arXiv subjects

Qi Zhou

Publications and source records attributed to Qi Zhou.

At least 109 records · Page 6Linked to original sources

arXiv2025

InfiR : Crafting Effective Small Language Models and Multimodal Small Language Models in Reasoning

Large Language Models (LLMs) and Multimodal Large Language Models (MLLMs) have made significant advancements in reasoning capabilities. However, they still face challenges such as high computational demands and privacy concerns. This paper focuses on developing efficient Small Language Models (SLMs) and Multimodal Small Language Models (MSLMs) that retain competitive reasoning abilities. We introduce a novel training pipeline that enhances reasoning capabilities and facilitates deployment on edge devices, achieving state-of-the-art performance while minimizing development costs. \InfR~ aims to advance AI systems by improving reasoning, reducing adoption barriers, and addressing privacy concerns through smaller model sizes. Resources are available at https://github. com/Reallm-Labs/InfiR.

cs.CL↗

arXiv2025

Exact new mobility edges

Mobility edges (ME), defined as critical energies that separate the extended states from the localized states, are a significant topic in quantum physics. In this paper, we demonstrate the existence of two exact new mobility edges for two physically realistic models: the first, referred to as Type II ME, represents the critical energy that separates the critical states from localized states; the second, referred to as Type III ME, marks the critical energy that separate the critical states from extended states. The proof is based on spectral analysis of singular Jacobi operator on the strip.

math.DS↗

arXiv2025

Phase Transitions in Quasi-Periodically Driven Quantum Critical Systems: Analytical Results

In this work, we study analytically the phase transitions in quasi-periodically driven one dimensional quantum critical systems that are described by conformal field theories (CFTs). The phase diagrams and phase transitions can be analytically obtained by using Avila's global theory in one-frequency quasiperiodic cocycles. Compared to the previous works where the quasiperiodicity was introduced in the driving time and no phase transitions were observed [1], here we propose a setup where the quasiperiodicity is introduced in the driving Hamiltonians. In our setup, one can observe the heating phases, non-heating phases, and the phase transitions. The phase diagram as well as the Lyapunov exponents that determine the entanglement entropy evolution can be analytically obtained. In addition, based on Avila's theory, we prove there is no phase transition in the previously proposed setup of quasi-periodically driven CFTs [1]. We verify our field theory results by studying the time evolution of entanglement entropy on lattice models.

cond-mat.stat-mech↗

arXiv2025

Error estimate of the u-series method for molecular dynamics simulations

This paper provides an error estimate for the u-series method of the Coulomb interaction in molecular dynamics simulations. We show that the number of truncated Gaussians $M$ in the u-series and the base of interpolation nodes $b$ in the bilateral serial approximation are two key parameters for the algorithm accuracy, and that the errors converge as $\mathcal{O}(b^{-M})$ for the energy and $\mathcal{O}(b^{-3M})$ for the force. Error bounds due to numerical quadrature and cutoff in both the electrostatic energy and forces are obtained. Closed-form formulae are also provided, which are useful in the parameter setup for simulations under a given accuracy. The results are verified by analyzing the errors of two practical systems.

math.NA↗

arXiv2024

A fast spectral sum-of-Gaussians method for electrostatic summation in quasi-2D systems

The quasi-2D electrostatic systems, characterized by periodicity in two dimensions with a free third dimension, have garnered significant interest in many fields. We apply the sum-of-Gaussians (SOG) approximation to the Laplace kernel, dividing the interactions into near-field, mid-range, and long-range components. The near-field component, singular but compactly supported in a local domain, is directly calculated. The mid-range component is managed using a procedure similar to nonuniform fast Fourier transforms in three dimensions. The long-range component, which includes Gaussians of large variance, is treated with polynomial interpolation/anterpolation in the free dimension and Fourier spectral solver in the other two dimensions on proxy points. Unlike the fast Ewald summation, which requires extensive zero padding in the case of high aspect ratios, the separability of Gaussians allows us to handle such case without any zero padding in the free direction. Furthermore, while NUFFTs typically rely on certain upsampling in each dimension, and the truncated kernel method introduces an additional factor of upsampling due to kernel oscillation, our scheme eliminates the need for upsampling in any direction due to the smoothness of Gaussians, significantly reducing computational cost for large-scale problems. Finally, whereas all periodic fast multipole methods require dividing the periodic tiling into a smooth far part and a near part containing its nearest neighboring cells, our scheme operates directly on the fundamental cell, resulting in better performance with simpler implementation. We provide a rigorous error analysis showing that upsampling is not required in NUFFT-like steps, achieving $O(N\log N)$ complexity with a small prefactor. The performance of the scheme is demonstrated via extensive numerical experiments.

