Sampling from exponential distributions in the time domain with superparamagnetic tunnel junctions
This work explores time domain statistics of superparamagnetic tunnel junctions (SMTJs). We leverage a temporal encoding scheme to experimentally sample exponential distributions from the dwell time statistics of these devices. We develop SMTJ-based circuits called probabilistic delay cells to produce stochastic time-delayed digital rising edges and describe circuits that can leverage these exponentially delayed signals to sample from other distributions of interest and use them as hardware primitives for probabilistic computations.
Authors
- M. D. Stiles (ORCID: https://orcid.org/0000-0001-8238-4156)
- Matthew W. Daniels (ORCID: https://orcid.org/0000-0002-3390-4714)
- William A. Borders (ORCID: https://orcid.org/0000-0002-0421-5282)
- Advait Madhavan (ORCID: https://orcid.org/0000-0002-4121-1336)
- Daniel P. Lathrop (ORCID: https://orcid.org/0000-0001-9282-3247)
- Sidra Gibeault (ORCID: https://orcid.org/0009-0005-0103-7168)
- U. Ebels (ORCID: https://orcid.org/0000-0001-5061-5538)
- Jabez J. McClelland (ORCID: https://orcid.org/0000-0001-5672-5965)
- Jason T. Ryan (ORCID: https://orcid.org/0000-0003-4501-0969)
- Philippe Talatchian (ORCID: https://orcid.org/0000-0003-2034-6140)
- Temitayo N. Adeyeye
Institutions
- National Institute of Standards and Technology (US)
- Spintronique et Technologie des Composants (FR)
- University of Maryland, College Park (US)
Publication Details
- Published
- 2025-09-18
- DOI
- https://doi.org/10.1117/12.3065707
- Primary Topic
- Quantum chaos and dynamical systems
- Type
- article
- Field-Weighted Citation Impact
- 0.00