arXiv subjects

Guoqiang Wang

Publications and source records attributed to Guoqiang Wang.

At least 19 recordsLinked to original sources

arXiv2026

Atria Dawn: The Dawn of Agentic Superintelligence

As AI agents become participants in the development of their successors, they reshape both the production of intelligence and the role of human researchers. We introduce Atria Dawn Preview, a foundation agentic language model designed for scientific research and engineering workflows, with the goal of expanding the frontier of agent productivity in the real world. This model is trained via a Verifiable Experience Pipeline that connects tool-mediated interactions to executable environments and externally verified outcomes. Across 16 benchmarks spanning real-world research, engineering, and digital work, Atria Dawn Preview is competitive with frontier agents and achieves the highest reported score on five of them. Beyond standalone performance, we examine the real research-and-development process behind this model as a case study of human--AI collaboration, analyzing 769 task records from 56 participants together with agent logs. When asked to evaluate completed tasks under comparable conditions, participants rated about one-third of completed AI-assisted tasks as infeasible without AI. More strikingly, agents frequently propose methods and implement revisions, while humans retain most final decisions and guide exploration through judgment and feedback. These observations indicate a shift from task-level execution to project-level partnership, with human effort concentrating on what is worth pursuing and how evidence should guide research. Progress toward more autonomous AI research must therefore advance both the capacity for discovery and the capacity for meaningful human oversight, preserving accountable human authority over the risks and direction of continued development.

cs.AI↗

arXiv2026

PolycubeNet: A Dual-latent Diffusion Model for Polycube-Based Hexahedral Mesh Generation

Hexahedral meshes are widely used in simulation pipelines, yet automatic generation remains challenging for complex CAD geometries. Polycube-based hexahedral meshing is a representative approach due to its regular, parameterization-friendly structure, but existing polycube construction methods often rely on intricate surface segmentation and local heuristics, which can produce artifacts or fail on difficult shapes. In this paper, we propose an end-to-end framework for polycube generation based on conditional diffusion models. Given an input geometry represented as a point cloud, our method directly produces a corresponding polycube point cloud, eliminating the need for explicit surface segmentation or predefined polycube templates. At the core of our approach is a dual-latent conditional diffusion architecture that confines computationally expensive self-attention operations to a fixed-capacity, low-dimensional latent space. This design effectively decouples computational complexity from the resolution of both the input geometry and the output polycube, thereby avoiding the quadratic cost typical of point cloud self-attention mechanisms while supporting flexible input and output resolutions. To obtain a hexahedral mesh, the generated polycube is aligned to the input shape via rigid and non-rigid point cloud registration to establish surface correspondence, followed by a polycube-to-hex pipeline. We additionally create and release a paired dataset of CAD meshes and their corresponding polycube meshes, together with the core implementation of our model. Experiments show that PolycubeNet generalizes to complex CAD models with arbitrary genus and produces high-quality polycube structures within seconds, improving robustness and efficiency over prior learning-based approaches.

cs.GR↗

arXiv2026

TREX: Automating LLM Fine-tuning via Agent-Driven Tree-based Exploration

While Large Language Models (LLMs) have empowered AI research agents to perform isolated scientific tasks, automating complex, real-world workflows, such as LLM training, remains a significant challenge. In this paper, we introduce TREX, a multi-agent system that automates the entire LLM training life-cycle. By orchestrating collaboration between two core modules-the Researcher and the Executor-the system seamlessly performs requirement analysis, open-domain literature and data research, formulation of training strategies, preparation of data recipes, and model training and evaluation. The multi-round experimental process is modeled as a search tree, enabling the system to efficiently plan exploration paths, reuse historical results, and distill high-level insights from iterative trials. To evaluate the capability of automated LLM training, we construct FT-Bench, a benchmark comprising 10 tasks derived from real-world scenarios, ranging from optimizing fundamental model capabilities to enhancing performance on domain-specific tasks. Experimental results demonstrate that the TREX agent consistently optimizes model performance on target tasks.

