IRIS-HEP Fellow: Artem Havryliuk



Fellowship dates: Jun – Sep, 2022
Jun – Sep, 2023
Jul – Oct, 2024

Home Institution: Kyiv Academic University


Project: Data Classification with PointNet++ for the Active-Target Time Projection Chamber at FRIB

We have an Active-Target Time Projection Chamber (AT-TPC) detector, which consists of a gas chamber (the detector gas is at the same time the target material), a source of magnetic and electric fields, and the pad plane on which particles are detected. The task is to extract information from the data that we receive on the detector, such as information about energy and angles, as well as the classification of events to identify particles, using a model built on the basis of the PointNet architecture.

More information: My project proposal

Mentors:
  • Michelle Kuchera (Davidson College)


Project: Using Diffusion Probabilistic Models for Denoising Tracks from AT-TPC Detector

Preliminary work in Dr. Kuchera’s group investigated using diffusion probabilistic models (dpm) as a surrogate simulator for generating point clouds from the AT-TPC detector. I propose to continue this idea, but in order to denoise and clean charged particle tracks recorded by the AT-TPC detector. We plan to model the problem as unpaired, event-to-event domain translation. One domain would comprise AT-TPC events generated via simulation; the other would comprise events recorded during actual experiments with the detector. We expect the trained model to be able to remove noise from tracks.

More information: My project proposal

Mentors:
  • Michelle Kuchera (Davidson College)


Project: Array-Oriented Python Interface for the Pythia Event Generator

Pythia is a cornerstone C++ package used in particle physics for simulating high-energy collisions. Although Pythia's current Python interface is excellent for prototyping due to its comprehensive feature set, including bi-directional bindings, it becomes inefficient when analyzing large datasets because of its one-event-at-a-time approach. This project aims to develop an array-oriented Python interface for Pythia, leveraging the Awkward Array library to enhance performance and usability, particularly in interactive environments like Jupyter notebooks. By focusing on a streamlined feature set optimized for speed, this new interface will facilitate more efficient data handling, enabling the usage of Pythia through the Python interface for large-scale production-level simulation.

More information: My project proposal

Mentors:
  • Jim Pivarski, Philip James Ilten

Presentations and Publications
Current Status
March 2023 - Data Scientist at ALPhA

Contact me: