BZPEER

preprint2022arXiv

A bound for twists of $\rm GL_3\times GL_2$ $L$ -functions with composite modulus

Let $π$ be a Hecke-Maass cusp form for $\rm SL_3(\mathbf{Z})$ and let $g$ be a holomorphic or Maass cusp form for $\rm SL_2(\mathbf{Z})$ . Let $χ$ be a primitive Dirichlet character of modulus $M=M_1M_2$ with $M_i$ prime, $i=1,2$ . Suppose that $M^{1/2+2η}<M_1<M^{1-2η}$ with $0<η<1/8$ . Then we have $L\left(\frac{1}{2},π\otimes g \otimes χ\right)\ll_{π,g,\varepsilon} M^{3/2-η+\varepsilon}.$

preprint2022arXiv

Multimodal Dialogue Response Generation

Responsing with image has been recognized as an important capability for an intelligent conversational agent. Yet existing works only focus on exploring the multimodal dialogue models which depend on retrieval-based methods, but neglecting generation methods. To fill in the gaps, we first present a multimodal dialogue generation model, which takes the dialogue history as input, then generates a textual sequence or an image as response. Learning such a model often requires multimodal dialogues containing both texts and images which are difficult to obtain. Motivated by the challenge in practice, we consider multimodal dialogue generation under a natural assumption that only limited training examples are available. In such a low-resource setting, we devise a novel conversational agent, Divter, in order to isolate parameters that depend on multimodal dialogues from the entire generation model. By this means, the major part of the model can be learned from a large number of text-only dialogues and text-image pairs respectively, then the whole parameters can be well fitted using the limited training examples. Extensive experiments demonstrate our method achieves state-of-the-art results in both automatic and human evaluation, and can generate informative text and high-resolution image responses.

preprint2022arXiv

On some estimates involving Fourier coefficients of Maass cusp forms

Let $f$ be a Hecke-Maass cusp form for $\rm SL_2(\mathbb{Z})$ with Laplace eigenvalue $λ_f(Δ)=1/4+μ^2$ and let $λ_f(n)$ be its $n$ -th normalized Fourier coefficient. It is proved that, uniformly in $α, β\in \mathbb{R}$ , $\sum_{n \leq X}λ_f(n)e\left(αn^2+βn\right) \ll X^{7/8+\varepsilon}λ_f(Δ)^{1/2+\varepsilon},$ where the implied constant depends only on $\varepsilon$ . We also consider the summation function of $λ_f(n)$ and under the Ramanujan conjecture we are able to prove $\sum_{n \leq X}λ_f(n)\ll X^{1/3+\varepsilon}λ_f(Δ)^{4/9+\varepsilon}$ with the implied constant depending only on $\varepsilon$ .

preprint2022arXiv

PromDA: Prompt-based Data Augmentation for Low-Resource NLU Tasks

This paper focuses on the Data Augmentation for low-resource Natural Language Understanding (NLU) tasks. We propose Prompt-based D}ata Augmentation model (PromDA) which only trains small-scale Soft Prompt (i.e., a set of trainable vectors) in the frozen Pre-trained Language Models (PLMs). This avoids human effort in collecting unlabeled in-domain data and maintains the quality of generated synthetic data. In addition, PromDA generates synthetic data via two different views and filters out the low-quality data using NLU models. Experiments on four benchmarks show that synthetic data produced by PromDA successfully boost up the performance of NLU models which consistently outperform several competitive baseline models, including a state-of-the-art semi-supervised model using unlabeled in-domain data. The synthetic data from PromDA are also complementary with unlabeled in-domain data. The NLU models can be further improved when they are combined for training.

preprint2022arXiv

Stylized Knowledge-Grounded Dialogue Generation via Disentangled Template Rewriting

