arXiv subjects

Jun Gao

Publications and source records attributed to Jun Gao.

At least 235 records · Page 13Linked to original sources

arXiv2021

On the rainbow matching conjecture for 3-uniform hypergraphs

Aharoni and Howard, and, independently, Huang, Loh, and Sudakov proposed the following rainbow version of Erdős matching conjecture: For positive integers $n,k,m$ with $n\ge km$, if each of the families $F_1,\ldots, F_m\subseteq {[n]\choose k}$ has size more than $\max\{\binom{n}{k} - \binom{n-m+1}{k}, \binom{km-1}{k}\}$, then there exist pairwise disjoint subsets $e_1,\dots, e_m$ such that $e_i\in F_i$ for all $i\in [m]$. We prove that there exists an absolute constant $n_0$ such that this rainbow version holds for $k=3$ and $n\geq n_0$. We convert this rainbow matching problem to a matching problem on a special hypergraph $H$. We then combine several existing techniques on matchings in uniform hypergraphs: find an absorbing matching $M$ in $H$; use a randomization process of Alon et al. to find an almost regular subgraph of $H-V(M)$; and find an almost perfect matching in $H-V(M)$. To complete the process, we also need to prove a new result on matchings in 3-uniform hypergraphs, which can be viewed as a stability version of a result of Łuczak and Mieczkowska and might be of independent interest.

math.CO↗

arXiv2021

Experimentally Detecting Quantized Zak Phases without Chiral Symmetry in Photonic Lattices

Symmetries play a major role in identifying topological phases of matter and in establishing a direct connection between protected edge states and topological bulk invariants via the bulk-boundary correspondence. One-dimensional lattices are deemed to be protected by chiral symmetry, exhibiting quantized Zak phases and protected edge states, but not for all cases. Here, we experimentally realize an extended Su-Schrieffer-Heeger model with broken chiral symmetry by engineering one-dimensional zigzag photonic lattices, where the long-range hopping breaks chiral symmetry but ensures the existence of inversion symmetry. By the averaged mean displacement method, we detect topological invariants directly in the bulk through the continuous-time quantum walk of photons. Our results demonstrate that inversion symmetry protects the quantized Zak phase, but edge states can disappear in the topological nontrivial phase, thus breaking the conventional bulk-boundary correspondence. Our photonic lattice provides a useful platform to study the interplay among topological phases, symmetries, and the bulk-boundary correspondence.

physics.optics↗

arXiv2021

Differential distributions for Single Top Quark Production at the LHeC

We present a phenomenological study of the single top (anti-)quark production with leptonic decays at the Large Hadron electron Collider (LHeC) at the next-to-leading-order (NLO) in QCD. We focus on various differential distributions in a fiducial region. The NLO corrections can reduce the fiducial cross section by 14%. We find the NLO predictions exhibit strong stability under scale variations for most observables considered while the scale variations at the leading-order (LO) dominated in the theoretical uncertainties. We propose a method of determining the top-quark mass using the measurement of the average transverse momentum of the charged lepton. The scale variations at the NLO induce a theoretical uncertainty of about 1.3 GeV of the extracted top-quark mass. The statistical error of the extracted top-quark mass amounts to 1.1 GeV. We also investigate the impact of the QCD corrections and the scale variations in searches of the anomalous Wtb couplings.

hep-ph↗

arXiv2021

3DStyleNet: Creating 3D Shapes with Geometric and Texture Style Variations

We propose a method to create plausible geometric and texture style variations of 3D objects in the quest to democratize 3D content creation. Given a pair of textured source and target objects, our method predicts a part-aware affine transformation field that naturally warps the source shape to imitate the overall geometric style of the target. In addition, the texture style of the target is transferred to the warped source object with the help of a multi-view differentiable renderer. Our model, 3DStyleNet, is composed of two sub-networks trained in two stages. First, the geometric style network is trained on a large set of untextured 3D shapes. Second, we jointly optimize our geometric style network and a pre-trained image style transfer network with losses defined over both the geometry and the rendering of the result. Given a small set of high-quality textured objects, our method can create many novel stylized shapes, resulting in effortless 3D content creation and style-ware data augmentation. We showcase our approach qualitatively on 3D content stylization, and provide user studies to validate the quality of our results. In addition, our method can serve as a valuable tool to create 3D data augmentations for computer vision tasks. Extensive quantitative analysis shows that 3DStyleNet outperforms alternative data augmentation techniques for the downstream task of single-image 3D reconstruction.