math.NA↗

arXiv2024

Irrational-window-filter projection method and application to quasiperiodic Schrödinger eigenproblems

In this paper, we propose a new algorithm, the irrational-window-filter projection method (IWFPM), for quasiperiodic systems with concentrated spectral point distribution. Based on the projection method (PM), IWFPM filters out dominant spectral points by defining an irrational window and uses a corresponding index-shift transform to make the FFT available. The error analysis on the function approximation level is also given. We apply IWFPM to 1D, 2D, and 3D quasiperiodic Schrödinger eigenproblems (QSEs) to demonstrate its accuracy and efficiency. IWFPM exhibits a significant computational advantage over PM for both extended and localized quantum states. More importantly, by using IWFPM, the existence of Anderson localization in 2D and 3D QSEs is numerically verified.

math.NA↗

arXiv2024

Mint: Cost-Efficient Tracing with All Requests Collection via Commonality and Variability Analysis

Distributed traces contain valuable information but are often massive in volume, posing a core challenge in tracing framework design: balancing the tradeoff between preserving essential trace information and reducing trace volume. To address this tradeoff, previous approaches typically used a '1 or 0' sampling strategy: retaining sampled traces while completely discarding unsampled ones. However, based on an empirical study on real-world production traces, we discover that the '1 or 0' strategy actually fails to effectively balance this tradeoff. To achieve a more balanced outcome, we shift the strategy from the '1 or 0' paradigm to the 'commonality + variability' paradigm. The core of 'commonality + variability' paradigm is to first parse traces into common patterns and variable parameters, then aggregate the patterns and filter the parameters. We propose a cost-efficient tracing framework, Mint, which implements the 'commonality + variability' paradigm on the agent side to enable all requests capturing. Our experiments show that Mint can capture all traces and retain more trace information while optimizing trace storage (reduced to an average of 2.7%) and network overhead (reduced to an average of 4.2%). Moreover, experiments also demonstrate that Mint is lightweight enough for production use.

cs.SE↗

arXiv2024

Variance-reduced random batch Langevin dynamics

The random batch method is advantageous in accelerating force calculations in particle simulations, but it poses a challenge of removing the artificial heating effect in application to the Langevin dynamics. We develop an approach to solve this issue by estimating the force variance, resulting in a variance-reduced random batch Langevin dynamics. Theoretical analysis shows the high-order local truncation error of the time step in the numerical discretization scheme, in consistent with the fluctuation-dissipation theorem. Numerical results indicate that the method can achieve a significant variance reduction since a smaller batch size provides accurate approximation, demonstrating the attractive feature of the variance-reduced random batch method for Langevin dynamics.

physics.comp-ph↗

arXiv2024

$\mathcal{R}^2$ curvature-squared corrections on Langevin diffusion coefficients

The effect of finite coupling corrections to the Langevin diffusion coefficients on a moving heavy quark in the Super Yang-Mills plasma is investigated. These corrections are related to curvature squared corrections in the corresponding gravity sector. We compare the results of both longitudinal and perpendicular Langevin diffusion coefficients with those in $\mathcal{N}$=4 Super Yang-Mills plasma. It is observed that the curvature-squared corrections influence the Langevin diffusion coefficients, and the corrections for both Langevin diffusion coefficients demonstrate the dependence on the velocity of the moving heavy quark and the specifics of the higher derivative correction. In addition, we conduct calculations for the Langevin diffusion coefficients of a moving heavy quark within the Gauss-Bonnet background.

hep-ph↗

arXiv2024

Unconstrained Model Merging for Enhanced LLM Reasoning

Recent advancements in building domain-specific large language models (LLMs) have shown remarkable success, especially in tasks requiring reasoning abilities like logical inference over complex relationships and multi-step problem solving. However, creating a powerful all-in-one LLM remains challenging due to the need for proprietary data and vast computational resources. As a resource-friendly alternative, we explore the potential of merging multiple expert models into a single LLM. Existing studies on model merging mainly focus on generalist LLMs instead of domain experts, or the LLMs under the same architecture and size. In this work, we propose an unconstrained model merging framework that accommodates both homogeneous and heterogeneous model architectures with a focus on reasoning tasks. A fine-grained layer-wise weight merging strategy is designed for homogeneous models merging, while heterogeneous model merging is built upon the probabilistic distribution knowledge derived from instruction-response fine-tuning data. Across 7 benchmarks and 9 reasoning-optimized LLMs, we reveal key findings that combinatorial reasoning emerges from merging which surpasses simple additive effects. We propose that unconstrained model merging could serve as a foundation for decentralized LLMs, marking a notable progression from the existing centralized LLM framework. This evolution could enhance wider participation and stimulate additional advancement in the field of artificial intelligence, effectively addressing the constraints posed by centralized models.

