Thesis (m/f/x) - 2D geophysical tomographic imaging by physics

• Comprehensive literature review on neural network based surrogate models and neural solvers for partial differential equations
• Get familiar with existing methodologies to solve numerical simulations by neural networks, e.g., based on the wave equation or Schrödinger’s equations, by neural solvers
• Adapt this (pre-trained) networks for geophysical forward and inverse modeling, e.g., 2D geophysical tomographic imaging
• Assess prediction error of the adapted neural solvers with respect to numerical schemes
• Quantitative and qualitative analysis of neural solvers for approaching 2D geophysical tomographic imaging

Your profile:

• Bachelor in Geophysics, Physics, Mathematics or Computer Science
• Experience in programming with Python (experience with deep learning tool boxes, e.g., PyTorch, is not required)
• Knowledge of numerical methods and linear algebra
• Knowledge of geophysical methods
• Fundamental knowledge of forward and inverse modeling
• Good will to work in an interdisciplinary team
• Fluency in English (verbal and written)
• Willingness to work at the Helmholtz Centers in Leipzig (mostly) and Dresden

We offer:

• A close and good supervision
• Exciting insights into the work of two leading research institutes