February 10th, 2017 BORIS BOURDONCLE · FLORIAN CURCHOD ICFO-The Institute of Photonic Sciences

BORIS BOURDONCLE 'Generating Randomness against an Adversary with Memory'

Non-local correlations can be obtained in quantum physics. They contain some intrinsic private randomness. Quantitative relations between non-locality and randomness can be obtained in various framework and under diverse assumptions. There is a trade-off between the strength of the hypotheses that one assumes and the amount of randomness that one certifies. In this work, we investigate the framework of time-ordered no-signalling: any information task, such as randomness generation, or more generally, cryptography, is based on a protocol, i.e., successive choices of measurements leading to successive results. Here, we address the problem of evaluating randomness when the correlations at each round are constrained by the natural assumption that information can propagate forward in time but not backward. We give adversarial strategies that makes use of this information flow, and thus gives a higher predictability on the results, compared with the case where one considers that no information at all propagate between the rounds of a protocol.

FLORIAN CURCHOD 'Random Numbers Certified by Bell's Theorem'

Quantum theory, contrary to the classical one, is intrinsically probabilistic. Indeed, Born's rule only assigns probabilities to the outcomes of physical processes. However, a (classical) monkey rolling some biased dice can reproduce the outcomes' probabilities of any single quantum process. How can one then be sure that some numbers that were produced are really random? We will see that Bell Nonlocality offers a way to certify that the outcomes of certain physical processes are truly unpredictable. Monkeys are thus not playing dice.

PhD Seminar, February 10, 2017, 17:00. Seminar Room