Artur Toshev

I'm a PhD Candidate at Technical University of Munich since April 2021 where I am supervised by Professor Nikolaus Adams. My research combines Machine Learning and Particle-Based Fluid Dynamics.

More about me can be found in my Resume. If you have some questions, just drop me an email or connect with me on one of the following platforms.

Email /  X  /  Google Scholar  /  Github  /  Linkedin

profile photo
Research

I'm interested in large-scale physics-based machine learning, geometric deep learning, implicit neural representations, turbulence modelling, and uncertainty quantification. The current goal of my research is to speed-up conventional smoothed particle hydrodynamics simulations with machine learning.

Publications

Neural SPH: Improved Neural Modeling of Lagrangian Fluid Dynamics
A. P. Toshev, J. A. Erbesdobler, N. A. Adams, J. Brandstetter
ICML 2024
project page / ICML'24 / arXiv / poster / code

JAX-SPH: A Differentiable Smoothed Particle Hydrodynamics Framework
A. P. Toshev, H. Ramachandran, J. A. Erbesdobler, G. Galletti, J. Brandstetter, N. A. Adams
AI4DiffEq Workshop at ICLR 2024
AI4DiffEq@ICLR'24 / arXiv / poster / code

LagrangeBench: A Lagrangian Fluid Mechanics Benchmarking Suite
A. P. Toshev* , G. Galletti* , F. Fritz, S. Adami, N. A. Adams
NeurIPS 2023 Track on Datasets and Benchmarks
NeurIPS'23 / arXiv / poster / video / code

Accelerating Molecular Graph Neural Networks via Knowledge Distillation
F. E. Kelvinius* , D. Georgiev* , A. P. Toshev* , J. Gasteiger
NeurIPS 2023 / LOG 2023 (oral)
NeurIPS'23 / arXiv / poster / video @ LOG'23

Learning Lagrangian Fluid Mechanics with E(3)-Equivariant Graph Neural Networks
A. P. Toshev, G. Galletti, J. Brandstetter, S. Adami, N. A. Adams
Geometric Science of Information (GSI) 2023
arXiv / poster / slides / code

E(3) Equivariant Graph Neural Networks for Particle-Based Fluid Mechanics
A. P. Toshev, G. Galletti, J. Brandstetter, S. Adami, N. A. Adams
Physics4ML Workshop at ICLR 2023
⇾ Workshop version of the GSI paper above.

On the Relationships between Graph Neural Networks for the Simulation of Physical Systems and Classical Numerical Methods
A. P. Toshev, L. Paehler, A. Panizza, N. A. Adams
AI4Science Workshop at ICML 2022
arXiv / poster / slides

Teaching

I am happy to be a teaching assistant for the following courses at TUM.

  • AI for Science Seminar
    Summer '23, '24
    ⇾ New Master's level course
  • Introduction to Scientific Machine Learning for Engineers (lecture/exercise)
    Winter '22/23, '23/24, '24/25
    ⇾ Check out our SciML Jupyter Book, which I'm maintaining
  • Turbulent flows (exercise)
    Summer '22
  • Turbulent flow simulation on HPC systems (practical course)
    Winter '21/22


In addition, I was supervising students as a tutor during my Master's studies.

  • Summer 20 - Engineering Mechanics 2 (MSE)


Last update 29.10.2024


Template by Jon Barron.