arXiv subjects

Tingxiang Ji

Publications and source records attributed to Tingxiang Ji.

5 recordsLinked to original sources

arXiv2026

CENTILE: A Telemetry Foundation Model Evaluated by the Decisions It Drives

Modern computing and networking infrastructure emits telemetry continuously, yet operators convert it into decisions with a separate predictor per task, entity, and horizon. One generative model, pretrained once over an operator's own event streams, could replace this fleet, an approach that already scales to high-cardinality streams in recommendation systems. However, point-forecast error on operational telemetry saturates near simple last-value baselines, so lower error alone need not improve the decisions it feeds. To close this gap, we present \sys, a generative foundation model for network and systems telemetry, evaluated by replaying the decisions its calibrated conditional quantiles drive. \sys treats heterogeneous telemetry as event-driven, irregularly timed entity streams and serves flexible forecast horizons in a single pass, requiring no future timestamps. To our knowledge, \sys is the first pretrained telemetry model to improve both HPC scheduling and network provisioning decisions under replay, its runtime estimator transferring zero-shot across months and its pretrained weights across domains from hours of target data. Extensive experiments on HPC job logs and network traffic confirm that \sys lowers the mean bounded slowdown of backfilling by up to approximately $77\%$ over deployed user estimates and roughly halves the deployed rule's violation rate. Our code is available at https://github.com/ZzZTripleZzZ/all-in-one.

cs.NI↗

arXiv2026

Towards Efficient Synthesis of Quantum Graph States by Fusing Graph Motifs

Photonic graph states with advanced topologies can enable measurement-based quantum computing, distributed quantum sensing, and quantum interconnects. However, the efficient generation of photonic graph states is limited by the probabilistic nature of photonic entangling operations and the exponential dependence of generation rate on resource cost. In this work, we study photonic graph state synthesis as a cost-aware decomposition problem, exploiting local Clifford (LC) equivalence to identify more synthesis-friendly representations of the target graph state before decomposition. Specifically, we propose Cost-aware Fusion-based Decomposition (CFD), a three-stage heuristic framework that decomposes a target graph state into ring, star, and linear motifs, and assembles them via Type-I fusion operations to minimize fusion overhead and physical-qubit consumption. We further show that selecting the LC-equivalent graph state with the minimum number of edges provides a highly effective proxy for near-optimal synthesis: in many cases it matches the best generation rate observed within the LC equivalence class under CFD, and in most remaining cases it remains close to it. Numerical evaluations on graph state orbit data and 2D and 3D lattice graph states demonstrate that CFD achieves up to 84.6\% reduction in resource overhead compared to baseline constructions, and yields improvements in photonic generation rate spanning multiple orders of magnitude. These results suggest that combining structure-aware motif decomposition with LC equivalence is a practical and scalable strategy for photonic graph state synthesis.

quant-ph↗

arXiv2025

Quantum-enhanced Information Retrieval from Reflective Intelligent Surfaces

Information retrieval from passive backscatter systems is widely used in digital applications with tight energy budgets, short communication distances, and low data rates. Due to the fundamental limits of classical wireless receivers, the achievable data rate cannot be increased without compromising either energy efficiency or communication range, thereby hindering the broader adoption of this technology. In this work, we present a novel time-resolving quantum receiver combined with a multi-mode probing signal to extract large-alphabet information modulated by a passive reconfigurable intelligent surface (RIS). The adaptive nature of the proposed receiver yields significant quantum advantages over classical receivers without relying on complex or fragile quantum resources such as entanglement. Simulation results show that the proposed technique surpasses the classical standard quantum limit (SQL) for modulation sizes up to M = 2^8, meanwhile halving the probing energy or increasing the communication distance by a factor of 1.41.

quant-ph↗

arXiv2025

Distributing Arbitrary Quantum Graph States by Graph Transformation

Quantum graph state is a special class of nonlocal state among multiple quantum particles, underpinning several nonclassical and promising applications such as quantum computing and quantum secret sharing. Recently, establishing quantum graph states among physically distant nodes has gained increasing popularity owing to its potential in expanding current quantum applications in scale. Existing research on this topic relies on a two-step approach: first distributing low-dimension elementary entanglement to target nodes, and then fusing them into a high-dimension quantum graph state. However, most existing studies focus solely on minimizing costs (e.g., the number of elementary entanglements consumed) to entangle target nodes, while neglecting the structure of the final quantum graph state. This can easily result in weak system entanglement, jeopardizing the graph state under partial measurement or noises. In this paper, we aim to establish any arbitrary quantum graph states of strong entanglement structures at a much lower cost than the state of the art. The method is to search for and establish an alternative state to the target state that is of lowest cost in creation. Subsequently, we transform such an alternative state back to the target state via compressed single-qubit Clifford operations. To verify the performance of our developed algorithm, we conduct comprehensive simulations based on an open dataset containing all graph state structures up to 8 qubits. The results demonstrate fast algorithm convergence, an increased success probability in distributing any graph states, and $53.57\%$ saving in ERP cost compared with the state-of-the-art baseline.

quant-ph↗

arXiv2024

Post-hoc Interpretability Illumination for Scientific Interaction Discovery

Model interpretability and explainability have garnered substantial attention in recent years, particularly in decision-making applications. However, existing interpretability tools often fall short in delivering satisfactory performance due to limited capabilities or efficiency issues. To address these challenges, we propose a novel post-hoc method: Iterative Kings' Forests (iKF), designed to uncover complex multi-order interactions among variables. iKF iteratively selects the next most important variable, the "King", and constructs King's Forests by placing it at the root node of each tree to identify variables that interact with the "King". It then generates ranked short lists of important variables and interactions of varying orders. Additionally, iKF provides inference metrics to analyze the patterns of the selected interactions and classify them into one of three interaction types: Accompanied Interaction, Synergistic Interaction, and Hierarchical Interaction. Extensive experiments demonstrate the strong interpretive power of our proposed iKF, highlighting its great potential for explainable modeling and scientific discovery across diverse scientific fields.

cs.LG↗