arXiv subjects

Yi Liao

Publications and source records attributed to Yi Liao.

At least 73 records · Page 4Linked to original sources

arXiv2021

Effective field theory approach to lepton number violating $τ$ decays

We continue our endeavor to investigate lepton number violating (LNV) processes at low energy in the framework of effective field theory (EFT). In this work we study the LNV tau decays $τ^+\rightarrow \ell^-P_i^{+}P_j^{+}$, where $\ell=e,~μ$ and $P^+_{i,j}$ are the lowest-lying charged pseudoscalars $π^+,~K^+$. We analyze the dominant contributions in a series of EFTs from high to low energy scales, namely, the standard model effective field theory (SMEFT), the low-energy effective field theory (LEFT), and the chiral perturbation theory ($χ$PT). The decay branching ratios are expressed in terms of the Wilson coefficients of dimension-five and -seven operators in SMEFT and hadronic low energy constants. These Wilson coefficients involve the first and second generations of quarks and all generations of leptons and thus cannot be explored in low energy processes such as nuclear neutrinoless double decay or LNV kaon decays. Unfortunately, the current experimental upper limits on the branching ratios are too weak to set useful constraints on those coefficients. Or, if we assume the new physics scale is larger than 1 TeV, the branching ratios are well below the current experimental bounds. We also estimate hadronic uncertainties incurred in applying $χ$PT to $τ$ decays by computing one-loop chiral logarithms and attempt to improve convergence of chiral perturbation by employing dispersion relations in the short-distance part of the decay amplitudes.

hep-ph↗

arXiv2021

JPEG Steganography with Embedding Cost Learning and Side-Information Estimation

A great challenge to steganography has arisen with the wide application of steganalysis methods based on convolutional neural networks (CNNs). To this end, embedding cost learning frameworks based on generative adversarial networks (GANs) have been proposed and achieved success for spatial steganography. However, the application of GAN to JPEG steganography is still in the prototype stage; its anti-detectability and training efficiency should be improved. In conventional steganography, research has shown that the side-information calculated from the precover can be used to enhance security. However, it is hard to calculate the side-information without the spatial domain image. In this work, an embedding cost learning framework for JPEG Steganography via a Generative Adversarial Network (JS-GAN) has been proposed, the learned embedding cost can be further adjusted asymmetrically according to the estimated side-information. Experimental results have demonstrated that the proposed method can automatically learn a content-adaptive embedding cost function, and use the estimated side-information properly can effectively improve the security performance. For example, under the attack of a classic steganalyzer GFR with quality factor 75 and 0.4 bpnzAC, the proposed JS-GAN can increase the detection error 2.58% over J-UNIWARD, and the estimated side-information aided version JS-GAN(ESI) can further increase the security performance by 11.25% over JS-GAN.

cs.MM↗

arXiv2021

PanGu-$α$: Large-scale Autoregressive Pretrained Chinese Language Models with Auto-parallel Computation

Large-scale Pretrained Language Models (PLMs) have become the new paradigm for Natural Language Processing (NLP). PLMs with hundreds of billions parameters such as GPT-3 have demonstrated strong performances on natural language understanding and generation with \textit{few-shot in-context} learning. In this work, we present our practice on training large-scale autoregressive language models named PanGu-$α$, with up to 200 billion parameters. PanGu-$α$ is developed under the MindSpore and trained on a cluster of 2048 Ascend 910 AI processors. The training parallelism strategy is implemented based on MindSpore Auto-parallel, which composes five parallelism dimensions to scale the training task to 2048 processors efficiently, including data parallelism, op-level model parallelism, pipeline model parallelism, optimizer model parallelism and rematerialization. To enhance the generalization ability of PanGu-$α$, we collect 1.1TB high-quality Chinese data from a wide range of domains to pretrain the model. We empirically test the generation ability of PanGu-$α$ in various scenarios including text summarization, question answering, dialogue generation, etc. Moreover, we investigate the effect of model scales on the few-shot performances across a broad range of Chinese NLP tasks. The experimental results demonstrate the superior capabilities of PanGu-$α$ in performing various tasks under few-shot or zero-shot settings.

