Hello there!
I'm Sebastian 👋

Postdoctoral Researcher at University of Regensburg  ·  PhD in Theoretical Physics

I'm a theoretical physicist with a background in numerical simulations, Bayesian inference, and statistical data analysis. Currently a postdoc at the University of Regensburg, transitioning into data science and machine learning.

This is my personal page. You'll find projects, publications, and a bit about me below.

bio

  1. 2025 —

    Postdoctoral Researcher in Theoretical Physics, University of Regensburg

    I currently work as a postdoctoral researcher in lattice QCD at the chair of Sara Collins. Day to day, that means Bayesian inference, uncertainty quantification, and statistical analysis of high-dimensional simulation data in Python.

  2. 2022 — 2025

    Doctoral Researcher in Theoretical Physics, University of Regensburg

    I completed my PhD in lattice QCD under the supervision of Christoph Lehner. My main research focused on the muon anomalous magnetic moment (g-2): I computed the short-distance window of the leading-order hadronic vacuum polarization for the first time from first-principles lattice QCD. I carried out the project end-to-end, including HPC simulations, data pipeline design, and a lot of statistical analysis.

    The resulting estimate is cited in the 2025 Muon g-2 Theory Initiative White Paper and contributes to the current Standard Model prediction of the muon g-2.

    In parallel, I collaborated with Gunnar Bali and Sara Collins on precision charmonium spectroscopy.

    Towards the end of my PhD, I squeezed in an internship in algorithm development, data analysis, and embedded software development at the start-up excav in Erlangen, building a Kalman filter for GNSS/IMU sensor fusion and deploying it on a microcontroller.

  3. 2018 — 2021

    M.Sc. in Physics, University of Regensburg

    During my master's degree, I specialized in theoretical physics with a focus on lattice QCD and computational methods. My thesis with Christoph Lehner focused on algorithmic improvements for Markov Chain Monte Carlo simulations in lattice field theory (Schwinger model) on HPC systems.

    In addition, I worked as a student assistant at Bertrandt in their Advanced Driver Assistance Systems team in Regensburg.

  4. 2014 — 2018

    B.Sc. in Physics, Friedrich-Alexander University Erlangen-Nürnberg

    I did my bachelor's thesis with Thorsten Glüsenkamp and Gisela Anton at ECAP, using unsupervised machine learning to explore novel particle detector geometries for the planned Gen2 expansion of the IceCube neutrino telescope at the South Pole.

projects

A selection of my projects. A more extensive list is available on GitHub.

python · uncertainty quantification

statpy

Python toolkit targeted at Markov Chain Monte Carlo data analysis in the context of lattice QCD.
monte carlo · control variates

caa-control-variates

Monte Carlo variance reduction (control-variate-style estimator) on real lattice-QCD data, with a measured 6.5–12.7× error reduction over baseline.
pytorch · transformer

shakespeare-transformer

Character-level GPT implementation in PyTorch from scratch. Trained on Shakespeare; KV caching gives a measured 1.83× inference speedup.
docker · sandboxing

docker-sandbox

Sandboxed Docker container for experimenting with AI coding assistants.
tooling · agentic coding

agent-config

My coding agent configuration including AGENTS.md and skills.

academia

My research focuses on lattice quantum chromodynamics (lattice QCD) — a first-principles approach to the theory of the strong interaction. Discretizing spacetime turns an otherwise intractable problem into one you can attack with numerical simulations, high-performance computing, and a lot of statistics.

publications

Phys. Rev. D 2025

High-precision continuum limit study of the HVP short-distance window

Sebastian Spiegel, Christoph Lehner

Physical Review D 111, 114517 (2025)

This work contributed to the 2025 Muon g-2 Theory Initiative White Paper. We performed the first high-precision, first-principles lattice QCD calculation of the hadronic vacuum polarization (HVP) short-distance window contribution to the muon anomalous magnetic moment $a_\mu = (g-2)/2$. The result directly enters the current Standard Model prediction of $a_\mu$, a key benchmark for precision tests of fundamental particle physics.

HVP SD Comparison
Comparison of HVP short-distance window contributions to the muon $g-2$. The result of our calculation is labeled SL-24. This plot is taken from the 2025 Muon $g-2$ Theory Initiative White Paper.

presentations & posters

personal

CrossFit Open 23.2
CrossFit Open 23.2

Outside the office, I enjoy being with good people or under a barbell.