Current Knowledge-Grounded Dialogue Generation (KDG) models specialize in producing rational and factual responses. However, to establish long-term relationships with users, the KDG model needs the capability to generate responses in a desired style or attribute. Thus, we study a new problem: Stylized Knowledge-Grounded Dialogue Generation (SKDG). It presents two challenges: (1) How to train a SKDG model where no <context, knowledge, stylized response> triples are available. (2) How to cohere with context and preserve the knowledge when generating a stylized response. In this paper, we propose a novel disentangled template rewriting (DTR) method which generates responses via combing disentangled style templates (from monolingual stylized corpus) and content templates (from KDG corpus). The entire framework is end-to-end differentiable and learned without supervision. Extensive experiments on two benchmarks indicate that DTR achieves a significant improvement on all evaluation metrics compared with previous state-of-the-art stylized dialogue generation methods. Besides, DTR achieves comparable performance with the state-of-the-art KDG methods in standard KDG evaluation setting.

preprint2021arXiv

Analytic twists of $\rm GL_3\times \rm GL_2$ automorphic forms

Let $π$ be a Hecke--Maass cusp form for $\rm SL_3(\mathbb{Z})$ with normalized Hecke eigenvalues $λ_π(n,r)$ . Let $f$ be a holomorphic or Maass cusp form for $\rm SL_2(\mathbb{Z})$ with normalized Hecke eigenvalues $λ_f(n)$ . In this paper, we are concerned with obtaining nontrivial estimates for the sum \begin{equation*} \sum_{r,n\geq 1}λ_π(n,r)λ_f(n)e\left(t\,φ(r^2n/N)\right)V\left(r^2n/N\right), \end{equation*} where $e(x)=e^{2πix}$ , $V(x)\in \mathcal{C}_c^{\infty}(0,\infty)$ , $t\geq 1$ is a large parameter and $φ(x)$ is some real-valued smooth function. As applications, we give an improved subconvexity bound for $\rm GL_3\times \rm GL_2$ $L$ -functions in the $t$ -aspect, and under the Ramanujan--Petersson conjecture we derive the following bound for sums of $\rm GL_3\times \rm GL_2$ Fourier coefficients \begin{equation*} \sum_{r^2n\leq x}λ_π(r,n)λ_f(n)\ll_{π, f, \varepsilon} x^{5/7-1/364+\varepsilon} \end{equation*} for any $\varepsilon>0$ , which breaks for the first time the barrier $O(x^{5/7+\varepsilon})$ in a work by Friedlander--Iwaniec.

preprint2020arXiv

A Bessel $δ$ -method and hybrid bounds for $\mathrm{GL}_2$

Let $g$ be a primitive holomorphic or Maass newform for $Γ_0(D)$ . In this paper, by studying the Bessel integrals associated to $g$ , we prove an asymptotic Bessel $δ$ -identity associated to $g$ . Among other applications, we prove the following hybrid subconvexity bound $L\left(1/2+it,g\otimes χ\right)\ll_{g,\varepsilon} (q(1+|t|))^{\varepsilon}q^{3/8}(1+|t|)^{1/3}$ for any $\varepsilon>0$ , where $χ\bmod q$ is a primitive Dirichlet character with $(q, D)=1$ . This improves the previous known result.

preprint2020arXiv

A subconvex bound for twisted $L$ -functions

Let $\mathfrak{q}>2$ be a prime number, $χ$ a primitive Dirichlet character modulo $\mathfrak{q}$ and $f$ a primitive holomorphic cusp form or a Hecke-Maass cusp form of level $\mathfrak{q}$ and trivial nebentypus. We prove the subconvex bound $L(1/2,f\otimes χ)\ll \mathfrak{q}^{1/2-1/12+\varepsilon},$ where the implicit constant depends only on the archimedean parameter of $f$ and $\varepsilon$ . The main input is a modifying trivial delta method developed in [1].

