Search arXiv⌕ Search

arXiv subjects

Da Xu

Publications and source records attributed to Da Xu.

At least 19 recordsLinked to original sources
arXiv2026

TransX: Scaling Transformer-based Recommendation via Behavioral and Serving Stream Crossings

Modern industrial recommender systems (RecSys) increasingly adopt Transformer-based sequence models, with an emerging paradigm that frames recommendation as next-token prediction over a unified monolithic user sequence. However, collapsing heterogeneous data sources -- such as long-term user behaviors and real-time serving events -- into a single monolithic token stream that obscures their distinct causal roles and temporal characteristics, leading to inefficient modeling and elevated training and serving costs. We propose TransX, a production-oriented encoder-decoder architecture that reformulates recommendation as a sequence-to-sequence action transduction problem. TransX explicitly decouples behavior-stream modeling from serving-event modeling and conditions next-action decoding on scalable cross-attention between nearline behavior encodings and real-time serving representations. To enable low-latency, high-QPS deployment, TransX is co-designed with an amortized serving strategy that combines incremental behavior encoding with per-request key-value caching, rendering serving latency insensitive to behavior sequence length. Extensive offline experiments and large-scale online A/B tests on LinkedIn's recommender systems show that TransX consistently outperforms state-of-the-art DLRMs and sequential baselines, and delivers substantial CTR lift (+6.0%) and conversion gain (+4.4%) while maintaining serving costs comparable to existing production models where our co-designed serving strategy reduces online computation by approximately 80%.

cs.IR↗
arXiv2026

Magnon squeezing in the quantum regime

Squeezed states, crucial for quantum metrology and emerging quantum technologies, have been demonstrated in various platforms, but quantum squeezing of magnons in macroscopic spin systems remains elusive. Here we report the experimental observation of quantum-level magnon squeezing in a millimeter-scale yttrium iron garnet (YIG) sphere. By engineering a strong dispersive magnon-superconducting qubit coupling via a microwave cavity, we implement a significant self-Kerr nonlinearity to generate squeezed magnon states with their mean magnon number less than one. Harnessing a magnon-assisted Raman process, we perform Wigner tomography, revealing quadrature variances of $\sim\!0.8$ ($\sim\!1.0$~dB squeezing) relative to the vacuum. These results lay the groundwork for quantum nonlinear magnonics and promise potential applications in quantum metrology.

quant-ph↗
arXiv2025

The Angular Momentum Penrose Inequality

We prove the Penrose inequality with angular momentum for asymptotically flat, axisymmetric vacuum initial data sets containing a stable marginally outer trapped surface. This inequality provides a lower bound for the ADM mass in terms of the area and angular momentum of the black hole horizon, with equality holding if and only if the initial data set corresponds to a slice of the Kerr spacetime. Our proof combines the Jang equation approach with the p-harmonic level set method. A key component of the analysis is a modified Hawking mass functional that incorporates angular momentum and exhibits monotonicity along the flow. We also establish the rigidity of the inequality using the Mars-Simon tensor.

gr-qc↗
arXiv2025

The Spacetime Penrose Inequality: Conditional Results for Stable MOTS and General Trapped Surfaces

We present a rigorous proof of the Spacetime Penrose Inequality relating the ADM mass to the area of trapped surfaces in asymptotically flat initial data sets satisfying the dominant energy condition. The main theorem establishes that the ADM mass is bounded below by the square root of the area divided by 16 pi for an area-maximizing marginally outer trapped surface (MOTS), subject to a distributional favorable jump condition which we prove is structurally guaranteed by KKT optimality. The extension to the outermost MOTS remains conditional on the hypothesis that the area maximizer coincides with the outermost MOTS, or equivalently on Weak Cosmic Censorship. We explicitly flag that without this condition, the proof for general trapped surfaces does not go through, as evidenced by binary merger counterexamples. We provide a complete double-limit analysis of the Agostiniani-Mazzieri-Oronzio level-set flow on the singular Jang space, resolving regularity and boundary-term obstructions. In the equality case, the initial data embed isometrically into the Schwarzschild spacetime.

