arXiv subjects

Zihang Wang

Publications and source records attributed to Zihang Wang.

31 records · Page 2Linked to original sources

arXiv2025

LVD-GS: Gaussian Splatting SLAM for Dynamic Scenes via Hierarchical Explicit-Implicit Representation Collaboration Rendering

3D Gaussian Splatting SLAM has emerged as a widely used technique for high-fidelity mapping in spatial intelligence. However, existing methods often rely on a single representation scheme, which limits their performance in large-scale dynamic outdoor scenes and leads to cumulative pose errors and scale ambiguity. To address these challenges, we propose \textbf{LVD-GS}, a novel LiDAR-Visual 3D Gaussian Splatting SLAM system. Motivated by the human chain-of-thought process for information seeking, we introduce a hierarchical collaborative representation module that facilitates mutual reinforcement for mapping optimization, effectively mitigating scale drift and enhancing reconstruction robustness. Furthermore, to effectively eliminate the influence of dynamic objects, we propose a joint dynamic modeling module that generates fine-grained dynamic masks by fusing open-world segmentation with implicit residual constraints, guided by uncertainty estimates from DINO-Depth features. Extensive evaluations on KITTI, nuScenes, and self-collected datasets demonstrate that our approach achieves state-of-the-art performance compared to existing methods.

cs.CV↗

arXiv2025

Non-Equilibrium Origin of Native Ring Anisotropy in Amorphous Systems

Native ring structures within amorphous networks play a critical role in determining structural and optical properties, in part due to their ability to host dopants such as rare earth ions in silicate systems. In this work, we demonstrate that the universal features of structural anisotropy in amorphous networks can be efficiently simulated using a model based on stochastically deformed, edge sharing N member native ring structures. This model isolates and characterizes the structural anisotropy generated during the annealing quenching process that is independent of any constituent specific interactions. We refer to this computational framework as Indistinguishable Simulated Folding (ISF), a stochastic process that mimics a simulated annealing quenching procedure. Formulated as a Markov process, ISF is governed by two physically meaningful parameters: the number of Markov steps, representing the mean duration of each ring folding event, and the stochastic deformation magnitude, which quantifies thermally induced structural changes per event. Furthermore, we show that the logarithm of any positive valued anisotropy measure generated by ISF is a skewed random variable, reflecting the growing entropy production rate during the Markov evolution. ISF provides both a conceptual framework for understanding the universal stochastic origin of structural anisotropy in amorphous networks and a practical tool for simulating constituent independent features, without requiring full scale molecular dynamics simulations.

cond-mat.dis-nn↗

arXiv2025

Correspondence between quasiparticle dissipation and quantum information decay in open quantum systems

Diagrammatic techniques simplify a weakly interacting many-body problem into an effective few-quasiparticle problem within a system of interest (SOI). If scattering events, mediated by a bath, between those quasiparticles can be approximated as density-density interactions, the bath behaves like an effective external potential. On the other hand, exchange interactions could entangle those quasiparticles and the bath, leading to an open quantum system that induces quantum decoherence and spectral broadening. We investigate the renormalized interaction between the SOI and the bath, employing a projection operator technique similar to the one used in the Nakajima-Zwanzig method. We find that the frequency variation of this renormalized interaction is analogous to the quasiparticle residue and provides a measure of the SOI-bath separability that serves as the lower bound of the SOI-bath entanglement entropy. In the weak-coupling regime and continuum limit, we demonstrate that the degree of SOI-bath separability corresponds to the quasiparticle spectral weight in the single-impurity Anderson model and find that the loss of quantum information to the continuum of the bath can be understood as a decay process where an initial single-impurity state escapes to a thermal bath. This work provides a direction for connecting energy dissipation in quasiparticles propagation to the loss of quantum information in open quantum systems.

quant-ph↗

arXiv2025

LLM-based Automated Theorem Proving Hinges on Scalable Synthetic Data Generation

Recent advancements in large language models (LLMs) have sparked considerable interest in automated theorem proving and a prominent line of research integrates stepwise LLM-based provers into tree search. In this paper, we introduce a novel proof-state exploration approach for training data synthesis, designed to produce diverse tactics across a wide range of intermediate proof states, thereby facilitating effective one-shot fine-tuning of LLM as the policy model. We also propose an adaptive beam size strategy, which effectively takes advantage of our data synthesis method and achieves a trade-off between exploration and exploitation during tree search. Evaluations on the MiniF2F and ProofNet benchmarks demonstrate that our method outperforms strong baselines under the stringent Pass@1 metric, attaining an average pass rate of $60.74\%$ on MiniF2F and $21.18\%$ on ProofNet. These results underscore the impact of large-scale synthetic data in advancing automated theorem proving.

