Soutenance de Nicolas Barros
| When |
Jul 21, 2025
from 01:30 to 04:30 |
|---|---|
| Where | Salle des thèses |
| Attendees |
Nicolas Barros |
| Add event to calendar |
vCal
iCal
|
This thesis aims to understand and optimize physical processes in small systems with few degrees of freedom, where thermal fluctuations play a central role. Stochastic thermodynamics provides a suitable framework to describe such random behaviors. We present a versatile and robust experimental setup designed to investigate the laws of thermodynamics at the nanoscale, using the fluctuations of a mechanical oscillator subjected to a tunable feedback potential. The system’s precision and reliability are demonstrated throughout our study by an excellent agree- ment with both simulations and theoretical predictions. We tackle several fundamental questions in stochastic thermodynamics related to energy exchanges and information processing. Thanks to great statistics and efficient out-of-equilibrium protocols, we show how the second law of thermodynamics can be locally surpassed in 95% of our experiments, while still holding on aver- age. We then apply tools from optimization theory and machine learning to perform irreversible logical operations efficiently in finite time, highlighting the benefits of this approach for broader objectives. Finally, we probe the experimental limits of our system to expand the range of oper- ations. By independently tuning the temperature and confinement of the cantilever, we pave the way for a comprehensive study of a one-particle heat engine. Overall, this work demonstrates that while classical thermodynamic principles remain valid at small scales, fluctuating systems reveal rich behaviors and significant opportunities for energetic optimization.