cs.AI↗

arXiv2025

Zeeman: A Deep Learning Regional Atmospheric Chemistry Transport Model

Atmospheric chemistry encapsulates the emission of various pollutants, the complex chemistry reactions, and the meteorology dominant transport, which form a dynamic system that governs air quality. While deep learning (DL) models have shown promise in capturing intricate patterns for forecasting individual atmospheric component - such as PM2.5 and ozone - the critical interactions among multiple pollutants and the combined influence of emissions and meteorology are often overlook. This study introduces an advanced DL-based atmospheric chemistry transport model Zeeman for multi-component atmospheric chemistry simulation. Leveraging an attention mechanism, our model effectively captures the nuanced relationships among these constituents. Performance metrics demonstrate that our approach rivals numerical models, offering an efficient solution for atmospheric chemistry. In the future, this model could be further integrated with data assimilation techniques to facilitate efficient and accurate atmospheric emission estimation and concentration forecast.

physics.ao-ph↗

arXiv2023

Interplanetary Coronal Mass Ejections and Stream Interaction Regions observed by Tianwen-1 and Maven at Mars

Tianwen-1 spacecraft (Wan et al. 2020) is China's first Mars exploration mission. The Mars Orbiter Magnetometer (MOMAG) is a scientific instrument aboard the Tianwen-1 mission that is designed to study magnetic fields at Mars, including the solar wind to the magnetosheath and the ionosphere. Using the first Tianwen-1/MOMAG data that is publicly available, we present interplanetary coronal mass ejection (ICME) and stream interaction region (SIR) catalogues based on in-situ observations at Mars between November 16, 2021, and December 31, 2021. We compared the magnetic field intensity and vector magnetic field measurements from Tianwen-1/MOMAG and Mars Atmospheric Volatile EvolutioN (MAVEN)/MAG during the ICME and SIR interval and found a generally good consistency between them. Due to MAVEN's orbital adjustment since 2019, the Tianwen-1/MOMAG instrument is currently the almost unique interplanetary magnetic field monitor at Mars. The observations indicate that the MOMAG instrument on Tianwen-1 is performing well and can provide accurate measurements of the vector magnetic field in the near-Mars solar wind space. The multi-point observations combining MOMAG, MINPA, and MEPA on board Tianwen-1 with MAG, SWIA, and STATIC on board MAVEN will open a window to systematically study the characteristic of ICMEs and SIRs at Mars, and their influences on the Martian atmosphere and ionosphere.

physics.space-ph↗

arXiv2023

In-flight Calibration of the Magnetometer on the Mars Orbiter of Tianwen-1

Mars Orbiter Magnetometer (MOMAG) is one of seven science payloads onboard Tianwen-1's orbiter. Unlike most of the satellites, Tianwen-1's orbiter is not magnetically cleaned, and the boom where placed the magnetometer's sensors is not long enough. These pose many challenges to the magnetic field data processing. In this paper, we introduce the in-flight calibration process of the Tianwen-1/MOMAG. The magnetic interference from the spacecraft, including spacecraft generated dynamic field and slowly-changing offsets are cleaned in sequence. Then the calibrated magnetic field data are compared with the data from the Mars Atmosphere and Volatile EvolutioN (MAVEN). We find that some physical structures in the solar wind are consistent between the two data sets, and the distributions of the magnetic field strength in the solar wind are very similar. These results suggest that the in-flight calibration of the MOMAG is successful and the MOMAG provides reliable data for scientific research.

astro-ph.EP↗

arXiv2023

The Mars Orbiter Magnetometer of Tianwen-1: In-flight Performance and First Science Results

Mars Orbiter MAGnetometer (MOMAG) is a scientifc instrument onboard the orbiter of China's first mission for Mars -- Tianwen-1. It started to routinely measure the magnetic field from the solar wind to magnetic pile-up region surrounding Mars since November 13, 2021. Here we present its in-flight performance and first science results based on the first one and a half months' data. By comparing with the magnetic field data in the solar wind from the Mars Atmosphere and Volatile EvolutioN (MAVEN), the magnetic field by MOMAG is at the same level in magnitude, and the same magnetic structures with the similar variations in three components could be found in MOMAG data. In the first one and a half months, we recognize 158 clear bow shock (BS) crossings from MOMAG data, whose locations statistically match well with the modeled average BS. We also identify 5 pairs of simultaneous BS crossings of the Tianwen-1's orbiter and MAVEN. These BS crossings confirm the global shape of modeled BS as well as the south-north asymmetry of the Martian BS. Two presented cases in this paper suggest that the BS is probably more dynamic at flank than near the nose. So far, MOMAG performs well, and provides accurate magnetic field vectors. MOMAG is continuously scanning the magnetic field surrounding Mars. These measurements complemented by observations from MAVEN will undoubtedly advance our understanding of the plasma environment of Mars.

