arXiv subjects

Qin Feng

Publications and source records attributed to Qin Feng.

6 recordsLinked to original sources

arXiv2026

Lemon Agent Technical Report

Recent advanced LLM-powered agent systems have exhibited their remarkable capabilities in tackling complex, long-horizon tasks. Nevertheless, they still suffer from inherent limitations in resource efficiency, context management, and multimodal perception. Based on these observations, Lemon Agent is introduced, a multi-agent orchestrator-worker system built on a newly proposed AgentCortex framework, which formalizes the classic Planner-Executor-Memory paradigm through an adaptive task execution mechanism. Our system integrates a hierarchical self-adaptive scheduling mechanism that operates at both the overall orchestrator layer and workers layer. This mechanism can dynamically adjust computational intensity based on task complexity. It enables orchestrator to allocate one or more workers for parallel subtask execution, while workers can further improve operational efficiency by invoking tools concurrently. By virtue of this two-tier architecture, the system achieves synergistic balance between global task coordination and local task execution, thereby optimizing resource utilization and task processing efficiency in complex scenarios. To reduce context redundancy and increase information density during parallel steps, we adopt a three-tier progressive context management strategy. To make fuller use of historical information, we propose a self-evolving memory system, which can extract multi-dimensional valid information from all historical experiences to assist in completing similar tasks. Furthermore, we provide an enhanced MCP toolset. Empirical evaluations on authoritative benchmarks demonstrate that our Lemon Agent can achieve a state-of-the-art 91.36% overall accuracy on GAIA and secures the top position on the xbench-DeepSearch leaderboard with a score of 77+.

cs.MA↗

arXiv2025

3D fast-mode Wave Propagation from Corona to Chromosphere: Triggering Mechanism for 3D Oscillations of filaments

Moreton waves are widely regarded as the chromospheric counterpart of extreme ultraviolet (EUV) waves propagating in the corona. However, direct observational evidence confirming their simultaneous propagation across multiple atmospheric layers from the corona through the transition region to the chromosphere has been lacking. In this study, we present comprehensive observational evidence of a three-dimensional (3D) fast-mode wave propagating from the corona through the transition region into the chromosphere, exhibiting a gradual deceleration. Additionally, this wave interacts with three filaments (F1, F2, and F3) along its path, inducing oscillation with multiple amplitudes: Filaments F1 and F2 exhibit simultaneous horizontal and vertical large-scale oscillations ($\sim$\speed{20}), while Filament F3 only exhibits vertical small-scale oscillation ($\sim$\speed{4}). Interestingly, F1 displays a similar oscillation period of about 500\,s in both horizontal and vertical directions, whereas F2 shows significantly different periods in these two dimensions (1100\,s and 750\,s), and F3 exhibits only a vertical oscillation with a period of about 450\,s. Based on this kinematic behavior, we propose that their oscillations were likely triggered by compression from the flanks of the dome-shaped wavefront. We further estimate the magnetic fields of the filaments. The radial (axial) magnetic fields for F1 and F2 are estimated to be 14.9\,G (28.6\,G) and 9.9\,G (18.6\,G), respectively. For F3, we estimate its radial magnetic field to be 16.6\,G.

astro-ph.SR↗

arXiv2021

Direct Fidelity Estimation of Quantum States using Machine Learning

In almost all quantum applications, one of the key steps is to verify that the fidelity of the prepared quantum state meets expectations. In this Letter, we propose a new approach solving this problem using machine-learning techniques. Compared to other fidelity estimation methods, our method is applicable to arbitrary quantum states, the number of required measurement settings is small, and this number does not increase with the size of the system. For example, for a general five-qubit quantum state, only four measurement settings are required to predict its fidelity with $\pm1\%$ precision in a nonadversarial scenario. This machine-learning-based approach for estimating quantum state fidelity has the potential to be widely used in the field of quantum information.

quant-ph↗

arXiv2021

A Steering Paradox for Einstein-Podolsky-Rosen Argument and its Extended Inequality

The Einstein-Podolsky-Rosen (EPR) paradox is one of the milestones in quantum foundations, arising from the lack of local realistic description of quantum mechanics. The EPR paradox has stimulated an important concept of "quantum nonlocality", which manifests itself by three different types: quantum entanglement, quantum steering, and Bell nonlocality. Although Bell nonlocality is more often used to show the "quantum nonlocality", the original EPR paradox is essentially a steering paradox. In this work, we formulate the original EPR steering paradox into a contradiction equality,thus making it amenable to an experimental verification. We perform an experimental test of the steering paradox in a two-qubit scenario. Furthermore, by starting from the steering paradox, we generate a generalized linear steering inequality and transform this inequality into a mathematically equivalent form, which is more friendly for experimental implementation, i.e., one may only measure the observables in $x$-, $y$-, or $z$-axis of the Bloch sphere, rather than other arbitrary directions. We also perform experiments to demonstrate this scheme. Within the experimental errors, the experimental results coincide with the theoretical predictions. Our results deepen the understanding of quantum foundations and provide an efficient way to detect the steerability of quantum states.

quant-ph↗

arXiv2020

Experimentally undoing an unknown single-qubit unitary

Undoing a unitary operation, $i.e$. reversing its action, is the task of canceling the effects of a unitary evolution on a quantum system, and it may be easily achieved when the unitary is known. Given a unitary operation without any specific description, however, it is a hard and challenging task to realize the inverse operation. Recently, a universal quantum circuit has been proposed [Phys.Rev.Lett. 123, 210502 (2019)] to undo an arbitrary unknown $d$-dimensional unitary $U$ by implementing its inverse with a certain probability. In this letter, we report the experimental reversing of three single-qubit unitaries $(d = 2)$ by linear optical elements. The experimental results prove the feasibility of the reversing scheme, showing that the average fidelity of inverse unitaries is $F=0.9767\pm0.0048$, in close agreement with the theoretical prediction.

quant-ph↗

arXiv2020

On-Chip Multiphoton Entangled States by Path Identity

Multiphoton entanglement, as a quantum resource, plays an essential role in linear optical quantum information processing. Krenn et al. (Phys. Rev. Lett. 118, 080401 2017) proposed an innovative scheme that generating entanglement by path identity, in which two-photon interference (called Hong-Ou-Mandel effect) is not necessary in experiment. However, the experiments in this scheme have strict requirements in stability and scalability, which is difficult to be realized in bulk optics. To solve this problem, in this paper we first propose an on-chip scheme to generate multi-photon polarization entangled states, including Greenberger-Horne-Zeilinger (GHZ) states and W states. Moreover, we also present a class of generalized graphs for W states (odd-number-photon) by path identity in theory. The on-chip scheme can be implemented in existing integrated optical technology which is meaningful for multi-party entanglement distribution in quantum communication networks.

quant-ph↗