cs.AI↗

arXiv2025

Wide Binary Evaporation by Dark Solitons: Implications from the GAIA Catalog

An analytic calculation is given for binary star evaporation under the tidal perturbation from randomly distributed, spatially extended dark objects. In particular, the Milky Way's wide binary star population is susceptible to such disruption from dark matter solitons of comparable and larger sizes. We identify high-probability `halo-like' wide binaries in GAIA EDR3 with separations larger than 0.1 parsec. Survival of the farthest-separated candidates will provide a novel gravitational probe to dark matter in the form of solitons. In the case of dilute axion-like boson stars, the observational sensitivity extends into the axion mass range $m_a \sim 10^{-17}-10^{-15}$ eV.

hep-ph↗

arXiv2024

Axion Minicluster Halo Limits from Wide Binary Disruption

Axionic dark matter can form miniclusters and minicluster halos from inhomogenuities in the early Universe. If MCHs are sufficiently massive, their existence can be revealed by small-scale gravitational tidal perturbation to halo-like binary star system in the Galaxy. The observed population of the Milky Way's wide-separation binaries with $a\gtrsim\mathcal{O}(0.1)\,$ parsec offer a sensitive test to dynamic evaporation from MCHs. Considering data from recent GAIA observations, we derive significant constraints on the MCH fraction of the galactic dark matter halo. For several scenarios including dense MCHs and isolated minicluster models, these limits will apply to axion-like particles in the mass range $m_{a}\sim 10^{-15}-10^{-12}\,\rm eV$ and $m_{a}\sim 10^{-19}-10^{-16}\,\rm eV$, respectively.

hep-ph↗

arXiv2024

A Study on Lampreys Population Based on Sex-Ratio-Related Growth-Balance Model

Lampreys are one of the oldest species in the world, living longer than dinosaurs, which is related to the ability to change the sex ratio during their lifespan. In this paper, to understand how sex ratio and food quantity affect the population growth rate of lampreys, the researchers draw inspiration from the logistics model and established a model called EcoSexChange(ESC), which results in a population initially increasing and then stabilizing, a reasonable outcome that may apply to other organisms with significant differences in consumption between sexes. Subsequently, this paper develops the Sex Ratio Adaptation Eco Impact (SRAEI) model based on the ESC model using the ABM algorithm to simulate how the population of lampreys, whose lives are divided into seven stages, grows and stabilizes. Then introduces a sudden disaster factor in the middle of the simulation, while also comparing lampreys that cannot adjust their sex ratio. The results of this paper are of great reference significance for people to analyze the population changes of lampreys in different living environments, and they are also easy to apply to other species with large differences between males and females.

q-bio.PE↗

arXiv2023

Exploiting Maximally Mixed States for Spectral Estimation by Time Evolution

We introduce a novel approach for estimating the spectrum of quantum many-body Hamiltonians, and more generally, of Hermitian operators, using quantum time evolution. In our approach we are evolving a maximally mixed state under the Hamiltonian of interest and collecting specific time-series measurements to estimate its spectrum. We demonstrate the advantage of our technique over currently used classical statistical sampling methods. We showcase our approach by experimentally estimating the spectral decomposition of a 2-qubit Heisenberg Hamiltonian on an IBM Quantum backend. For this purpose, we develop a hardware-efficient decomposition that controls $n$-qubit Pauli rotations against the physically closest qubit alongside expressing two-qubit rotations in terms of the native entangling interaction. This substantially reduced the accumulation of errors from noisy two-qubit operations in time evolution simulation protocols. We conclude by discussing the potential impact of our work and the future directions of research it opens.

quant-ph↗

arXiv2023

Enhancing data efficiency in reinforcement learning: a novel imagination mechanism based on mesh information propagation

