Relational quantum causality
Operator algebras, quantum causal models, and emergent geometry as a language for spacetime without a fixed background.
TheoryAI for science · Foundational physics · Cosmology
Tracing the causal fabric of spacetime.
I connect operator-algebraic models of quantum causality with cosmological observation—and build physics-informed AI to reveal faint structure in massive scientific data.
ResearchYX
CausalstructureCosmic
webPhysics-
informed AIOne question · three scales
From first principles
09Selected research papers
05Research programmes
01%MCM Finalist, 2026
01 · Research
A research programme across scales.
“How can we trace the causal fabric of quantum spacetime within the macroscopic cosmic web?”
My work moves between mathematical structure, observable signatures, and computational tools. The goal is not merely to make AI faster, but to make it respect—and help uncover—the laws that shape physical reality.
Cosmological observables
Large-scale structure and weak lensing as empirical probes of foundational physics and causal signatures.
ObservationPhysics-informed AI
Generative and context-aware models designed to preserve known constraints while extracting faint physical signals.
MethodScientific computing
Reproducible, high-performance workflows for simulation, inference, and data mining at survey scale.
Infrastructure02 · Selected papers
Work in theory, vision, and scientific AI.
Quantum gravity2026
Relational Quantum Causal Processes for Quantum Gravity: From Operator-Algebraic Facts to Causal Geometry, Vacuum-Response Sequestering, and Noncommutative Matter
Preprint · Zenodo
Scientific discoveryWorking paper
Neuro-symbolic discovery of interpretable subgrid-scale closures for large-eddy simulation via constrained Monte Carlo tree search
Intended for Journal of Fluid Mechanics
Embodied AI2025
A Unified World Model is the cornerstone for integrating perception, reasoning, and decision-making in embodied AI
NeurIPS 2025 Workshop · Embodied World Models for Decision Making
Video reasoningUnder review
ActiveLook: Progressive Multi-Scale Active Perception for Long Video Reasoning
IEEE Transactions on Multimedia
Computer vision2025
GCA-SUNet: A Gated Context-Aware Swin-UNet for Exemplar-Free Counting
IEEE ICME 2025
AI for engineeringAccepted
Data-driven machine learning for scale-up of bubbling and turbulent fluidized beds: Flow hydrodynamics and reactor performance
Engineering Applications of Computational Fluid Mechanics
Time-series forecasting2026
An Empirical Study of Turbidity Forecasting on Open Aquaponics Sensor Data
IEEE COMPSAC · Trustworthy Agri-Data Management
Computer visionUnder review
LLaVA-SEGA: LLaVA-guided Semantic-Enhanced Group-Attended Framework for Class-Agnostic Counting
IEEE Transactions on Image Processing
Multi-agent systemsWorking paper
Hybrid STGAT-MAPPO: Robust Multi-Agent Graph Reinforcement Learning for Scalable LEO Satellite Packet Routing
Intended for IEEE Transactions on Network and Service Management
* Equal-contribution authorship applies where noted in the full CV.
03 · Research experience
Building ideas into working science.
May 2024 — PresentCo-first contributor
Open-world object counting
AIOP Research Group · UNNC
Developing context-aware and cross-modal counting systems, including GCA-SUNet and LLaVA-SEGA, to reason beyond fixed visual categories.
Jun — Oct 2025Project leader
AI foundation models for NEV batteries
UNNC · Challenge Cup National Grand Award
Led a multidisciplinary team building an edge–cloud system for battery state-of-health forecasting and thermal-runaway early warning.
Jun 2023 — Apr 2026Key contributor
Physics-inspired machine learning for fluidized beds
UNNC · Western University
Built physics-inspired neural models for industrial scale-up, preserving flow and reaction behavior while reducing computational cost.
Oct 2024 — Aug 2025Group leader
Interactive automatic image labeling
Computer Vision and Intelligent Perception Laboratory · UNNC
Led the development of a deployable annotation platform integrating model ensembles, fine-tuning, and human-centered post-processing.
Jul 2022 — Jun 2023Key contributor
Semi-supervised galaxy morphology classification
National Astronomical Observatories · CAS
Designed multi-scale contrastive and adversarial training strategies for astronomical images with sparse, noisy, and imbalanced labels.
04 · About
Mechanics taught me systems. Computer science taught me models.
I began in mechanical engineering before transferring to computer science, carrying a physical intuition for complex systems into machine learning. At Cambridge, I plan to bring those threads together through quantum gravity, computational cosmology, and physics-informed AI.
Education
- 2026–27MPhil, Data Intensive Science
University of Cambridge · Expected
- 2023–26BSc, Computer Science
University of Nottingham Ningbo China · First-Class Honours
- 2021–23BEng coursework, Mechanical Engineering
University of Nottingham Ningbo China · Transferred to CS
Selected recognition
- MCM Finalist · Top 1%
- Zhejiang Provincial Outstanding Graduate
- Challenge Cup · National Grand Award
- Chinese Astronomical Society Innovation Contest · First Place
- Zhejiang Provincial Government Scholarship · Top 2%
- Dream Scholarship · Science & Technology
- NAOC Undergraduate Research Fellowship
05 · Contact
Let's investigate something fundamental.
I welcome conversations around quantum foundations, cosmology, scientific machine learning, and ambitious interdisciplinary work.
yipeng.xu@ieee.org