arXiv subjects

Xufei Wu

Publications and source records attributed to Xufei Wu.

10 recordsLinked to original sources

arXiv2026

TEngineDB-V: An OLAP-Native Vector Search System for Large-$k$ Workloads at Tencent

Vector search systems are essential infrastructure for modern data-driven applications. Large-$k$ analytical vector search, which retrieves $k=10^3$--$10^5$ results for analytics (e.g., aggregation, filtering, joins), is increasingly important for emerging workloads, including LLM data management and advertising analysis at Tencent. Existing systems remain inadequate: specialized vector databases often cap $k$ (e.g., $k \leq 10^4$) to satisfy tail-latency constraints and offer limited analytical support, while OLAP systems typically embed per-segment vector indexes as black boxes, causing severe read/compute amplification and preventing native query optimization. This paper presents TEngineDB-V, an OLAP-native vector search system for large-$k$ workloads. TEngineDB-V makes vector search a first-class analytical primitive in Tencent's OLAP engine through a global segment-decoupled index materialized as relational tables, eliminating scatter-gather execution, reducing amplification, and enabling native storage optimizations. It decomposes IVFPQ-based search into relational operators, integrates OLAP optimizations, and introduces DPPQ, which combines direction-aware quantization with hierarchical residual refinement to improve recall while preserving relational efficiency. TEngineDB-V further incorporates index-aware query rewriting and a distributed-aware cost model for efficient distributed execution. Experiments show that TEngineDB-V achieves up to a $145\times$ speedup over competitive systems such as StarRocks, and up to a $52\times$ improvement in 10-billion-scale production deployments.

cs.DB↗

arXiv2025

PARROT: A Benchmark for Evaluating LLMs in Cross-System SQL Translation

Large language models (LLMS) have shown increasing effectiveness in Text-to-SQL tasks. However, another closely related problem, Cross-System SQL Translation (a.k.a., SQL-to-SQL), which adapts a query written for one database system (e.g., MySQL) into its equivalent one for another system (e.g., ClickHouse), is of great practical importance but remains underexplored. Existing SQL benchmarks are not well-suited for SQL-to-SQL evaluation, which (1) focus on a limited set of database systems (often just SQLite) and (2) cannot capture many system-specific SQL dialects (e.g., customized functions, data types, and syntax rules). Thus, in this paper, we introduce PARROT, a Practical And Realistic BenchmaRk for CrOss-System SQL Translation. PARROT comprises 598 translation pairs from 38 open-source benchmarks and real-world business services, specifically prepared to challenge system-specific SQL understanding (e.g., LLMS achieve lower than 38.53% accuracy on average). We also provide multiple benchmark variants, including PARROT-Diverse with 28,003 translations (for extensive syntax testing) and PARROT-Simple with 5,306 representative samples (for focused stress testing), covering 22 production-grade database systems. To promote future research, we release a public leaderboard and source code at: https://code4db.github.io/parrot-bench/.

cs.DB↗

arXiv2017

Magnon and Phonon Dispersion, Lifetime and Thermal Conductivity of Iron from Spin-Lattice Dynamics Simulations

In recent years, the fundamental physics of spin-thermal (i.e., magnon-phonon) interaction has attracted significant experimental and theoretical interests given its potential paradigm-shifting impacts in areas like spin-thermoelectrics, spin-caloritronics and spintronics. Modelling studies of the transport of magnons and phonons in magnetic crystals are very rare. In this paper, we use spin-lattice dynamics (SLD) simulations to model ferromagnetic crystalline iron, where the spin and lattice systems are coupled through the atomic position-dependent exchange function, and thus the interaction between magnon and phonon is naturally considered. We then present a method combining SLD simulations with spectral energy analysis to calculate the magnon and phonon harmonic (e.g., dispersion, specific heat, group velocity) and anharmonic (e.g., scattering rate) properties, based on which their thermal conductivity values are calculated. This work represents an example of using SLD simulations to understand the transport properties involving coupled magnon and phonon dynamics.

cond-mat.mtrl-sci↗

arXiv2016

How to Characterize Thermal Transport Capability of 2D Materials Fairly? - Sheet Thermal Conductance and the Choice of Thickness

Ever since the discovery of the record-high thermal conductivity of single layer graphene, thermal transport capability of monolayer 2D materials has been under constant spotlight. Since thermal conductivity is an intensive property for 3D materials and the thickness of 2D materials is not well defined, different definitions of thickness in literature have led to ambiguity towards predicting thermal conductivity values and thus in understanding the heat transfer capability of different monolayer 2D materials. We argue that if conventional definition of thermal conductivity should be used as the quantity to compare the heat transfer capability of various monolayer 2D materials, then the same thickness should be used. Alternatively, to circumvent the problem of ambiguous thickness completely, we also suggest that a "sheet thermal conductance" to be defined as an intensive 2D material property when characterizing the heat transfer capability of 2D materials. When converting literature thermal conductivity values of monolayer materials to this new property, some new features that were not displayed when using different thicknesses show up.

cond-mat.mtrl-sci↗

arXiv2016

Hydrogenation of Penta-Graphene Leads to Unexpected Large Improvement in Thermal Conductivity