astro-ph.EP↗

arXiv2022

On the nature of photospheric horizontal magnetic field increase in major solar flares

Rapid increase of horizontal magnetic field ($B_h$) around the flaring polarity inversion line is the most prominent photospheric field change during flares. It is considered to be caused by the contraction of flare loops, the details behind which is still not fully understood. Here we investigate the $B_h$-increase in 35 major flares using HMI high-cadence vector magnetograms. We find that $B_h$-increase is always accompanied by the increase of field inclination. It usually initiates near the flare ribbons, showing step-like change in between the ribbons. In particular, its evolution in early flare phase shows close spatio-temporal correlation to flare ribbons. We further find that $B_h$-increase tends to have similar intensity in confined and eruptive flares, but larger spatial-extent in eruptive flares in a statistical sense. Its intensity and timescale have inverse and positive correlations to the initial ribbon separations, respectively. The results altogether are well consistent with a recent proposed scenario which suggests that the reconnection-driven contraction of flare loops enhances photospheric $B_h$ according to the ideal induction equation, providing statistical evidence to the reconnection-driven origin for $B_h$-increase for the first time.

astro-ph.SR↗

arXiv2022

Calculating Question Similarity is Enough: A New Method for KBQA Tasks

Knowledge Base Question Answering (KBQA) aims to answer natural language questions with the help of an external knowledge base. The core idea is to find the link between the internal knowledge behind questions and known triples of the knowledge base. Traditional KBQA task pipelines contain several steps, including entity recognition, entity linking, answering selection, etc. In this kind of pipeline methods, errors in any procedure will inevitably propagate to the final prediction. To address this challenge, this paper proposes a Corpus Generation - Retrieve Method (CGRM) with Pre-training Language Model (PLM) for the KBQA task. The major novelty lies in the design of the new method, wherein our approach, the knowledge enhanced T5 (kT5) model aims to generate natural language QA pairs based on Knowledge Graph triples and directly solve the QA by retrieving the synthetic dataset. The new method can extract more information about the entities from PLM to improve accuracy and simplify the processes. We test our method on NLPCC-ICCPOL 2016 KBQA dataset, and the results show that our method improves the performance of KBQA and the out straight-forward method is competitive with the state-of-the-art.

cs.CL↗

arXiv2022

A Roadmap for Big Model

With the rapid development of deep learning, training Big Models (BMs) for multiple downstream tasks becomes a popular paradigm. Researchers have achieved various outcomes in the construction of BMs and the BM application in many fields. At present, there is a lack of research work that sorts out the overall progress of BMs and guides the follow-up research. In this paper, we cover not only the BM technologies themselves but also the prerequisites for BM training and applications with BMs, dividing the BM review into four parts: Resource, Models, Key Technologies and Application. We introduce 16 specific BM-related topics in those four parts, they are Data, Knowledge, Computing System, Parallel Training System, Language Model, Vision Model, Multi-modal Model, Theory&Interpretability, Commonsense Reasoning, Reliability&Security, Governance, Evaluation, Machine Translation, Text Generation, Dialogue and Protein Research. In each topic, we summarize clearly the current studies and propose some future research directions. At the end of this paper, we conclude the further development of BMs in a more general view.

cs.LG↗

arXiv2021

The configuration and failed eruption of a complex magnetic flux rope above a δ sunspot region