cs.CL↗

arXiv2021

A new derivation of the relationship between diffusion coefficient and entropy in classical Brownian motion by the ensemble method

The diffusion coefficient--a measure of dissipation, and the entropy--a measure of fluctuation are found to be intimately correlated in many physical systems. Unlike the fluctuation dissipation theorem in linear response theory, the correlation is often strongly non-linear. To understand this complex dependence, we consider the classical Brownian diffusion in this work. Under certain rational assumption, i.e. in the bi-component fluid mixture, the mass of the Brownian particle $M$ is far greater than that of the bath molecule $m$, we can adopt the weakly couple limit. Only considering the first-order approximation of the mass ratio $m/M$, we obtain a linear motion equation in the reference frame of the observer as a Brownian particle. Based on this equivalent equation, we get the Hamiltonian at equilibrium. Finally, using canonical ensemble method, we define a new entropy that is similar to the Kolmogorov-Sinai entropy. Further, we present an analytic expression of the relationship between the diffusion coefficient $D$ and the entropy $S$ in the thermal equilibrium, that is to say, $D =\frac{\hbar}{eM} \exp{[S/(k_Bd)]}$, where $d$ is the dimension of the space, $k_B$ the Boltzmann constant, $\hbar $ the reduced Planck constant and $e$ the Euler number. This kind of scaling relation has been well-known and well-tested since the similar one for single component is firstly derived by Rosenfeld with the expansion of volume ratio.

cond-mat.stat-mech↗

arXiv2020

An explicit construction of the dimension-9 operator basis in the standard model effective field theory

We investigate systematically dimension-9 operators in the standard model effective field theory which contains only standard model fields and respects its gauge symmetry. With the help of the Hilbert series approach to classifying operators according to their lepton and baryon numbers and their field contents, we construct the basis of operators explicitly. We remove redundant operators by employing various kinematic and algebraic relations including integration by parts, equations of motion, Schouten identities, Dirac matrix and Fierz identities, and Bianchi identities. We confirm counting of independent operators by analyzing their flavor symmetry relations. All operators violate lepton or baryon number or both, and are thus non-Hermitian. Including Hermitian conjugated operators there are $384|^{ΔL=\pm 2}_{ΔB=0}+10|^{ΔL=0}_{ΔB=\pm 2}+4|^{ΔL=\pm3}_{ΔB=\pm1}+236|^{ΔL=\mp 1}_{ΔB=\pm1}$ operators without referring to fermion generations, and $44874|^{ΔL=\pm 2}_{ΔB=0}+2862|^{ΔL=0}_{ΔB=\pm 2}+486|^{ΔL=\pm3}_{ΔB=\pm1}+42234|^{ΔL=\pm 1}_{ΔB=\mp1}$ operators when three generations of fermions are referred to, where $ΔL,~ΔB$ denote the net lepton and baryon numbers of the operators. Our result provides a starting point for consistent phenomenological studies associated with dimension-9 operators.

hep-ph↗

arXiv2020

Imprint of a new light particle at KOTO?

Recently, the KOTO experiment reported their new preliminary result of searching for the decay $K_L\toπ^0ν\barν$. Three candidate events were observed in the signal region, which exceed significantly the expectation based on the standard model. On the other hand, the new NA62 and previous BNL-E949 experiments yielded a consistent result and confirmed the standard model prediction in the charged meson decay $K^+\toπ^+ν\barν$. Furthermore, the two decays are bound by a well-motivated relation from an analysis of isospin symmetry that is hard to break by the new physics of heavy particles. In this work, we study the issue by a systematic effective field theory approach with three of the simplest scenarios, in which the $K_L$ may decay into a new light neutral particle $X$, i.e., $K_L\toπ^0X$, $K_L\to γγX$, or $K_L\toπ^0XX$. We assess the feasibility of the scenarios by simulations and by incorporating constraints coming from NA62 and other relevant experiments. Our main conclusion is that the scenario $K\toπXX$ for a long lived scalar $X$ seems more credible than the other two when combining distributions and other experimental constraints while the region below the KOTO's blind box provides a good detection environment to search for all three scenarios for a relatively heavy $X$.

