arXiv subjects

Junjie Xia

Publications and source records attributed to Junjie Xia.

5 recordsLinked to original sources

arXiv2026

Future of Artificial Intelligence for Science in Japan 2024 Community Report

This white paper summarizes scientific challenges and AI/ML research opportunities identified through the FAIRS Japan 2024 unconference process. The discussion focuses on three major physics domains: accelerator physics, cosmology and astrophysics, and neutrino physics. Although each domain has distinct scientific goals and experimental constraints, several common technical themes emerge: high-dimensional reconstruction, fast and accurate simulation, uncertainty propagation, simulation-to-data mismatch, anomaly detection, real-time decision-making, and shared infrastructure.

hep-ph↗

arXiv2026

ForceVLA2: Unleashing Hybrid Force-Position Control with Force Awareness for Contact-Rich Manipulation

Embodied intelligence for contact-rich manipulation has predominantly relied on position control, while explicit awareness and regulation of interaction forces remain under-explored, limiting stability, precision, and robustness in real-world tasks. We propose ForceVLA2, an end-to-end vision-language-action framework that equips robots with hybrid force-position control and explicit force awareness. ForceVLA2 introduces force-based prompts into the VLM expert to construct force-aware task concepts across stages, and employs a Cross-Scale Mixture-of-Experts (MoE) in the action expert to adaptively fuse these concepts with real-time interaction forces for closed-loop hybrid force-position regulation. To support learning and evaluation, we construct ForceVLA2-Dataset, containing 1,000 trajectories over 5 contact-rich tasks, including wiping, pressing, and assembling, with multi-view images, task prompts, proprioceptive state, and force signals. Extensive experiments show that ForceVLA2 substantially improves success rates and reliability in contact-rich manipulation, outperforming pi0 and pi0.5 by 48.0% and 35.0%, respectively, across the 5 tasks, and mitigating common failure modes such as arm overload and unstable contact, thereby actively advancing force-aware interactive physical intelligence in VLAs. The project page is available at https://sites.google.com/view/force-vla2/home.

cs.RO↗

arXiv2025

Neutrino Oscillation Tomography of the Earth with the Hyper-Kamiokande Detector

Using PREM as a reference model for the Earth density distribution we investigate the sensitivity of the Hyper-Kamiokande (HK) detector to deviations of the Earth i) core average density $\barρ_C$, ii) lower mantle average density $\barρ_{lman}$) and iii) upper mantle average density $\barρ_{uman}$, from their respective PREM densities. The analysis is performed by studying the effects of the Earth matter on the oscillations of atmospheric $ν_μ$, $ν_e$, $\barν_μ$ and $\barν_e$. We implement the constraints on the variations of $ρ_C$, $ρ_{lman}$ and $ρ_{uman}$ following from the precise knowledge of the Earth mass $M_\oplus$ and moment of inertia $I_\oplus$, as well as from the requirement that the Earth be in hydrostatic equilibrium (EHE). These constraints limit in the case of the three layer Earth density structure we are considering the maximal positive deviation of $\barρ_C$ from its PREM value to $10\%$. Considering the case of normal ordering (NO) of neutrino masses, we present results which illustrate the dependence of sensitivity to the core, lower and upper mantle average densities on the energy and zenith angle resolutions and on the value of $θ_{23}$. We show, in particular, that in the ''nominal'' case of neutrino energy resolution $E_{res} = 30\%$ and zenith angle resolution $θ_{zres} = 20^\circ$ and for, e.g., $\sin^2θ_{23}=0.45~(0.58)$, HK can determine the average core density $\barρ_C$ at $2σ$ C.L. after 6500 days of operation with an uncertainty of (-14.5\%)/+39.5\% ((-9.3\%/+31.7\%). In the ''more favorable'' case of $E_{res}= 20\%$ and $θ_{zres} = 10^\circ$, and if $\sin^2θ_{23}=0.58~(0.45)$, the core density would be determined at $2σ$ C.L. with an uncertainty of (-8.3\%)/+9.8\% ((-9.2\%)/+11.3\%).

hep-ex↗

arXiv2022

Maximum likelihood reconstruction of water Cherenkov events with deep generative neural networks

Large water Cherenkov detectors have shaped our current knowledge of neutrino physics and nucleon decay, and will continue to do so in the foreseeable future. These highly capable detectors allow for directional and topological, as well as calorimetric information to be extracted from signals on their photosensors. The current state-of-the-art approach to water Cherenkov reconstruction relies on maximum-likelihood estimation, with several simplifying assumptions employed to make the problem tractable. In this paper, we describe neural networks that produce probability density functions for the signals at each photosensor, given a set of inputs that characterizes a particle in the detector. The neural networks we propose allow for likelihood-based approaches to event reconstruction with significantly fewer assumptions compared to traditional methods, and are thus expected to improve on the current performance of water Cherenkov detectors.

hep-ex↗

arXiv2018

Spatially Resolved Spectroscopic Study of nearby Seyfert Galaxies: Implications for a Population of "Missed" Seyferts at High-$\textit{z}$

We present mosaicked long-slit spectral maps of 18 nearby Active Galactic Nuclei (AGNs), 2 LINERs, and 4 star-forming galaxies. With the resulting data cubes taken using the Kast dual spectrograph on the 3 m Shane telescope of the Lick Observatory, we measure the aperture effects on the spectroscopic classification of AGNs. With more starlight included in a larger aperture, the nuclear spectrum that is Seyfert-like may become contaminated. We generated standard spectroscopic classification diagrams in different observing apertures. These show quantitatively how the ensemble of Seyferts migrates toward the H $\scriptsize{\textrm{II}}$ region classification when being observed with increasing aperture sizes. But the effect ranges widely in individual active galaxies. Some of the less luminous Seyferts shfit by a large amount, while some other barely move or even shift in different directions. We find that those Seyfert galaxies with the fraction of nuclear H$α$ emission lower than 0.2 of the host galaxy, 2-10 keV hard X-ray luminosity lower than $10^{43}$ erg s$^{-1}$, and the observed nuclear [O $\scriptsize{\textrm{III}}$] luminosity lower than $10^{40.5}$ erg s$^{-1}$, are more likely to change activity classification type when the entire host galaxy is included. Overall, 4 of our 24 galaxies (18 Seyferts) change their spectral activity classification type when observed with a very large aperture.

astro-ph.GA↗