cs.CV↗

arXiv2021

Rainbow independent sets in graphs with maximum degree two

Given a graph $G$, let $f_{G}(n,m)$ be the minimal number $k$ such that every $k$ independent $n$-sets in $G$ have a rainbow $m$-set. Let $\mathcal{D}(2)$ be the family of all graphs with maximum degree at most two. Aharoni et al. (2019) conjectured that (i) $f_G(n,n-1)=n-1$ for all graphs $G\in\mathcal{D}(2)$ and (ii) $f_{C_t}(n,n)=n$ for $t\ge 2n+1$. Lv and Lu (2020) showed that the conjecture (ii) holds when $t=2n+1$. In this article, we show that the conjecture (ii) holds for $t\ge\frac{1}{3}n^2+\frac{44}{9}n$. Let $C_t$ be a cycle of length $t$ with vertices being arranged in a clockwise order. An ordered set $I=(a_1,a_2,\ldots,a_n)$ on $C_t$ is called a $2$-jump independent $n$-set of $C_t$ if $a_{i+1}-a_i=2\pmod{t}$ for any $1\le i\le n-1$. We also show that a collection of 2-jump independent $n$-sets $\mathcal{F}$ of $C_t$ with $|\mathcal{F}|=n$ admits a rainbow independent $n$-set, i.e. (ii) holds if we restrict $\mathcal{F}$ on the family of 2-jump independent $n$-sets. Moreover, we prove that if the conjecture (ii) holds, then (i) holds for all graphs $G\in\mathcal{D}(2)$ with $c_e(G)\le 4$, where $c_e(G)$ is the number of components of $G$ isomorphic to cycles of even lengths.

math.CO↗

arXiv2021

Evaluating Deep Graph Neural Networks

Graph Neural Networks (GNNs) have already been widely applied in various graph mining tasks. However, they suffer from the shallow architecture issue, which is the key impediment that hinders the model performance improvement. Although several relevant approaches have been proposed, none of the existing studies provides an in-depth understanding of the root causes of performance degradation in deep GNNs. In this paper, we conduct the first systematic experimental evaluation to present the fundamental limitations of shallow architectures. Based on the experimental results, we answer the following two essential questions: (1) what actually leads to the compromised performance of deep GNNs; (2) when we need and how to build deep GNNs. The answers to the above questions provide empirical insights and guidelines for researchers to design deep and well-performed GNNs. To show the effectiveness of our proposed guidelines, we present Deep Graph Multi-Layer Perceptron (DGMLP), a powerful approach (a paradigm in its own right) that helps guide deep GNN designs. Experimental results demonstrate three advantages of DGMLP: 1) high accuracy -- it achieves state-of-the-art node classification performance on various datasets; 2) high flexibility -- it can flexibly choose different propagation and transformation depths according to graph size and sparsity; 3) high scalability and efficiency -- it supports fast training on large-scale graphs. Our code is available in https://github.com/zwt233/DGMLP.