preprint2020arXiv

The Burgess bound via a trivial delta method

Let $g$ be a fixed Hecke cusp form for $\mathrm{SL}(2,\mathbb{Z})$ and $χ$ be a primitive Dirichlet character of conductor $M$ . The best known subconvex bound for $L(1/2,g\otimes χ)$ is of Burgess strength. The bound was proved by a couple of methods: shifted convolution sums and the Petersson/Kuznetsov formula analysis. It is natural to ask what inputs are really needed to prove a Burgess-type bound on $\rm GL(2)$ . In this paper, we give a new proof of the Burgess-type bounds ${L(1/2,g\otimes χ)\ll_{g,\varepsilon} M^{1/2-1/8+\varepsilon}}$ and $L(1/2,χ)\ll_{\varepsilon} M^{1/4-1/16+\varepsilon}$ that does not require the basic tools of the previous proofs and instead uses a trivial delta method.

preprint2016arXiv

A note on simultaneous nonvanishing of Dirichlet $L$ -functions and twists of Hecke-Maass $L$ -functions

In this note, we prove that given a Hecke-Maass cusp form $f$ for $SL_2(\mathbb{Z})$ and a sufficiently large integer $q=q_1q_2$ with $q_j\asymp \sqrt{q}$ being prime numbers for $j=1,2$ , there exists a primitive Dirichlet character $χ$ of conductor $q$ such that $L\left(\frac{1}{2},f\otimes χ\right)L\left(\frac{1}{2},χ\right)\neq 0$ . To prove this, we establish asymptotic formulas of $L\left(\frac{1}{2},f\otimes χ\right)L\left(\frac{1}{2},χ\right)$ over the family of even primitive Dirichlet characters $χ$ of conductor $q$ for more general $q$ .

preprint2016arXiv

Majorana fermions in a superconducting Mobius strip

Recently, much attention has been paid to search for Majorana fermions in solid-state systems. Among various proposals there is one based on radio-frequency superconducting quantum interference devices (rf-SQUIDs), in which the appearance of 4 $π$ -period energy-phase relations is regarded as smoking-gun evidence of Majorana fermion states. Here we report the observation of truncated 4 $π$ -period (i.e., 2 $π$ -period but fully skewed) oscillatory patterns of contact resistance on rf-SQUIDs constructed on the surface of three-dimensional topological insulator Bi $_2$ Te $_3$ . The results reveal the existence of 1/2 fractional modes of Cooper pairs and the occurrence of parity switchings, both of which are necessary signatures accompanied with the formation of Majorana fermion states.

preprint2016arXiv

Shifted convolution sums of $GL_3$ cusp forms with $θ$ -series

Let $A_f(1,n)$ be the normalized Fourier coefficients of a Hecke-Maass cusp form $f$ for $SL_3(\mathbb{Z})$ and $r_3(n)=\#\left\{(n_1,n_2,n_3)\in \mathbb{Z}^3:n_1^2+n_2^2+n_3^2=n\right\}.$ Let $1\leq h\leq X$ and $ϕ(x)$ be a smooth function compactly supported on $[1/2,1]$ . It is shown that $\sum_{n\geq 1}A_f(1,n+h)r_3(n)ϕ\left(\frac{n}{X}\right) \ll_{f,\varepsilon} X^{\frac{3}{2}-\frac{1}{8}+\varepsilon}$ uniformly with respect to the shift $h$ .

preprint2015arXiv

Sums of the triple divisor function over values of a ternary quadratic form

Let $τ_3(n)$ be the triple divisor function which is the number of solutions of the equation $d_1d_2d_3=n$ in natural numbers. It is shown that $\sum_{1\leq n_1,n_2,n_3\leq \sqrt{x}}τ_3(n_1^2+n_2^2+n_3^2)=c_1x^{\frac{3}{2}}(\log x)^2+ c_2x^{\frac{3}{2}}\log x +c_3x^{\frac{3}{2}} +O_{\varepsilon}(x^{\frac{11}{8}+\varepsilon})$ for some constants $c_1$ , $c_2$ and $c_3$ .

