Fusion & real-time control
Disruption prediction, reinforcement-learning plasma control, tokamak surrogate transport models, KSTAR and ITER data
Program
ICDDPS-8 runs from 5 to 9 April 2027 at KAIST in Daejeon. Plenary and invited speakers, the session structure and the daily timetable will be announced on this page.
Scientific scope
Disruption prediction, reinforcement-learning plasma control, tokamak surrogate transport models, KSTAR and ITER data
Etch and deposition modelling, virtual metrology, run-to-run and chamber-matching control, semiconductor process data
Neural operators, physics-informed networks, reduced-order models, uncertainty quantification, accelerated kinetic simulation
Image and spectral analysis, tomography, sensor fusion, automated interpretation of high-rate diagnostic streams
FAIR data practice, cross-section and reaction databases, shared benchmarks, reproducibility in plasma ML
LLM agents for experiment planning, self-driving plasma experiments, active learning and Bayesian optimisation
Format
ICDDPS-8 is planned as a 4.5-day event following the shape of previous editions. The elements below describe that format; the ICDDPS-8 timetable is not yet fixed.
The conference language is English. Recent editions have been accompanied by Physics of Plasmas special topics open to conference participants.
Still to be announced