gr-qc↗
arXiv2025

Sharp Spectral Zeta Asymptotics on Graphs of Quadratic Growth

We investigate the spectral properties of the Dirichlet Laplacian on large finite metric balls within irregular infinite graphs of quadratic volume growth. We consider an exhaustion $G_n = B_{R_n}(x_0)$ and the spectral zeta value $Z_n(1) = \operatorname{tr}(L_n^{-1})$ of the killed generator $L_n$. We establish a sharp asymptotic law under the assumptions that the graph satisfies uniform quadratic volume growth (VG(2)) and a Poincare inequality (PI). These analytic-geometric hypotheses imply large-scale regularity. Additionally, we assume a standard quantitative homogenisation property: a uniform local central limit theorem with a polynomial convergence rate. This hypothesis holds for our main example classes and implies the existence of a global heat-kernel constant $\mathcal{G} > 0$ (independent of $x$). In particular, the lazy simple random walk (LSRW) satisfies $p_t(x,x) \sim \mathcal{G}/t$ as $t \to \infty$. Our main theorem establishes the sharp asymptotic $Z_n(1) = \mathcal{G},N_n \log N_n + O(N_n)$, where $N_n := |V(G_n)| \to \infty$ as $n \to \infty$. This implies a relative error of $O(1/\log N_n)$, with constants depending only on the structural parameters of $G$. This result extends far beyond homogeneous lattices. For $\mathbb{Z}^2$, this yields the constant identification $\mathcal{G} = 2/\pi$, providing a new limit formula that recovers $\pi$ without $\pi$ appearing in the input (a "pi-free" limit). Our techniques highlight the robustness of spectral asymptotics under homogenisation in this critical, recurrent setting.

math.FA↗
arXiv2025

Skew-symmetric super-biderivations of the special Lie superalgebra S(m, n; t)

This paper aims to study the skew-symmetric super-biderivations of the special Lie superalgebra S(m, n; t). Let S denote the special Lie superalgebra S(m, n; t) over a field of characteristic p > 2.Utilizing the abelian subalgebra TS and the weight space decomposition of S with respect to TS ,we show the action of a skew-symmetric super-biderivation on the elements of TS and some specific elements of S. Moreover, we prove that all skew-symmetric super-biderivations of S are inner.

math.RA↗
arXiv2025

Skew-symmetric super-biderivations of the Hamiltonian superalgebra H(m, n; t)

This paper aims to study the skew-symmetric super-biderivations of the Hamiltonian superalgebra H(m, n; t). Let H denote the Hamiltonian Lie superalgebra H(m, n; t) over a field of characteristic p > 2. Utilizing the abelian subalgebra TH and the weight space decomposition of H with respect to TH , we show the action of a skew-symmetric super-biderivation on the elements of TH and some specific elements of H. Moreover, we prove that all skew-symmetric super-biderivations of H are inner.

math.RA↗
arXiv2025

Skew-symmetric super-biderivations of the special odd Hamiltonian superalgebra SHO(n, n; t)

This paper aims to study the skew-symmetric super-biderivations of the special odd Hamiltonian superalgebra SHO(n, n; t). Let HO denote the odd Hamiltonian Lie superalgebra HO(n, n; t) and SHO the special odd Hamiltonian Lie superalgebra SHO(n, n; t) over a field of characteristic p > 2.Utilizing the abelian subalgebra TSHO and the weight space decomposition of HO with respect to TSHO , we show the action of a skew-symmetric super-biderivation on the elements of TSHO and some specific elements of SHO. Moreover, we prove that all skew-symmetric super-biderivations of SHO are inner.

math.RA↗
arXiv2025

Quantum Entanglement Theory and Its Generic Searches in High Energy Physics

We propose a new formalism for quantum entanglement (QE), and study its generic searches at the colliders. For a general quantum system with $N$ particles, we show that the quantum space (the total spin polarization parameter space) is complex projective space, and the classical space (the spin polarization parameter space for classical theory) is the cartesian product of the complex projective spaces. Thus, the quantum entanglement space is the difference of these two spaces. For the $ff$, $AA$, $Af$, $fff$, and $ffA$ systems, we propose their discriminants $\Delta_i$. The corresponding classical spaces are the discriminant locus $\Delta=0$ for $ff$ system, and intersections of the discriminant loci $\Delta_i=0$ for $AA$, $Af$, $fff$, and $ffA$ systems in the quantum space. In particular, for two fermion $ff$ system, we prove that our discriminant criterion is equivalent to the original Peres-Horodecki criterion and the CHSH criterion. And thus our quantum entanglement space is indeed Bell non-local. With the collider searches, we can reconstruct the discriminants from various measurements, and probe the quantum entanglement spaces via a fundamental approach at exact level. In addition, for the specific approach, we present a comprehensive framework to detect quantum entanglement in high-energy multi-particle systems, spanning fermion pairs ($t\bar{t}$, $\tau^{+}\tau^{-}$), bosonic pairs ($W^{-}W^{+}$), and hybrid or three-body systems ($W^{-}t$, $ttt$, $t\bar{t}W^{-}$), by diverse observables through angular correlations in decay products. These results establish model-independent methodologies for probing QE across collider experiments, bridging quantum information principles with high-energy phenomenology, while offering novel pathways to explore exotic particles and quantum properties in multi-particle systems.

