Kerem Y. Camsari, PhD

  kcamsari AT purdue DOT edu

My name is Kerem Camsari and I am a post-doctoral researcher at the School of Electrical and  Computer Engineering at Purdue University with the Supriyo Datta group, where I have also received my PhD degree.


My research is broadly about connecting novel materials and phenomena (Physics) to functional devices and circuits (Systems).


My PhD work was about developing a modular circuit framework to explore and understand emerging phenomena to evaluate new spintronic devices and circuits. Each module is carefully benchmarked with established physical methods and available experiments in the fields of spintronics and magnetics. This framework has been in continuous development to incorporate new physics and materials and is used to understand, evaluate and propose new experiments, functional devices and circuits.


My PhD thesis is here.


My post PhD work has been about establishing a new kind of computing platform based on probabilistic (p)-bits and (p)-circuits. We have shown that p-bits can be compactly built using present day technology  as energy efficient, small-footprint building blocks to be used for a range of applications including combinatorial optimization, Bayesian Networks and stochastic neural networks that can perform  “invertible” Boolean logic. In particular, invertible logic allows the design of Boolean circuits that can be operated in reverse: A hardware multiplier can be used as a  factorizer when operated in its inverse mode, with potentially intriguing applications.


In addition to detailed device level simulations, we have experimentally built and investigated the behavior of p-circuits using microcontrollers, FPGAs and more recently with nanodevices such as stochastic Magnetic Tunnel Junctions (s-MTJs).


We have also shown that a p-bit can function as a "poor man’s q-bit"  to accelerate the simulation of a class of quantum circuits amenable to Quantum Monte Carlo treatments, using highly efficient autonomous hardware based on integrated MTJs.


Take a look at the IEEE Spectrum Article on our recent work on p-bits.


Check out a recent talk in the Spintronics meets Neuromorphics Workshop for an overview of p-bits and p-circuits.


Update (09/18/2019): Check out our recent paper on implementing an 8 p-bit computer with stochastic Magnetic Tunnel Junctions, along with an insightful commentary by Dmitri Nikonov and an Editorial.


My CV is here.