Internship - Scalable Numerical Solvers for Systems of Learned Dynamical Components

MERL seeks a motivated graduate student to develop scalable numerical solvers for systems composed of learned dynamical components. The internship will focus on the numerical challenges that arise when multiple neural-network-based component models are interconnected through physical constraints and feedback, forming large-scale coupled ODE/DAE systems. The intern will develop numerical methods to improve solver robustness and simulation speed for these large-scale systems. Candidates should have a strong foundation in numerical analysis and scientific computing, together with a solid background in machine learning and data-driven methods. Strong programming skills in Python, Julia, C/C++, or Matlab are expected. Senior PhD students in applied mathematics, scientific computing, computational engineering, or related fields are encouraged to apply. The internship is expected to last 3 months, with a flexible start date, and may be extended depending on research progress and mutual interest.

The pay range for this internship position will be 6-8K per month.

Research Areas:

• Dynamical Systems
• Machine Learning
• Multi-Physical Modeling

Host: Hongtao Qiao