hep-ph↗
arXiv2025

Reinforcement-learning-assisted control of four-roll mills: geometric symmetry and inertial effect

Embedding the intrinsic symmetry of a flow system in training its machine learning algorithms has become a significant trend in the recent surge of their application in fluid mechanics. This paper leverages the geometric symmetry of a four-roll mill (FRM) to enhance its training efficiency. Stabilizing and precisely controlling droplet trajectories in a FRM is challenging due to the unstable nature of the extensional flow with a saddle point. Extending the work of Vona & Lauga, this study applies Deep Reinforcement Learning (DRL) to effectively guide a displaced droplet to the center of the FRM. Through direct numerical simulations, we explore the applicability of DRL in controlling FRM flow with moderate inertial effects, i.e., Reynolds number $\sim\mathcal{O}(1)$, a nonlinear regime previously unexplored. The FRM's geometric symmetry allows control policies trained in one of the eight sub-quadrants to be extended to the entire domain, reducing training costs. Our results indicate that the DRL-based control method can successfully guide a displaced droplet to the target center with robust performance across various starting positions, even from substantially far distances. The work also highlights potential directions for future research, particularly focusing on efficiently addressing the delay effects in flow response caused by inertia. This study presents new advances in controlling droplet trajectories in more nonlinear and complex situations, with potential applications to other nonlinear flows. The geometric symmetry used in this cutting-edge reinforcement learning approach can also be applied to other control methods.

physics.flu-dyn↗
arXiv2025

A Deep Learning Framework for Boundary-Aware Semantic Segmentation

As a fundamental task in computer vision, semantic segmentation is widely applied in fields such as autonomous driving, remote sensing image analysis, and medical image processing. In recent years, Transformer-based segmentation methods have demonstrated strong performance in global feature modeling. However, they still struggle with blurred target boundaries and insufficient recognition of small targets. To address these issues, this study proposes a Mask2Former-based semantic segmentation algorithm incorporating a boundary enhancement feature bridging module (BEFBM). The goal is to improve target boundary accuracy and segmentation consistency. Built upon the Mask2Former framework, this method constructs a boundary-aware feature map and introduces a feature bridging mechanism. This enables effective cross-scale feature fusion, enhancing the model's ability to focus on target boundaries. Experiments on the Cityscapes dataset demonstrate that, compared to mainstream segmentation methods, the proposed approach achieves significant improvements in metrics such as mIOU, mDICE, and mRecall. It also exhibits superior boundary retention in complex scenes. Visual analysis further confirms the model's advantages in fine-grained regions. Future research will focus on optimizing computational efficiency and exploring its potential in other high-precision segmentation tasks.

cs.CV↗
arXiv2025

Search potential for direct slepton pair production at the CEPC with $\sqrt{s}$ = 360 GeV

The Circular Electron Positron Collider (CEPC) is designed to operate at the key center-of-mass energies: 91.2 GeV as a Z factory for precision Z boson studies,$\approx$ 160 GeV at the threshold for W boson pair production, and 240 GeV as a Higgs factory for copious Higgs boson production. It can be upgraded to 360 GeV (CEPC-360GeV) for enabling top quark-antiquark ($t\bar{t}$) pair production. Beyond enabling high-precision measurements of the Standard Model (SM), CEPC-360GeV is uniquely position to perform searches for new physics beyond the SM (BSM) physics, serving as a valuable complement to hadron colliders. This paper presents a sensitivity study on the direct pair production of staus and smuons at the CEPC with $\sqrt{s}$ = 360 GeV, conducted via full Monte Carlo (MC) simulation. Under the assumptions of integrated luminosity 1.0 ab^{-1} and a flat 5% systematic uncertainty, CEPC-360GeV could potentially discover the combined production of left-handed and right-handed staus up to a mass of 170 GeV (if they exist), or up to 169 GeV for pure left-handed staus and 162 GeV for pure right-handed staus. For direct smuon production, the discovery potential reaches up to 178 GeV under the same conditions.

hep-ex↗
arXiv2024

Exploring Nanoscale Photoresponse Mechanisms for Enhanced Photothermoelectric Effects in van der Waals Interfaces