cs.LG↗

arXiv2021

A Survey on Contact Tracing: the Latest Advancements and Challenges

Infectious diseases are caused by pathogenic microorganisms, such as bacteria, viruses, parasites or fungi, which can be spread, directly or indirectly, from one person to another. Infectious diseases pose a serious threat to human health, especially COVID-19 that has became a serious worldwide health concern since the end of 2019. Contact tracing is the process of identifying, assessing, and managing people who have been exposed to a disease to prevent its onward transmission. Contact tracing can help us better understand the transmission link of the virus, whereby better interrupting its transmission. Given the worldwide pandemic of COVID-19, contact tracing has become one of the most critical measures to effectively curb the spread of the virus. This paper presents a comprehensive survey on contact tracing, with a detailed coverage of the recent advancements the models, digital technologies, protocols and issues involved in contact tracing. The current challenges as well as future directions of contact tracing technologies are also presented.

cs.SI↗

arXiv2021

Topologically Protecting Squeezed Light on a Photonic Chip

Squeezed light is a critical resource in quantum sensing and information processing. Due to the inherently weak optical nonlinearity and limited interaction volume, considerable pump power is typically needed to obtain efficient interactions to generate squeezed light in bulk crystals. Integrated photonics offers an elegant way to increase the nonlinearity by confining light strictly inside the waveguide. For the construction of large-scale quantum systems performing many-photon operations, it is essential to integrate various functional modules on a chip. However, fabrication imperfections and transmission crosstalk may add unwanted diffraction and coupling to other photonic elements, reducing the quality of squeezing. Here, by introducing the topological phase, we experimentally demonstrate the topologically protected nonlinear process of spontaneous four-wave mixing enabling the generation of squeezed light on a silica chip. We measure the cross-correlations at different evolution distances for various topological sites and verify the non-classical features with high fidelity. The squeezing parameters are measured to certify the protection of cavity-free, strongly squeezed states. The demonstration of topological protection for squeezed light on a chip brings new opportunities for quantum integrated photonics, opening novel approaches for the design of advanced multi-photon circuits.

quant-ph↗

arXiv2021

Image GANs meet Differentiable Rendering for Inverse Graphics and Interpretable 3D Neural Rendering

Differentiable rendering has paved the way to training neural networks to perform "inverse graphics" tasks such as predicting 3D geometry from monocular photographs. To train high performing models, most of the current approaches rely on multi-view imagery which are not readily available in practice. Recent Generative Adversarial Networks (GANs) that synthesize images, in contrast, seem to acquire 3D knowledge implicitly during training: object viewpoints can be manipulated by simply manipulating the latent codes. However, these latent codes often lack further physical interpretation and thus GANs cannot easily be inverted to perform explicit 3D reasoning. In this paper, we aim to extract and disentangle 3D knowledge learned by generative models by utilizing differentiable renderers. Key to our approach is to exploit GANs as a multi-view data generator to train an inverse graphics network using an off-the-shelf differentiable renderer, and the trained inverse graphics network as a teacher to disentangle the GAN's latent code into interpretable 3D properties. The entire architecture is trained iteratively using cycle consistency losses. We show that our approach significantly outperforms state-of-the-art inverse graphics networks trained on existing datasets, both quantitatively and via user studies. We further showcase the disentangled GAN as a controllable 3D "neural renderer", complementing traditional graphics renderers.

cs.CV↗

arXiv2021

DatasetGAN: Efficient Labeled Data Factory with Minimal Human Effort

We introduce DatasetGAN: an automatic procedure to generate massive datasets of high-quality semantically segmented images requiring minimal human effort. Current deep networks are extremely data-hungry, benefiting from training on large-scale datasets, which are time consuming to annotate. Our method relies on the power of recent GANs to generate realistic images. We show how the GAN latent code can be decoded to produce a semantic segmentation of the image. Training the decoder only needs a few labeled examples to generalize to the rest of the latent space, resulting in an infinite annotated dataset generator! These generated datasets can then be used for training any computer vision architecture just as real datasets are. As only a few images need to be manually segmented, it becomes possible to annotate images in extreme detail and generate datasets with rich object and part segmentations. To showcase the power of our approach, we generated datasets for 7 image segmentation tasks which include pixel-level labels for 34 human face parts, and 32 car parts. Our approach outperforms all semi-supervised baselines significantly and is on par with fully supervised methods, which in some cases require as much as 100x more annotated data as our method.