Penta-graphene (PG) has been identified as a novel 2D material with an intrinsic bandgap, which makes it especially promising for electronics applications. In this work, we use first-principles lattice dynamics and iterative solution of the phonon Boltzmann transport equation (BTE) to determine the thermal conductivity of PG and its more stable derivative - hydrogenated penta-graphene (HPG). As a comparison, we also studied the effect of hydrogenation on graphene thermal conductivity. In contrast to hydrogenation of graphene, which leads to a dramatic decrease in thermal conductivity (from 3590 to 1328 W/mK - a 63% reduction), HPG shows a notable increase in thermal conductivity (615 W/mK), which is 76% higher than that of PG (350 W/mK). The high thermal conductivity of HPG makes it more thermally conductive than most other semi-conducting 2D materials, such as the transition metal chalcogenides. Our detailed analyses show that the primary reason for the counter-intuitive hydrogenation-induced thermal conductivity enhancement is the weaker bond anharmonicity in HPG than PG. This leads to weaker phonon scattering after hydrogenation, despite the increase in the phonon scattering phase space. The high thermal conductivity of HPG may inspire intensive research around HPG and other derivatives of PG as potential materials for future nanoelectronic devices. The fundamental physics understood from this study may open up a new strategy to engineer thermal transport properties of other 2D materials by controlling bond anharmonicity via functionalization.

cond-mat.mtrl-sci↗

arXiv2015

Unusual Isotope Effect on Thermal Transport of Single Layer Molybdenum Disulphide (MoS2)

Thermal transport in single layer molybdenum disulfide (MoS2) is critical to advancing its applications. In this paper, we use molecular dynamics (MD) simulations with first-principles force constants to study the isotope effect on the thermal transport of single layer MoS2. Through phonon modal analysis, we found that isotopes can strongly scatter phonons with intermediate frequencies, and the scattering behavior can be radically different from that predicted by conventional scattering model based on perturbation theory (Tamura's formula). Such a discrepancy becomes smaller for low isotope concentrations. Natural isotopes can lead to a 30% reduction in thermal conductivity for large size samples. However, for small samples where boundary scattering becomes significant, the isotope effect can be greatly suppressed. It was also found that the Mo isotopes, which contribute more to the phonon eigenvectors in the intermediate frequency range, have stronger impact on thermal conductivity than S isotopes.

cond-mat.mtrl-sci↗

arXiv2015

Effect of Electron-Phonon Coupling on Thermal Transport across Metal-Nonmetal Interface - A Second Look

The effect of electron-phonon (e-ph) coupling on thermal transport across metal-nonmetal interfaces is yet to be completely understood. In this paper, we use a series of molecular dynamics (MD) simulations with e-ph coupling effect included by Langevin dynamics to calculate the thermal conductance at a model metal-nonmetal interface. It is found that while e-ph coupling can present additional thermal resistance on top of the phonon-phonon thermal resistance, it can also make the phonon-phonon thermal conductance larger than the pure phonon transport case. This is because the e-ph interaction can disturb the phonon subsystem and enhance the energy communication between different phonon modes inside the metal. This facilitates redistributing phonon energy into modes that can more easily transfer energy across the interfaces. Compared to the pure phonon thermal conduction, the total thermal conductance with e-ph coupling effect can become either smaller or larger depending on the coupling factor. This result helps clarify the role of e-ph coupling in thermal transport across metal-nonmetal interface.

cond-mat.mtrl-sci↗

arXiv2015

Five-Partite Entanglement Generation in A High-Q Microresonator

We propose to produce five-partite entanglement via cascaded four-wave mixing in a high-Q microresonator that may become a key to future one-way quantum computation on chip. A theoretical model is presented for the underlying continuous-variable entanglement among the generated comb modes that is expansible to more complicated scenarios. We analyze the entanglement condition when the van Loock and Furusawa criteria are violated, and discuss the device parameters for potential experimental realization that may be utilized to build an integrated compact five-partite entanglement generator. The proposed approach exhibits great potential for future large-scale integrated full optical quantum computation on chip.

quant-ph↗

arXiv2014

Polymer Nanofibers with Outstanding Thermal Conductivity and Thermal Stability: Fundamental Linkage between Molecular Characteristics and Macroscopic Thermal Properties

Polymer nanofibers with high thermal conductivities and outstanding thermal stabilities are highly desirable in heat transfer-critical applications such as thermal management, heat exchangers and energy storage. In this work, we unlock the fundamental relations between the thermal conductivity and thermal stability of polymer nanofibers and their molecular characteristics by studying the temperature-induced phase transitions and thermal transport of a series of polymer nanofibers. Ten different polymer nanofibers with systematically chosen molecular structures are studied using large scale molecular dynamics simulations. We found that high thermal conductivity and good thermal stability can be achieved in polymers with rigid backbones, exemplified by π-conjugated polymers, due to suppressed segmental rotations and large phonon group velocities. The low probability of segmental rotation does not only prevent temperature-induced phase transition but also enables long phonon mean free paths due to reduced disorder scattering. Although stronger inter-chain interactions can also improve the thermal stability, polymers with such a feature usually have heavier atoms, weaker backbone bonds, and segments vulnerable to random rotations, which lead to low thermal conductivities. This work elucidates the underlying linkage between the molecular nature and macroscopic thermal properties of polymer nanofibers, which is instrumental to the design of thermally conductive polymer nanofibers with high temperature stabilities.

physics.chem-ph↗

arXiv2011

Bright six-partite continuous variable entanglement using cascaded four-wave mixing processes in a four-level atomic system

We theoretically show that bright six-partite continuous-variable entanglement can be generated using cascaded four-wave mixing effects of third-order nonlinearity atomic systems above threshold. The six-partite continuous-variable entanglement among the six cavity fields with different frequencies is analyzed by applying optimized inseparability criteria proposed by Van Loock and Furusawa.

quant-ph↗