Supriyo Datta

We have always been intrigued by Feynman’s (Int. J. Theor. Phys. 21, 1982) observation that "The only difference between a probabilistic classical world and the equations of the quantum world is that somehow or other it appears as if the probabilities would have to go negative .. “The awesome power of quantum computing comes from exploiting these negative (more generally complex) probabilities, which in turn requires stringent experimental conditions to protect the phase. A probabilistic computer by contrast can be built with existing technology to operate at room temperature to provide a benchmark for quantum computing. Which algorithms can and cannot be implemented without the magic of complex probabilities? With this question in mind, our group has introduced the concept of a p-bit that is intermediate between the bits of digital computing and the qubits of quantum computing. We are using this approach to accelerate a wide variety of problems including Bayesian networks, optimization, Ising models, quantum Monte Carlo, to name a few.