Aims. We investigate the configuration of a complex flux rope above a δ sunspot region in NOAA AR 11515, and its eruptive expansion during a confined M5.3-class flare. Methods. We study the formation of the δ sunspot using continuum intensity images and photospheric vector magnetograms provided by SDO/HMI. We use EUV and UV images provided by SDO/AIA, and hard X-ray emission recorded by RHESSI to investigate the eruptive details. The coronal magnetic field is extrapolated with a non-linear force free field (NLFFF) method, based on which the flux rope is identified by calculating the twist number Tw and squashing factor Q. We search the null point via a modified Powell hybrid method. Results. The collision between two emerging spot groups form the δ sunspot. A bald patch (BP) forms at the collision location, above which a complex flux rope is identified. The flux rope has multiple layers, with one compact end and one bifurcated end, having Tw decreasing from the core to the boundary. A null point is located above the flux rope. The eruptive process consists of precursor flaring at a 'v'-shaped coronal structure, rise of the filament, and flaring below the filament, corresponding well with the NLFFF topological structures, including the null point and the flux rope with BP and hyperbolic flux tube (HFT). Two sets of post-flare loops and three flare ribbons support the bifurcation configuration of the flux rope. Conclusions. The precursor reconnection, which occurs at the null point, weakens the overlying confinement to allow the flux rope to rise, fitting the breakout model. The main phase reconnection, which may occur at the BP or HFT, facilitates the flux rope rising. The results suggest that the δ spot configuration presents an environment prone to the formation of complex magnetic configurations which will work together to produce activities.

astro-ph.SR↗

arXiv2020

PAL-Hom method for QP and an application to LP

In this paper, a proximal augmented Lagrangian homotopy (PAL-Hom) method for solving convex quadratic programming problems is proposed. This method takes the proximal augmented Lagrangian method as the outer iteration. To solve the proximal augmented Lagrangian subproblems, a homotopy method is presented as the inner iteration. The homotopy method tracks the piecewise-linear solution path of a parametric quadratic programming problem whose start problem takes an approximate solution as its solution and the target problem is the subproblem to be solved. To improve the performance of the homotopy method, the accelerated proximal gradient method is used to obtain a fairly good approximate solution that implies a good prediction of the optimal active set. Moreover, a sorting technique for the Cholesky factor update as well as an $\varepsilon$-relaxation technique for checking primal-dual feasibility and correcting the active sets are presented to improve the efficiency and robustness of the homotopy method. Simultaneously, a proximal-point-based AL-Hom method which is shown to converge in finite number of steps, is applied to linear programming. Numerical experiments on randomly generated problems and the problems from the CUTEr and Netlib test collections, support vector machines (SVMs) and contact problems of elasticity demonstrate that PAL-Hom is faster than the active-set methods and the parametric active set methods and is competitive to the interior-point methods and the specialized algorithms designed for specific models (e.g., sequential minimal optimization (SMO) method for SVMs).

math.OC↗

arXiv2020

APP-Hom Method for Box Constrained Quadratic Programming

In this paper, based on a $Q$-linear convergence analysis and an estimate of the linear convergence factor of the proximal point (PP) algorithm for solving box constrained quadratic programming (BQP) problems, an accelerated proximal point (APP) algorithm for solving BQP problems is presented. To solve the strictly convex BQP problems in each step of the APP algorithm, an efficient homotopy method, which tracks the solution path of a parametric quadratic program, is given. The algorithm with APP algorithm as outer iteration and the homotopy method as inner iteration is named by APP-Hom. The inner homotopy method is efficient by implementing, a warm-start technique based on the accelerated proximal gradient (APG) method, an $\varepsilon$-relaxation technique for checking prime and dual feasibility and determining/correcting the active set. Numerical tests for randomly generated dense and sparse BQPs, BQPs arising from image deblurring, BQPs in SVM, as well as discretized obstacle problem, elastic-plastic torsion problem, and the journal bearing problem show that the APP algorithm takes much less steps than the PP algorithm, the homotopy method is very efficient for strictly-convex BQP, and in consequence, that the APP-Hom is very efficient for non-convex BQP.

