Page of Simon Pfahler

Publications

• A Scalable Framework for Pan-Cancer Tumor Evolution Analysis Enables Transfer of Progression Mechanisms Across Tumor Entities
  Simon Pfahler, Andreas Lösch, Y. Linda Hu, Rudolf Schill, Rainer Spang, Tilo Wettig
  bioRxiv (2026)
• Easy colorblind-safe typesetting: General guidelines and a helpful LaTeX package
  Simon Pfahler
  TUGboat 45:2 (2024)
• Taming numerical imprecision by adapting the KL divergence to negative probabilities
  Simon Pfahler, Peter Georg, Rudolf Schill, Maren Klever, Lars Grasedyck, Rainer Spang, Tilo Wettig
  Stat. Comput. 34 (2024)
  also on arXiv:2312.103021 [stat.CO] (2023)
• Modeling metastatic progression from cross-sectional cancer genomics data
  Kevin Rupp, Andreas Lösch, Y. Linda Hu, Chenxi Nie, Rudolf Schill, Maren Klever, Simon Pfahler, Lars Grasedyck, Tilo Wettig, Niko Beerenwinkel, Rainer Spang
  Bioinformatics (2024)
  also on bioRxiv:2024.01.30.577989 (2024)
• Reconstructing Disease Histories in Huge Discrete State Spaces
  Rudolf Schill, Maren Klever, Kevin Rupp, Y. Linda Hu, Andreas Lösch, Peter Georg, Simon Pfahler, Stefan Vocht, Stefan Hansch, Tilo Wettig, Lars Grasedyck, Rainer Spang
  Künstl Intell (2024)

Posters & presentations

• Lattice 2025: A novel gauge-equivariant neural network architecture for preconditioners in lattice QCD
• RECOMB 2025: Exploiting weak modularity in cancer progression to infer large Mutual Hazard Networks
• Seminar Mathematik des Maschinellen Lernens: Exploiting symmetries to achieve fast model convergence
• COMPSTAT2024: Taming numerical imprecision by adapting the KL divergence to negative probabilities
• TUG24: Easy colorblind-safe typesetting - General guidelines and a helpful LaTeX package
• MECO49: Taming numerical imprecision by adapting the KL divergence to negative probabilities
• ISMB/ECCB 2023: Using low-rank tensor formats to enable computations of cancer progression models in large state spaces

Side projects

• Easy colorblind-safe typesetting in LaTeX: the colorblind package
• Rendering and exploration of Mandelbrot and Julia sets
• EQL÷ network implementation for neural symbolic regression, written in TensorFlow2