preprint2012arXiv

Shot noise of spin current and spin transfer torque

We report the theoretical investigation of noise spectrum of spin current and spin transfer torque for non-colinear spin polarized transport in a spin-valve device which consists of normal scattering region connected by two ferromagnetic electrodes. Our theory was developed using non-equilibrium Green's function method and general non-linear $S^σ-V$ and $S^τ-V$ relations were derived as a function of angle $θ$ between magnetization of two leads. We have applied our theory to a quantum dot system with a resonant level coupled with two ferromagnetic electrodes. It was found that for the MNM system, the auto-correlation of spin current is enough to characterize the fluctuation of spin current. For a system with three ferromagnetic layers, however, both auto-correlation and cross-correlation of spin current are needed to characterize the noise spectrum of spin current. Furthermore, the spin transfer torque and the torque noise were studied for the MNM system. For a quantum dot with a resonant level, the derivative of spin torque with respect to bias voltage is proportional to $\sinθ$ when the system is far away from the resonance. When the system is near the resonance, the spin transfer torque becomes non-sinusoidal function of $θ$ . The derivative of noise spectrum of spin transfer torque with respect to the bias voltage $N_τ$ behaves differently when the system is near or far away from the resonance. Specifically, the differential shot noise of spin transfer torque $N_τ$ is a concave function of $θ$ near the resonance while it becomes convex function of $θ$ far away from resonance. For certain bias voltages, the period $N_τ(θ)$ becomes $π$ instead of $2π$ . For small $θ$ , it was found that the differential shot noise of spin transfer torque is very sensitive to the bias voltage and the other system parameters.

Qingfeng Sun

Research overview

Is this your work?

14 publications

A bound for twists of $\rm GL_3\times GL_2$ $L$ -functions with composite modulus

Multimodal Dialogue Response Generation

On some estimates involving Fourier coefficients of Maass cusp forms

PromDA: Prompt-based Data Augmentation for Low-Resource NLU Tasks

Stylized Knowledge-Grounded Dialogue Generation via Disentangled Template Rewriting

Analytic twists of $\rm GL_3\times \rm GL_2$ automorphic forms

A Bessel $δ$ -method and hybrid bounds for $\mathrm{GL}_2$

A subconvex bound for twisted $L$ -functions

The Burgess bound via a trivial delta method

A note on simultaneous nonvanishing of Dirichlet $L$ -functions and twists of Hecke-Maass $L$ -functions

Majorana fermions in a superconducting Mobius strip

Shifted convolution sums of $GL_3$ cusp forms with $θ$ -series

Sums of the triple divisor function over values of a ternary quadratic form

Shot noise of spin current and spin transfer torque

See the researcher in context

Building this map preview

Qingfeng Sun

Research overview

Is this your work?

14 publications

A bound for twists of GL3×GL2\rm GL_3\times GL_2GL3​×GL2​ LLL-functions with composite modulus

Multimodal Dialogue Response Generation

On some estimates involving Fourier coefficients of Maass cusp forms

PromDA: Prompt-based Data Augmentation for Low-Resource NLU Tasks

Stylized Knowledge-Grounded Dialogue Generation via Disentangled Template Rewriting

Analytic twists of GL3×GL2\rm GL_3\times \rm GL_2GL3​×GL2​ automorphic forms

A Bessel δδδ-method and hybrid bounds for GL2\mathrm{GL}_2GL2​

A subconvex bound for twisted LLL-functions

The Burgess bound via a trivial delta method

A note on simultaneous nonvanishing of Dirichlet LLL-functions and twists of Hecke-Maass LLL-functions

Majorana fermions in a superconducting Mobius strip

Shifted convolution sums of GL3GL_3GL3​ cusp forms with θθθ-series

Sums of the triple divisor function over values of a ternary quadratic form

Shot noise of spin current and spin transfer torque

See the researcher in context

Building this map preview

A bound for twists of $\rm GL_3\times GL_2$ $L$ -functions with composite modulus

Analytic twists of $\rm GL_3\times \rm GL_2$ automorphic forms

A Bessel $δ$ -method and hybrid bounds for $\mathrm{GL}_2$

A subconvex bound for twisted $L$ -functions

A note on simultaneous nonvanishing of Dirichlet $L$ -functions and twists of Hecke-Maass $L$ -functions

Shifted convolution sums of $GL_3$ cusp forms with $θ$ -series