Past Events
24. June 2025, 16:00 until 17:00
Robert Stamps, Department of Physics and Astronomy, University of Manitoba, Canada
Seminar
Numerical models of nanomagnet arrays are used to demonstrate an approach for predictive coding that relies on stochastic exploration of states. Possible transitions to new, optimal magnetic configurations are sampled using a predictive coding algorithm which can be augmented using a feedback technique called “active inference”.[1] Numerical models of nanomagnets in artificial spin ice geometries are used to illustrate implementation of the algorithm as an hierarchical network and show how nanomagnets can be used to generate and navigate a complex state space.[2,3] Examples of tracking and PID control are demonstrated and shown to be consistent with a stochastic differential equation formulation of active inference.[4] Finally, micromagnetic simulations of a nanomagnet array are used to illustrate possible applications for feedback control of processes, selection of specific magnetic configurations, and as a method for real time optimisation of pulsed field nanomagnet switching protocols.
This work is supported by the Natural Sciences and Engineering Research Council of Canada RG- 430 and the Canada Foundation for Innovation JELF.
1. K Friston and S Kiebel, Philos Trans R Soc Lond B Biol Sci 364 (2009)
2. RB Popy, M Hamdi, RL Stamps, arXiv:2407.11165
3. J Frank, J van Lierop, RL Stamps, arXiv:2503.20983
4. RL Stamps, R. Popy, J. van Lierop, ACS Nano 19 (2025)
Event details
- Event location
-
SEM.R. DB gelb 05 B
1040 Wien
Wiedner Hauptstraße 8-10/E134 - Organiser
-
IAP
Manuela Marik
marik@iap.tuwien.ac.at - Public
- Yes
- Entrance fee
- No
- Registration required
- No