cs.CV↗

arXiv2021

Quantum Computation for Pricing the Collateralized Debt Obligations

Collateralized debt obligation (CDO) has been one of the most commonly used structured financial products and is intensively studied in quantitative finance. By setting the asset pool into different tranches, it effectively works out and redistributes credit risks and returns to meet the risk preferences for different tranche investors. The copula models of various kinds are normally used for pricing CDOs, and the Monte Carlo simulations are required to get their numerical solution. Here we implement two typical CDO models, the single-factor Gaussian copula model and Normal Inverse Gaussian copula model, and by applying the conditional independence approach, we manage to load each model of distribution in quantum circuits. We then apply quantum amplitude estimation as an alternative to Monte Carlo simulation for CDO pricing. We demonstrate the quantum computation results using IBM Qiskit. Our work addresses a useful task in finance instrument pricing, significantly broadening the application scope for quantum computing in finance.

q-fin.RM↗

arXiv2021

A strengthening on odd cycles in graphs of given chromatic number

Resolving a conjecture of Bollobás and Erdős, Gyárfás proved that every graph $G$ of chromatic number $k+1\geq 3$ contains cycles of $\lfloor\frac{k}{2}\rfloor$ distinct odd lengths. We strengthen this prominent result by showing that such $G$ contains cycles of $\lfloor\frac{k}{2}\rfloor$ consecutive odd lengths. Along the way, combining extremal and structural tools, we prove a stronger statement that every graph of chromatic number $k+1\geq 7$ contains $k$ cycles of consecutive lengths, except that some block is $K_{k+1}$. As corollaries, this confirms a conjecture of Verstraëte and answers a question of Moore and West.

math.CO↗

arXiv2021

Top-quark mass determination from t-channel single top production at the LHC

We study the determination of the top-quark mass using leptonic observables in t-channel single top-quark production at the LHC. We demonstrate sensitivity of transverse momentum of the charged lepton on the input top-quark mass. We present predictions at next-to-next-to-leading order (NNLO) in QCD with narrow width approximation and structure function approach. Further corrections due to parton shower and hadronization, non-resonant and non-factorized contributions are discussed. To reduce impact of SM backgrounds we propose to use the charge weighted distribution for the measurement, i.e., differences between distributions of charged lepton with positive and negative charges. By modeling both signal and background processes, we found the projections for (HL-)LHC to be promising, with a total theoretical uncertainty on the extracted top-quark mass of about 1 $\sim$ 2 GeV.

hep-ph↗

arXiv2021

Minimizing the number of edges in $\mathcal{C}_{\ge r}$-saturated graphs

Given a family of graphs $\mathcal{F}$, a graph $G$ is said to be $\mathcal{F}$-saturated if $G$ does not contain a copy of $F$ as a subgraph for any $F\in\mathcal{F}$ but the addition of any edge $e\notin E(G)$ creates at least one copy of some $F\in\mathcal{F}$ within $G$. The minimum size of an $\mathcal{F}$-saturated graph on $n$ vertices are called the saturation number, denoted by $\sat(n, \mathcal{F})$. Let $\mathcal{C}_{\ge r}$ be the family of cycles of length at least $r$. Ferrara et al. (2012) gave lower and upper bounds of $\sat(n, C_{\ge r})$ and determined the exact values of $\sat(n, C_{\ge r})$ for $3\le r\le 5$. In this paper, we determine the exact value of $\sat(n,\mathcal{C}_{\ge r})$ for $r=6$ and $28\le \frac{n}2\le r\le n$ and give new upper and lower bounds for the other cases.

math.CO↗

arXiv2021

Energy-energy correlation in hadronic Higgs decays: analytic results and phenomenology at NLO