cs.CL↗

arXiv2024

On an impulsive faecal-oral model in a moving infected environment

This paper develops an impulsive faecal-oral model with free boundary to in order to understand how the exposure to a periodic disinfection and expansion of the infected region together influences the spread of faecal-oral diseases. We first check that this impulsive model has a unique globally nonnegative classical solution. The principal eigenvalues of the corresponding periodic eigenvalue problem at the initial position and infinity are defined as $λ^{\vartriangle}_{1}(h_{0})$ and $λ^{\vartriangle}_{1}(\infty)$, respectively. They both depend on the impulse intensity $1-G'(0)$ and expansion capacities $μ_{1}$ and $μ_{2}$. The possible long time dynamical behaviours of the model are next explored in terms of $λ^{\vartriangle}_{1}(h_{0})$ and $λ^{\vartriangle}_{1}(\infty)$: if $λ^{\vartriangle}_{1}(\infty)\geq 0$, then the diseases are vanishing; if $λ^{\vartriangle}_{1}(\infty)<0$ and $λ^{\vartriangle}_{1}(h_{0})\leq 0$, then the disease are spreading; if $λ^{\vartriangle}_{1}(\infty)<0$ and $λ^{\vartriangle}_{1}(h_{0})> 0$, then for any given $μ_{1}$, there exists a $μ_{0}$ such that spreading happens as $μ_{2}\in( μ_{0},+\infty)$, and vanishing happens as $μ_{2}\in(0, μ_{0})$. Finally, numerical examples are presented to corroborate the correctness of the obtained theoretical findings and to further understand the influence of an impulsive intervention and expansion capacity on the spreading of the diseases. Our results show that both the increase of impulse intensity and the decrease of expansion capacity have a positive contribution to the prevention and control of the diseases.

math.AP↗

arXiv2024

Recovering dark states by non-Hermiticity

Dark states, which are incapable of absorbing and emitting light, have been widely applied in multiple disciplines of physics. However, the existence of dark states relies on certain strict constraints on the system. For instance, in the fundamental Λ system, a perturbation breaking the degeneracy between two energy levels may destroy the destructive interference and demolish the dark state. Here, we show that non-Hermiticity can be exploited as a constructive means to restore a dark state. By compensating for the undesired perturbations, non-Hermiticity produces unidirectional couplings such that the dark state remains decoupled from the rest of the system. Implementing this scheme in many-body systems, flat bands and edge states can be recovered by losses and gains. Further taking into account interactions, a range of novel quantum phases could arise in such non-Hermitian systems.

quant-ph↗

arXiv2024

LogParser-LLM: Advancing Efficient Log Parsing with Large Language Models

Logs are ubiquitous digital footprints, playing an indispensable role in system diagnostics, security analysis, and performance optimization. The extraction of actionable insights from logs is critically dependent on the log parsing process, which converts raw logs into structured formats for downstream analysis. Yet, the complexities of contemporary systems and the dynamic nature of logs pose significant challenges to existing automatic parsing techniques. The emergence of Large Language Models (LLM) offers new horizons. With their expansive knowledge and contextual prowess, LLMs have been transformative across diverse applications. Building on this, we introduce LogParser-LLM, a novel log parser integrated with LLM capabilities. This union seamlessly blends semantic insights with statistical nuances, obviating the need for hyper-parameter tuning and labeled training data, while ensuring rapid adaptability through online parsing. Further deepening our exploration, we address the intricate challenge of parsing granularity, proposing a new metric and integrating human interactions to allow users to calibrate granularity to their specific needs. Our method's efficacy is empirically demonstrated through evaluations on the Loghub-2k and the large-scale LogPub benchmark. In evaluations on the LogPub benchmark, involving an average of 3.6 million logs per dataset across 14 datasets, our LogParser-LLM requires only 272.5 LLM invocations on average, achieving a 90.6% F1 score for grouping accuracy and an 81.1% for parsing accuracy. These results demonstrate the method's high efficiency and accuracy, outperforming current state-of-the-art log parsers, including pattern-based, neural network-based, and existing LLM-enhanced approaches.

cs.SE↗

arXiv2024

RBMD: A molecular dynamics package enabling to simulate 10 million all-atom particles in a single graphics processing unit

