Lars Ruthotto
Scholar

Lars Ruthotto

Google Scholar ID: EaBF6r0AAAAJ
Emory University
Scientific ComputingInverse ProblemsPDE constrained optimizationMachine LearningImage Registration
Citations & Impact
All-time
Citations
4,319
 
H-index
25
 
i10-index
42
 
Publications
20
 
Co-authors
32
list available
Resume (English only)
Academic Achievements
  • Chair of the SIAM Activity Group on Data Science (since January 2024).
  • Section Editor for Machine Learning journal.
  • Co-organizer of the 'Future of AI and the Mathematical and Physical Sciences Workshop'.
  • Co-organizer of the NSF Computational Mathematics PI Meeting (May 2025, Salt Lake City).
  • Teaches courses including Math 785R: Deep Generative Modeling and a Deep Generative Modeling Workshop.
Research Experience
  • Research on numerical algorithms for high-dimensional differential equations, optimization, and inference.
  • Focus areas: generative models, continuous-time deep learning, mixed-precision training, and efficient numerical optimization.
  • Treats neural networks as dynamical systems, analyzed and trained via numerical methods.
  • Develops optimal-transport–based generative models, structure-exploiting optimizers, lean architectures, and mixed-precision algorithms for quantized networks.
  • Uses neural networks to approximate value functions and transport maps for high-dimensional optimal control, mean field games, and Bayesian inverse problems.
  • Designs learnable iterative solvers to accelerate PDE simulators.
  • Collaborates with national laboratories and industry partners; open to new collaborations.