In this work we complete the investigation of the recently introduced energy-energy correlation (EEC) function in hadronic Higgs decays at next-to-leading order (NLO) in fixed-order perturbation theory in the limit of vanishing light quark masses. The full analytic NLO result for the previously unknown EEC in the $H \to q \bar{q} + X$ channel is given in terms of classical polylogarithms and cross-checked against a numerical calculation. In addition to that, we discuss further corrections to predictions of the Higgs EEC event shape variable, including quark mass corrections, effects of parton shower and hadronization. We also estimate the statistical error on the measurements of the Higgs EEC at future Higgs factories and compare with the current perturbative uncertainty.

hep-ph↗

arXiv2021

A unified proof of conjectures on cycle lengths in graphs

In this paper, we prove a tight minimum degree condition in general graphs for the existence of paths between two given endpoints, whose lengths form a long arithmetic progression with common difference one or two. This allows us to obtain a number of exact and optimal results on cycle lengths in graphs of given minimum degree, connectivity or chromatic number. More precisely, we prove the following statements by a unified approach. (1) Every graph $G$ with minimum degree at least $k+1$ contains cycles of all even lengths modulo $k$; in addition, if $G$ is 2-connected and non-bipartite, then it contains cycles of all lengths modulo $k$. (2) For all $k\geq 3$, every $k$-connected graph contains a cycle of length zero modulo $k$. (3) Every 3-connected non-bipartite graph with minimum degree at least $k+1$ contains $k$ cycles of consecutive lengths. (4) Every graph with chromatic number at least $k+2$ contains $k$ cycles of consecutive lengths. The first statement is a conjecture of Thomassen, the second is a conjecture of Dean, the third is a tight answer to a question of Bondy and Vince, and the fourth is a conjecture of Sudakov and Verstraëte. All of the above results are best possible.

math.CO↗

arXiv2021

Constraints on neutrino non-standard interactions from LHC data with large missing transverse momentum

The possible non-standard interactions (NSIs) of neutrinos with matter plays important role in the global determination of neutrino properties. In our study we select various data sets from LHC measurements at 13 TeV with integrated luminosities of $35 \sim 139$ fb$^{-1}$, including production of a single jet, photon, $W/Z$ boson, or charged lepton accompanied with large missing transverse momentum. We derive constraints on neutral-current NSIs with quarks imposed by different data sets in a framework of either effective operators or simplified $Z'$ models. We use theoretical predictions of productions induced by NSIs at next-to-leading order in QCD matched with parton showering which stabilize the theory predictions and result in more robust constraints. In a simplified $Z'$ model we obtain a 95% CLs upper limit on the conventional NSI strength $ε$ of 0.042 and 0.0028 for a $Z'$ mass of 0.2 and 2 TeV respectively. We also discuss possible improvements from future runs of LHC with higher luminosities.

hep-ph↗

arXiv2020

Quantum Advantage with Timestamp Membosonsampling

Quantum computer, harnessing quantum superposition to boost a parallel computational power, promises to outperform its classical counterparts and offer an exponentially increased scaling. The term "quantum advantage" was proposed to mark the key point when people can solve a classically intractable problem by artificially controlling a quantum system in an unprecedented scale, even without error correction or known practical applications. Boson sampling, a problem about quantum evolutions of multi-photons on multimode photonic networks, as well as its variants, has been considered as a promising candidate to reach this milestone. However, the current photonic platforms suffer from the scaling problems, both in photon numbers and circuit modes. Here, we propose a new variant of the problem, timestamp membosonsampling, exploiting the timestamp information of single photons as free resources, and the scaling of the problem can be in principle extended to infinitely large. We experimentally verify the scheme on a self-looped photonic chip inspired by memristor, and obtain multi-photon registrations up to 56-fold in 750,000 modes with a Hilbert space up to $10^{254}$. Our work exhibits an integrated and cost-efficient shortcut stepping into the "quantum advantage" regime in a photonic system far beyond previous scenarios, and provide a scalable and controllable platform for quantum information processing.

quant-ph↗