• Microstructure analysis of textile-reinforced composites
• Machine learning for materials science
• CNN/Attention-based prediction of Fiber orientation from volumetric data
• Bayesian optimization and Gaussian process regression for materials design
• Spectral methods (FFT) for homogenization and fracture
• Generative models for textile architecture generation

• Diffusion models for realistic textile microstructure generation
• VAE-guided generative modeling of composite architectures
• Bayesian optimization with Gaussian processes for microstructure-property exploration
• CNN-based structure tensor prediction for orientation estimation in low-resolution CT scans
• Crack detection and tracking using deep learning
• Multiscale modeling of damage and fracture in composites

Project	Description
	FFT-based spectral solvers for mechanics and materials modeling implemented in JAX
	PyTorch-based utilities for 3D segmentation of volumetric datasets
	Diffusion models for generating realistic textile architectures and microstructure fields

Languages

Scientific Computing

Semantic Segmentation | DDPM | FFT methods | Spectral solvers | Phase-field fracture | HPC

Tools

ParaView • Plotly • Blender • LaTeX • Git • HPC clusters

Yes… the username.

Before you ask: no, it’s not a deep cryptographic reference, a research acronym, or a secret composite-material formula.

It was 2020, peak Corona lockdown.
I was bored, opened a random username generator, pasted my name into it…

…and choROPeNt was born.

At this point it has accumulated too many commits to change it, so here we are.

Please direct all questions about fiber orientations, machine learning, or fracture mechanics to me —
but the username is a historical artifact.

• 🏔️ Hiking and exploring mountains
• 📷 Photography and imaging
• 🏀 Sports
• ✈️ Traveling and discovering new cultures

TU Dresden – Institute of Lightweight Engineering and Polymer Technology (ILK)

Christian Düreth