Reinforcement learning(RL) algorithms face the challenge of limited data efficiency, particularly when dealing with high-dimensional state spaces and large-scale problems. Most of RL methods often rely solely on state transition information within the same episode when updating the agent's Critic, which can lead to low data efficiency and sub-optimal training time consumption. Inspired by human-like analogical reasoning abilities, we introduce a novel mesh information propagation mechanism, termed the 'Imagination Mechanism (IM)', designed to significantly enhance the data efficiency of RL algorithms. Specifically, IM enables information generated by a single sample to be effectively broadcasted to different states across episodes, instead of simply transmitting in the same episode. This capability enhances the model's comprehension of state interdependencies and facilitates more efficient learning of limited sample information. To promote versatility, we extend the IM to function as a plug-and-play module that can be seamlessly and fluidly integrated into other widely adopted RL algorithms. Our experiments demonstrate that IM consistently boosts four mainstream SOTA RL algorithms, such as SAC, PPO, DDPG, and DQN, by a considerable margin, ultimately leading to superior performance than before across various tasks. For access to our code and data, please visit https://github.com/OuAzusaKou/imagination_mechanism

cs.LG↗

arXiv2023

Patterning of morphogenetic anisotropy fields

Orientational order, encoded in anisotropic fields, plays an important role during the development of an organism. A striking example of this is the freshwater polyp Hydra, where topological defects in the muscle fiber orientation have been shown to localize to key features of the body plan. This body plan is organized by morphogen concentration gradients, raising the question how muscle fiber orientation, morphogen gradients and body shape interact. Here, we introduce a minimal model that couples nematic orientational order to the gradient of a morphogen field. We show that on a planar surface alignment to a radial concentration gradient can induce unbinding of topological defects, as observed during budding and tentacle formation in Hydra, and stabilize aster/vortex-like defects, as observed at a Hydra's mouth. On curved surfaces mimicking the morphologies of Hydra in various stages of development -- from spheroid to adult -- our model reproduces the experimentally observed reorganization of orientational order. Our results suggest how gradient alignment and curvature effects may work together to control orientational order during development and lays the foundations for future modeling efforts that will include the tissue mechanics that drive shape deformations.

physics.bio-ph↗

arXiv2022

Dark radiation as a probe for phase transition in the early universe

The cosmological constant is not necessarily small in the early universe. If a scalar field obtains a vacuum expectation value after a phase transition (PT), a possibly large cosmological constant could present before PT. The early cosmological constant (ECC) and the PT process may be detectable from dark radiation (DR) today, such as in the cosmic axion background. We show that for a broad class of DR models, the DR density and spectrum are significantly modified by the presence of an ECC. From the density and the spectrum of the DR today, we can deduce the temperature and the strength of the PT.

hep-ph↗

arXiv2021

Axion induced spin effective couplings

Detecting axionic dark matter induced electron or nucleon oscillating electric dipole moment (OEDM) has become a new way for dark matter searches. We re-examine such axion-spin couplings in external electromagnetic fields. We point out that axion-photon interaction induces an electron spin effective coupling, which is different from an OEDM. In particular, the axion-spin effective coupling is directly related to magnetic field rather than electric field. For axion-electron or axion-nucleon couplings, an OEDM of fermion is introduced, whose effect in ultralight axion cases depends on whether axion shift symmetry is manifest. Specifically, ultralight axionic dark matter interactions that do not obey the shift symmetry will be strongly constrained. We also extend the results to the case where axion has a finite velocity.

hep-ph↗

arXiv2020

Fast Intent Classification for Spoken Language Understanding

Spoken Language Understanding (SLU) systems consist of several machine learning components operating together (e.g. intent classification, named entity recognition and resolution). Deep learning models have obtained state of the art results on several of these tasks, largely attributed to their better modeling capacity. However, an increase in modeling capacity comes with added costs of higher latency and energy usage, particularly when operating on low complexity devices. To address the latency and computational complexity issues, we explore a BranchyNet scheme on an intent classification scheme within SLU systems. The BranchyNet scheme when applied to a high complexity model, adds exit points at various stages in the model allowing early decision making for a set of queries to the SLU model. We conduct experiments on the Facebook Semantic Parsing dataset with two candidate model architectures for intent classification. Our experiments show that the BranchyNet scheme provides gains in terms of computational complexity without compromising model accuracy. We also conduct analytical studies regarding the improvements in the computational cost, distribution of utterances that egress from various exit points and the impact of adding more complexity to models with the BranchyNet scheme.

cs.CL↗