hep-ph↗

arXiv2020

Probabilistically Masked Language Model Capable of Autoregressive Generation in Arbitrary Word Order

Masked language model and autoregressive language model are two types of language models. While pretrained masked language models such as BERT overwhelm the line of natural language understanding (NLU) tasks, autoregressive language models such as GPT are especially capable in natural language generation (NLG). In this paper, we propose a probabilistic masking scheme for the masked language model, which we call probabilistically masked language model (PMLM). We implement a specific PMLM with a uniform prior distribution on the masking ratio named u-PMLM. We prove that u-PMLM is equivalent to an autoregressive permutated language model. One main advantage of the model is that it supports text generation in arbitrary order with surprisingly good quality, which could potentially enable new applications over traditional unidirectional generation. Besides, the pretrained u-PMLM also outperforms BERT on a set of downstream NLU tasks.

cs.CL↗

arXiv2020

Geometric phase and topological phase diagram of the one-dimensional $XXZ$ Heisenberg spin chain in a longitudinal field

In this paper, we determine the geometric phase for the one-dimensional $XXZ$ Heisenberg chain with spin-$1/2$, the exchange couple $J$ and the spin anisotropy parameter $Δ$ in a longitudinal field(LF) with the reduced field strength $h$. Using the Jordan-Wigner transformation and the mean-field theory based on the Wick's theorem, a semi-analytical theory has been developed in terms of order parameters which satisfy the self-consistent equations. The values of the order parameters are numerically computed using the matrix-product-state(MPS) method. The validity of the mean-filed theory could be checked through the comparison between the self-consistent solutions and the numerical results. Finally, we draw the the topological phase diagrams in the case $J<0$ and the case $J>0$.

cond-mat.stat-mech↗

arXiv2020

Effective field theory approach to lepton number violating decays $K^\pm\rightarrow π^\mp l^{\pm}_αl^{\pm}_β$: long-distance contribution

This is a sequel to our recent work [1] in which we calculated the lepton number violating (LNV) $K^\pm$ decays due to contact dimension-9 (dim-9) quark-lepton effective interactions that are induced at a high energy scale. In this work we investigate the long-distance contribution to the decays arising from the exchange of a neutrino. These decays can probe LNV interactions involving the second generation of fermions that are not reachable in nuclear neutrinoless double-$β$ decays. Our study is completely formulated in the framework of effective field theories (EFTs), from the standard model effective field theory (SMEFT) through the low energy effective field theory (LEFT) to chiral perturbation theory. We work to the first nontrivial orders in each effective field theory, collect along the way the matching conditions and renormalization group effects, and express the decay branching ratios in terms of the Wilson coefficients associated with the dim-5 and dim-7 operators in SMEFT. Our result is general in that it does not depend on dynamical details of physics at a high scale that induce the effective interactions in SMEFT and in that it does not appeal to any hadronic models. We find that the long-distance contribution overwhelmingly dominates over the contact or short-distance one. Assuming the new physics scale to be around a TeV, the branching ratios are predicted to be below the current experimental upper bounds by several orders of magnitude.

hep-ph↗

arXiv2020

Effective field theory approach to lepton number violating decays $K^\pm\rightarrow π^\mp l^{\pm}l^{\pm}$: short-distance contribution

This is the first paper of our systematic efforts on lepton number violating (LNV) hadronic decays in the effective field theory approach. These decays provide information complementary to popular nuclear neutrinoless double-$β$ ($0νββ$) decay in that they can probe LNV interactions involving heavier quarks and charged leptons. We may call them hadronic $0νββ$ decays in short, though $β$ refers to all charged leptons. In this work we investigate the decays $K^\pm\rightarrowπ^\mp l^{\pm}l^{\pm}$ that arise from short-distance or contact interactions involving four quark fields and two charged lepton fields, which have canonical dimension nine (dim-9) at leading order in low energy effective field theory (LEFT). We make a complete analysis on the basis of all dim-9 operators that violate lepton number by two units, and compute their one-loop QCD renormalization effects. We match these effective interactions in LEFT to those in chiral perturbation theory ($χ$PT) for pseudoscalar mesons, and determine the resulting hadronic low energy constants (LECs) by chiral symmetry and lattice results in the literature. The obtained decay rate is general in that all physics at and above the electroweak scale is completely parameterized by the relevant Wilson coefficients in LEFT and hadronic LECs in $χ$PT. Assuming the standard model effective field theory (SMEFT) is the appropriate effective field theory between some new physics scale and the electroweak scale, we match our LEFT results to SMEFT whose leading effective interactions arise from LNV dim-7 operators. This connection to SMEFT simplifies significantly the interaction structures entering in the kaon decays, and we employ the current experimental bounds to set constraints on the relevant Wilson coefficients in SMEFT.

