better constrain the Reggeon parton content. The experimental study of Reggeon can be carried out in certain kinematic windows.

hep-ph↗

arXiv2020

Phenomenology at the LHC of composite particles from strongly interacting Standard Model fermions via four-fermion operators of NJL type

A new physics scenario shows that four-fermion operators of Nambu-Jona-Lasinio (NJL) type have a strong-coupling UV fixed point, where composite fermions $F$ (bosons $Π$) form as bound states of three (two) SM elementary fermions and they couple to their constituents via effective contact interactions at the composite scale $Λ\approx {\cal O} $(TeV). We present a phenomenological study to investigate such composite particles at the LHC by computing the production cross sections and decay widths of composite fermions in the context of the relevant experiments at the LHC with $pp$ collisions at $\sqrt{s}={\rm 13}$ TeV and $\sqrt{s}={\rm 14}$ TeV. Systematically examining all the different composite particles $F$ and the signatures with which they can manifest, we found a vast spectrum of composite particles $F$ that has not yet been explored at the LHC. Recasting the recent CMS results of the resonant channel $pp\rightarrow e^+F \rightarrow e^+e^- q\bar{q}'$, we find that the composite fermion mass $m_F$ below 4.25 TeV is excluded for $Λ$/$m_F$ = 1. We further highlight the region of parameter space where this specific composite particle $F$ can appear using 3 ab$^{-1}$, expected by the High-Luminosity LHC, computing 3 and 5 $σ$ contour plots of its statistical significance.

hep-ph↗