Integrated photodetectors are crucial for their high speed, sensitivity, and efficient power consumption. In these devices, photocurrent generation is primarily attributed to the photovoltaic (PV) effect, driven by electron hole separations, and the photothermoelectric (PTE) effect, which results from temperature gradients via the Seebeck effect. As devices shrink, the overlap of these mechanisms-both dependent on the Fermi level and band structure-complicates their separate evaluation at the nanoscale. This study introduces a novel 3D photocurrent nano-imaging technique specifically designed to distinctly map these mechanisms in a Schottky barrier photodiode featuring a molybdenum disulfide and gold (MoS2 Au) interface. We uncover a significant PTE-dominated region extending several hundred nanometers from the electrode edge, a characteristic facilitated by the weak electrostatic forces typical in 2D materials. Unexpectedly, we find that incorporating hexagonal boron nitride (hBN), known for its high thermal conductivity, markedly enhances the PTE response. This counterintuitive enhancement stems from an optimal overlap between thermal and Seebeck profiles, presenting a new pathway to boost device performance. Our findings highlight the capability of this imaging technique to not only advance optoelectronic applications but also to deepen our understanding of light matter interactions within low-dimensional systems.

cond-mat.mtrl-sci↗
arXiv2024

Survey for Landing Generative AI in Social and E-commerce Recsys -- the Industry Perspectives

Recently, generative AI (GAI), with their emerging capabilities, have presented unique opportunities for augmenting and revolutionizing industrial recommender systems (Recsys). Despite growing research efforts at the intersection of these fields, the integration of GAI into industrial Recsys remains in its infancy, largely due to the intricate nature of modern industrial Recsys infrastructure, operations, and product sophistication. Drawing upon our experiences in successfully integrating GAI into several major social and e-commerce platforms, this survey aims to comprehensively examine the underlying system and AI foundations, solution frameworks, connections to key research advancements, as well as summarize the practical insights and challenges encountered in the endeavor to integrate GAI into industrial Recsys. As pioneering work in this domain, we hope outline the representative developments of relevant fields, shed lights on practical GAI adoptions in the industry, and motivate future research.

cs.IR↗
arXiv2024

Demonstration of 3 V Programmable Josephson Junction Arrays Using Non-Integer-Multiple Logic

This article demonstrates a new kind of programmable logic for the representation of an integer that can be used for the programmable Josephson voltage standard. It can enable the numbers of junctions in most bits to be variable integer values, which is different from normal binary logic or ternary logic. Consequently, missing junctions due to superconducting short circuits can be tolerated under this logic. This logic can also have nearly the same segmentation efficiency as ternary logic. The completeness of the sequences using this logic is proven by the recursive method in mathematics in this paper. After that, a new algorithm for the representation of integers is presented according to the proven process, and an analysis of the number of fault-tolerant junctions for each bit is provided. Although the first and second bits are not tolerant to missing junctions, bits beyond these can tolerate one to hundreds of missing junctions. Due to the non-fixed multiples between the bits of the sequence, this logic is called non-integer-multiple logic. Finally, the design and fabrication of a 3 V programmable Josephson junction array using this logic are described, and the measurements and analysis of the characteristic parameters are presented.

cs.ET↗
arXiv2023

Macroscopic Bell state between a millimeter-sized spin system and a superconducting qubit

Entanglement is a fundamental property in quantum mechanics that systems share inseparable quantum correlation regardless of their mutual distances. Owing to the fundamental significance and versatile applications, the generation of quantum entanglement between {\it macroscopic} systems has been a focus of current research. Here we report on the deterministic generation and tomography of the macroscopically entangled Bell state in a hybrid quantum system containing a millimeter-sized spin system ($\sim 1\times10^{19}$ atoms) and a micrometer-sized superconducting qubit. The deterministic generation is realized by coupling the macroscopic spin system and the qubit via a microwave cavity. Also, we develop a joint tomography approach to confirming the deterministic generation of the Bell state, which gives a generation fidelity of $0.90\pm0.01$. Our work makes the macroscopic spin system the {\it largest} system (in the sense of atom number) capable of generating the maximally entangled quantum state.

quant-ph↗
arXiv2023

Pretrained Embeddings for E-commerce Machine Learning: When it Fails and Why?

The use of pretrained embeddings has become widespread in modern e-commerce machine learning (ML) systems. In practice, however, we have encountered several key issues when using pretrained embedding in a real-world production system, many of which cannot be fully explained by current knowledge. Unfortunately, we find that there is a lack of a thorough understanding of how pre-trained embeddings work, especially their intrinsic properties and interactions with downstream tasks. Consequently, it becomes challenging to make interactive and scalable decisions regarding the use of pre-trained embeddings in practice. Our investigation leads to two significant discoveries about using pretrained embeddings in e-commerce applications. Firstly, we find that the design of the pretraining and downstream models, particularly how they encode and decode information via embedding vectors, can have a profound impact. Secondly, we establish a principled perspective of pre-trained embeddings via the lens of kernel analysis, which can be used to evaluate their predictability, interactively and scalably. These findings help to address the practical challenges we faced and offer valuable guidance for successful adoption of pretrained embeddings in real-world production. Our conclusions are backed by solid theoretical reasoning, benchmark experiments, as well as online testings.

cs.LG↗