math.OC↗

arXiv2017

Realizing ICN in 3GPP's 5G NextGen Core Architecture

The proposed 3GPP's 5G Next-generation (NextGen) Core architecture (5GC) enables the ability to introduce new user and control plane functions within the context of network slicing to allow greater flexibility in handling of heterogeneous devices and applications. In this paper, we discuss the integration of such architecture with future networking technologies by focusing on the information centric networking (ICN) technology. For that purpose, we first provide a short description of the proposed 5GC, which is followed by a discussion on the extensions to 5GC's control and user planes to support Protocol Data Unit (PDU) sessions from ICN. To illustrate the value of enabling ICN within 5GC, we focus on two important network services that can be enabled by ICN data networks. The first case targets mobile edge computing for a connected car use case, whereas the second case targets seamless mobility support for ICN sessions. We present these discussions in consideration with the procedures proposed by 3GPP's 23.501 and 23.502 technical specifications.

cs.NI↗

arXiv2017

ICN-aware Network Slicing Framework for Mobile Data Distribution

Network slicing offers an opportunity to realize ICN as a slice in 5G deployment. We demonstrate this through a generic service orchestration framework operating on commodity compute, storage and bandwidth resource pool to realize multiple ICN service slices. Specifically, we show the dynamic creation of real-time audio/video conferencing slices, over which multi-participant communication is enabled. These slices leverage ICN features of name-based routing, integrated security, inherent support for multicasting and mobility, and in-network caching-and-computing to scale and deliver services efficiently, while dynamically adapting to varying service demands. Proposed framework also enables mobility-on-demand feature as a service over an ICN slice to more effectively support producer mobility over multi-access links, such as LTE, Wifi and Ethernet, as will be demonstrated with our demo.

cs.NI↗

arXiv2017

Demo Abstract: CDMA-based IoT Services with Shared Band Operation of LTE in 5G

With the vision of deployment of massive Internet-of-Things (IoTs) in 5G network, existing 4G network and protocols are inefficient to handle sporadic IoT traffic with requirements of low-latency, low control overhead and low power. To suffice these requirements, we propose a design of a PHY/MAC layer using Software Defined Radios (SDRs) that is backward compatible with existing OFDM based LTE protocols and supports CDMA based transmissions for low power IoT devices as well. This demo shows our implemented system based on that design and the viability of the proposal under different network scenarios.

cs.NI↗

arXiv2017

Realization of CDMA-based IoT Services with Shared Band Operation of LTE in 5G

5G network is envisioned to deploy a massive Internet-of-Things (IoTs) with requirements of low-latency, low control overhead and low power. Current 4G network is optimized for large bandwidth applications and inefficient to handle short sporadic IoT messages. The challenge here spans multiple layer including the radio access and the network layer. This paper focus on reusing CDMA access for IoT devices considering event-driven and latency sensitive traffic profile. We propose a PHY/MAC layer design for CDMA based communication for low power IoT devices. We propose and evaluate coexisting operation of CDMA based IoT network in presence of the exiting LTE network. Our proposed design will integrate IoT traffic with legacy system by minimal modification at the edge network, essentially eNodeB. We show that the underlay CDMA IoT network meets IoT data traffic requirements with minimal degradation (3%) in the LTE throughput. We also implement the proposed design using Software Defined Radios and show the viability of the proposal under different network scenarios.

cs.NI↗

arXiv2016

5G-ICN : Delivering ICN Services over 5G using Network Slicing

The challenging requirements of 5G--from both the applications and the architecture perspectives--motivate the need to explore the feasibility of delivering services over new network architectures. As 5G proposes application-centric network slicing, which enables the use of new data planes realizable over a programmable compute, storage, and transport infrastructure, we consider Information-centric Networking (ICN) as a candidate network architecture to realize 5G objectives. This can co-exist with end-to-end IP services that are offered today. To this effect, we first propose a 5G-ICN architecture and compare its benefits (i.e., innovative services offered by leveraging ICN features) to current 3GPP-based mobile architectures. We then introduce a general application-driven framework that emphasizes on the flexibility afforded by Network Function Virtualization (NFV) and Software Defined Networking (SDN) over which 5G-ICN can be realized. We specifically focus on the issue of how mobility-as-a-service (MaaS) can be realized as a 5G-ICN slice, and give an in-depth overview on resource provisioning and inter-dependencies and -coordinations among functional 5G-ICN slices to meet the MaaS objectives.

cs.NI↗