Scientific AI · Geomechanics · Subsurface Energy

Yulong Liu

My research connects scientific AI with computational geomechanics, with applications in subsurface energy systems.

I work on differentiable programming, hybrid neural PDE solvers, uncertainty-aware scientific AI, and AI for scientific discovery. These methods complement my work in homogenization and coupled multiphysics simulation to study geothermal reservoirs, hydraulic fracturing, and the interactions between fluid flow, heat transport, and rock deformation.

At Cornell, I work with Chloé Arson as a Ph.D. candidate in Earth Science, with minors in Computer Science and Scientific Computing. My projects span neural operators for geothermal reservoirs, physics-informed modeling of heterogeneous rock, and verification of AI-generated simulation code.

Before Cornell, I earned my B.S. in Mining Engineering at Northeastern University in 2024, where I worked on deep learning for microseismic signal analysis and tunnel boring machines (TBMs).

Explore my work Curriculum vitae

Research directions

AI methods & scientific discovery

  • Differentiable programming
  • Hybrid neural PDE solvers
  • Uncertainty-aware scientific AI
  • AI for Science & scientific discovery

Engineering & subsurface energy

  • Homogenization & multiscale modeling
  • Coupled multiphysics simulation
  • Geothermal energy & hydraulic fracturing

News

Research experience

  1. Scientific AI & computational geomechanics

    Cornell University · Advised by Chloé Arson.

    AI methods and scientific discovery; homogenization and coupled simulation for geothermal and hydraulic fracturing systems.

  2. Microseismic signal analysis

    Undergraduate thesis · Northeastern University, 2024.

    Deep learning for classification and arrival-time picking.

Education

  1. Cornell University

    2024 – present. Ph.D. in Earth Science.

    Minors in Computer Science and Scientific Computing.

  2. Northeastern University

    2020 – 2024. B.S. in Mining Engineering.

    College of Engineering · Shenyang, China.

Selected Recent Publications

Conference Papers

First page of Operator Learning Surrogate Modeling of Hydraulically Fractured Geothermal Injection-Production Systems: A Cornell Case Study

Operator Learning Surrogate Modeling of Hydraulically Fractured Geothermal Injection-Production Systems: A Cornell Case Study

Yulong Liu, Jonah Botvinick-Greenhouse, Yunan Yang, Chloé Arson

ARMA US Rock Mechanics/Geomechanics Symposium, Tucson, 2026

PaperPDF
First page of Physics-Informed Neural Network Surrogate Modeling of Pressurized Cavity in Homogeneous and Bilayered Media

Physics-Informed Neural Network Surrogate Modeling of Pressurized Cavity in Homogeneous and Bilayered Media

Yulong Liu (corresponding author), Chloé Arson

ARMA US Rock Mechanics/Geomechanics Symposium, 2025 · D022S018R006

PaperPDF

Journal Publications

First page of A Physics-Informed Neural Network for Pressurized Cavities of Arbitrary Shape in Heterogeneous Rock

A Physics-Informed Neural Network for Pressurized Cavities of Arbitrary Shape in Heterogeneous Rock

Yulong Liu, Chloé Arson

Rock Mechanics and Rock Engineering, 2026

Paper
First page of Numerical Assessment of Plasticity Development and Energy Expenditure of Ant-Like Microtunnelling

Numerical Assessment of Plasticity Development and Energy Expenditure of Ant-Like Microtunnelling

M. Belachew, Yulong Liu, J. D. Frost, Chloé Arson

Tunnelling and Underground Space Technology, 172, 107501, 2026

PaperPDF

Preprints & Manuscripts

First page of Your Simulation Runs but Solves the Wrong Physics: PDE-Grounded Intent Verification for LLM-Generated Multiphysics Simulation Code

Your Simulation Runs but Solves the Wrong Physics: PDE-Grounded Intent Verification for LLM-Generated Multiphysics Simulation Code

Zhenghan Song, Yulong Liu (co-first and corresponding author), et al.

arXiv preprint · Under review at NeurIPS 2026

arXivPDF
First page of Prefix-Safe Bayesian Belief Tracking for LLM Reasoning Reliability: Separating Calibration from Ranking

Prefix-Safe Bayesian Belief Tracking for LLM Reasoning Reliability: Separating Calibration from Ranking

Zhenghan Song, Yunyi Li, Yulong Liu (corresponding author)

Under review at ACL ARR 2026

arXivPDF

Awards & Research Support

Talks & Presentations

Academic Service

Reviewer: Rock Mechanics and Rock Engineering; Underground Space; International Conference on Artificial Neural Networks; Geothermal Rising Conference.

Professional memberships: ARMA (2025–present), ISRM (2025–present), CSRME (2024–present), and Geothermal Rising (2025–present).