This paper introduces a random-batch molecular dynamics (RBMD) package for fast simulations of particle systems at the nano/micro scale. Different from existing packages, the RBMD uses random batch methods for nonbonded interactions of particle systems. The long-range part of Coulomb interactions is calculated in Fourier space by the random batch Ewald algorithm, which achieves linear complexity and superscalability, surpassing classical lattice-based Ewald methods. For the short-range part, the random batch list algorithm is used to construct neighbor lists, significantly reducing both computational and memory costs. The RBMD is implemented on GPU-CPU heterogeneous architectures, with classical force fields for all-atom systems. Benchmark systems are used to validate accuracy and performance of the package. Comparison with the particle-particle particle-mesh method and the Verlet list method in the LAMMPS package is performed on three different NVIDIA GPUs, demonstrating high efficiency of the RBMD on heterogeneous architectures. Our results also show that the RBMD enables simulations on a single GPU with a CPU core up to 10 million particles. Typically, for systems of one million particles, the RBMD allows simulating all-atom systems with a high efficiency of 8.20 ms per step, demonstrating the attractive feature for running large-scale simulations of practical applications on a desktop machine.

physics.comp-ph↗

arXiv2024

HiMA: Hierarchical Quantum Microarchitecture for Qubit-Scaling and Quantum Process-Level Parallelism

Quantum computing holds immense potential for addressing a myriad of intricate challenges, which is significantly amplified when scaled to thousands of qubits. However, a major challenge lies in developing an efficient and scalable quantum control system. To address this, we propose a novel Hierarchical MicroArchitecture (HiMA) designed to facilitate qubit scaling and exploit quantum process-level parallelism. This microarchitecture is based on three core elements: (i) discrete qubit-level drive and readout, (ii) a process-based hierarchical trigger mechanism, and (iii) multiprocessing with a staggered triggering technique to enable efficient quantum process-level parallelism. We implement HiMA as a control system for a 72-qubit tunable superconducting quantum processing unit, serving a public quantum cloud computing platform, which is capable of expanding to 6144 qubits through three-layer cascading. In our benchmarking tests, HiMA achieves up to a 4.89x speedup under a 5-process parallel configuration. Consequently, to the best of our knowledge, we have achieved the highest CLOPS (Circuit Layer Operations Per Second), reaching up to 43,680, across all publicly available platforms.

cs.AR↗

arXiv2024

Hidden curved spaces in Bosonic Kitaev model

Quantum matter in curved spaces exhibits remarkable properties unattainable in flat spaces. To access curved spaces in laboratories, the conventional wisdom is that physical distortions need to be implemented into a system. In contrast to this belief, here, we show that two hyperbolic surfaces readily exist in bosonic Kitaev model in the absence of any physical distortions and give rise to a range of intriguing phenomena, such as chiral quantum transport or chiral reaction-diffusion. A finite chemical potential couples these two hyperbolic surfaces, delivering a quantum sensor whose sensitivity grows exponentially with the size of the system. Our results provide experimentalists with an unprecedented opportunity to explore intriguing quantum phenomena in curve spaces without distortion or access non-Hermitian phenomena without dissipation. Our work also suggests a new class of quantum sensors in which geometry amplifies small signals.

quant-ph↗

arXiv2024

Evolutionary Trigger Detection and Lightweight Model Repair Based Backdoor Defense

Deep Neural Networks (DNNs) have been widely used in many areas such as autonomous driving and face recognition. However, DNN model is fragile to backdoor attack. A backdoor in the DNN model can be activated by a poisoned input with trigger and leads to wrong prediction, which causes serious security issues in applications. It is challenging for current defenses to eliminate the backdoor effectively with limited computing resources, especially when the sizes and numbers of the triggers are variable as in the physical world. We propose an efficient backdoor defense based on evolutionary trigger detection and lightweight model repair. In the first phase of our method, CAM-focus Evolutionary Trigger Filter (CETF) is proposed for trigger detection. CETF is an effective sample-preprocessing based method with the evolutionary algorithm, and our experimental results show that CETF not only distinguishes the images with triggers accurately from the clean images, but also can be widely used in practice for its simplicity and stability in different backdoor attack situations. In the second phase of our method, we leverage several lightweight unlearning methods with the trigger detected by CETF for model repair, which also constructively demonstrate the underlying correlation of the backdoor with Batch Normalization layers. Source code will be published after accepted.

cs.CR↗

arXiv2024

Exact local distribution of the absolutely continuous spectral measure

It is well-established that the spectral measure for one-frequency Schrödinger operators with Diophantine frequencies exhibits optimal $1/2$-Hölder continuity within the absolutely continuous spectrum. This study extends these findings by precisely characterizing the local distribution of the spectral measure for dense small potentials, including a notable result for any subcritical almost Mathieu operators. Additionally, we investigate the stratified Hölder continuity of the spectral measure at subcritical energies.

math-ph↗