The workshop will take place over five days, structured around the main themes that enter into the fitting of a machine learning potential. In particular:
ML interatomic potentials introduction
Representation and fitting
Configuration sampling and active learning (two days)
Validation and scientific applications
The structured sessions will each include several invited speakers, chosen from young and early-career researchers who have applied the theme in their work. Each speaker will be given one hour to present an interactive lesson with the goal of teaching participants how to use and apply a given method or software on an example problem, and to get started applying it to their own research problem if they find it useful.
There will additionally be "study room" time, where participants will be able to test out and experiment with the methods they have learned so far, with access to the organizers over chat and Zoom if needed. Poster sessions will take place at the end of two of the workshop days, during the study-room time. We will also encourage participants to join breakout rooms for networking and collaboration during these activities.
Virtual social activities
Even though ML-IP is a purely virtual conference, we have included a social dinner on Wednesday, 17th November (or the equivalent in local time), to help foster community and social interaction. This will involve participants bringing or ordering food and joining an online meeting to mimic a traditional social dinner or networking setting. There will additionally be social games and activities to encourage interaction and create a friendly atmosphere.
In addition, there will be plenty of opportunity for interaction and collaboration during the poster sessions and study/work time. We hope that these events can replace some of the spontaneous social interaction and potential collaborations that are often otherwise missing from purely virtual events.