Efficient Python

by Prof. Jacek Gatlik (AGH University of Krakow), Katarzyna Foryszewska (AGH University of Krakow)

Tuesday 7 Jul 2026, 09:00 → Tuesday 14 Jul 2026, 16:00 Europe/Zurich

593/R-010 - Salle 11 (CERN)

Course abstract

Efficient Python is a practice-oriented course on building reliable, readable, and high-performance workflows for scientific computing. It begins by treating vectors, matrices, and tensors as computational array objects, with particular emphasis on shapes, axes, indexing, broadcasting, and memory-aware operations. Participants then learn how to turn arrays into inspectable numerical experiments using NumPy and Matplotlib, and explore the numerical foundations of machine-learning workflows, including data organization, model fitting, losses, gradients, scaling, stability, and validation.

The course then moves from individual calculations to maintainable programs and reproducible data pipelines, covering explicit data flow, function interfaces, project organization, robust file handling, and labelled analysis with pandas. Its final part develops a disciplined approach to performance: benchmark first, locate bottlenecks through profiling and memory inspection, and only then select an appropriate acceleration strategy. Topics include algorithmic improvement, vectorization, memory control, selective JIT compilation with Numba, and concurrent or parallel execution. Throughout the course, scientific correctness, transparency, and reproducibility remain as important as speed.

Lecturer biography

Jacek Gatlik is an assistant professor at the Faculty of Physics and Applied Computer Science at AGH University of Krakow. He received his PhD in physics in 2024. His research lies at the intersection of nonlinear dynamics, computational physics, and scientific computing, with a particular focus on solitons and kink propagation in heterogeneous and time-dependent systems. His work combines analytical modelling with numerical simulations and effective reduced descriptions of nonlinear field equations, and has been published in journals such as Physical Review E and Chaos. He teaches Python programming, Unix and Linux systems, and has previously taught statistical physics and quantum mechanics. He has also carried out several extended research stays at INSA Rouen Normandie and the University of Massachusetts Amherst, collaborating on nonlinear-wave models and computational methods.

More info: www: https://galaxy.agh.edu.pl/~jgatlik/