hep-ph↗

arXiv2020

Thermodynamics and phase transition in rotational Kiselev black hole

In this work, we investigate the thermodynamic properties of rotational Kiselev black holes (KBH). Specifically, we use the first-order approximation of the event horizon (EH) to calculate thermodynamic properties for general equations of state $ω$. These thermodynamic properties include areas, entropies, horizon radii, surface gravities, surface temperatures, Komar energies and irreducible masses at the Cauchy horizon (CH) and EH. We study the products of these thermodynamic quantities, we find that these products are determined by the equation of state $ω$ and strength parameter $α$. In the case of the quintessence matter ($ω=-2/3$), radiation ($ω=1/3$) and dust ($ω=0$), we discuss their properties in detail. We also generalize the Smarr mass formula and Christodoulou-Ruffini mass formula to rotational KBH. Finally we study the phase transition and thermodynamic geometry for rotational KBH with radiation ($ω=1/3$). Through analysis, we find that this phase transition is a second order phase transition. Furthermore, we also obtain the scalar curvature in the thermodynamic geometry framework, indicating that the radiation matter may change the phase transition condition and properties for Kerr black hole.

gr-qc↗

arXiv2019

Zero-Shot Paraphrase Generation with Multilingual Language Models

Leveraging multilingual parallel texts to automatically generate paraphrases has drawn much attention as size of high-quality paraphrase corpus is limited. Round-trip translation, also known as the pivoting method, is a typical approach to this end. However, we notice that the pivoting process involves multiple machine translation models and is likely to incur semantic drift during the two-step translations. In this paper, inspired by the Transformer-based language models, we propose a simple and unified paraphrasing model, which is purely trained on multilingual parallel data and can conduct zero-shot paraphrase generation in one step. Compared with the pivoting approach, paraphrases generated by our model is more semantically similar to the input sentence. Moreover, since our model shares the same architecture as GPT (Radford et al., 2018), we are able to pre-train the model on large-scale unparallel corpus, which further improves the fluency of the output sentences. In addition, we introduce the mechanism of denoising auto-encoder (DAE) to improve diversity and robustness of the model. Experimental results show that our model surpasses the pivoting method in terms of relevance, diversity, fluency and efficiency.

cs.CL↗

arXiv2019

Geometric phase of the one-dimensional Ising chain in a longitudinal field

For the one-dimensional Ising chain with spin-$1/2$ and exchange couple $J$ in a steady transverse field(TF), an analytical theory has well been developed in terms of some topological order parameters such as Berry phase(BP). For a TF Ising chain, the nonzero BP which depends on the exchange couple and the field strength characterizes the corresponding symmetry breaking of parity and time reversal(PT). However, there seems to exist a topological phase transition for the one-dimensional Ising chain in a longitudinal field(LF) with the reduced field strength $ε$. If the LF is added at zero temperature, researchers believe that the LF also could influence the PT-symmetry and there exists the discontinuous BP. But the theoretic characterization has not been well founded. This paper tries to aim at this problem. With the Jordan-Wigner transformation, we give the four-fermion interaction form of the Hamiltonian in the one-dimensional Ising chain with a LF. Further by the method of Wick's theorem and the mean-field theory, the four-fermion interaction is well dealt with. We solve the ground state energy and the ground wave function in the momentum space. We discuss the BP and suggest that there exist nonzero BPs when $ε=0$ in the paramagnetic case where $J<0$ and when $-1<ε<1$, in the diamagnetic case where $J>0$.

cond-mat.stat-mech↗

arXiv2019

Dark matter and LHC phenomenology of a scale invariant scotogenic model

We study the phenomenology of a model that addresses the neutrino mass, dark matter, and generation of the electroweak scale in a single framework. Electroweak symmetry breaking is realized via the Coleman-Weinberg mechanism in a classically scale invariant theory, while the neutrino mass is generated radiatively through interactions with dark matter in a typically scotogenic manner. The model introduces a scalar triplet and singlet and a vector-like fermion doublet that carry an odd parity of $Z_2$, and an even parity scalar singlet that helps preserve classical scale invariance. We sample over the parameter space by taking into account various experimental constraints from the dark matter relic density and direct detection, direct scalar searches, neutrino mass, and charged lepton flavor violating decays. We then examine by detailed simulations possible signatures at the LHC to find some benchmark points of the free parameters. We find that the future high-luminosity LHC will have a significant potential in detecting new physics signals in the dilepton channel.

hep-ph↗

arXiv2019

GPT-based Generation for Classical Chinese Poetry

We present a simple yet effective method for generating high quality classical Chinese poetry with Generative Pre-trained Language Model (GPT). The method adopts a simple GPT model, without using any human crafted rules or features, or designing any additional neural components. While the proposed model learns to generate various forms of classical Chinese poems, including Jueju, Lüshi, various Cipai and Couples, the generated poems are of very high quality. We also propose and implement a method to fine-tune the model to generate acrostic poetry. To the best of our knowledge, this is the first to employ GPT in developing a poetry generation system. We have released an online mini demonstration program on Wechat to show the generation capability of the proposed method for classical Chinese poetry.

cs.CL↗

arXiv2019

Constraints on dimension-seven operators with a derivative in effective field theory for Dirac dark matter

The effective field theory (EFT) for dark matter (DM) has been widely used to investigate dark matter detection in both theoretical prediction and experimental analysis. To form a complete basis of effective operators for Dirac DM EFT at dimension seven, eight new four-fermion operators with a derivative in DM currents have recently been introduced. We discuss the experimental observables and constraints for the theoretical predictions of these new operators to constrain the DM mass and relevant energy scale. The observables from thermal relic abundance, indirect and direct detection, and LHC constraints are presented.

hep-ph↗

arXiv2019

Coupling of homogeneous and heterogeneous melting kinetics in polycrystalline materials

Melting kinetics of polycrystalline materials is analyzed on the basis of a new model which explicitly couples homogeneous and heterogeneous melting mechanisms. The distinct feature of this approach lies in its ability to evaluate not only grain-size-distribution effects on the overall melting kinetics but also competitions between the two melting mechanisms. For the first time, we reveal the three-part structure of temperature-time-transformation diagrams for melting of polycrystalline materials, through which it is possible to determine a critical temperature across which the dominant melting mechanism switches. The critical temperature increases as the mean-grain-diameter decreases following a negative power-law. The results are qualitatively consistent with experimental observations.

physics.app-ph↗

arXiv2019

Renormalization Group Evolution of Dimension-seven Operators in Standard Model Effective Field Theory and Relevant Phenomenology

We showed in a previous publication that there are six independent dimension-seven operators violating both lepton and baryon numbers ($L=-B=1$) and twelve ones violating lepton but preserving baryon number ($L=2,~B=0$) in standard model effective field theory, and we calculated one-loop renormalization for the former six operators. In this work we continue our efforts on renormalization of the operators. It turns out this could become subtle because the operators are connected by nontrivial relations when fermion flavors are counted. This kind of relations does not appear in lower dimensional operators. We show how we can extract anomalous dimension matrix for a flavor-specified basis of operators from counterterms computed for the above flavor-blind operators without introducing singular inverse Yukawa coupling matrices. As a phenomenological application, we investigate renormalization group effects on nuclear neutrinoless double $β$ decay. We also discuss very briefly its analog in the meson sector, $K^\pm\toπ^\mpμ^\pmμ^\pm$, and indicate potential difficulties to compute its decay width.

hep-ph↗