We borrow a Bayesian global optimisation algorithm from the machine learning community in order to find the optimum conditions in different physical systems quicker than textbook optimisation algorithms. This algorithm is specifically designed to iteratively find the optimum of expensive functions in two steps: 1. An auxiliary function (expressed as a Gaussian process) is used to mimic the original function as more and more observations are made; this function is relatively cheap to evaluate, and more importantly has analytic derivatives. 2. By finding the minimum of this auxiliary function using textbook gradient-based minimisation algorithms it is possible to have a better guess about where the minimum of the original function might be.
In recent years, social media and online social networking sites have become a major disseminator of false facts, urban legends, fake news, or, more generally, misinformation. To overcome this problem, online platforms are, on the one hand, empowering their users—the crowd—with the ability to evaluate the content they are exposed to and, on the other hand, resorting to trusted third parties for fact checking stories. However, given the noise in the evaluations provided by the crowd and the high cost of fact checking, the above mentioned measures require careful reasoning and smart algorithms. In this talk, I will first describe a modeling framework based on marked temporal point process that links noisy evaluations provided by the crowd to robust, unbiased and interpretable notions of information reliability and source trustworthiness. Then, I will introduce a scalable online algorithm, CURB, to select which stories to send for fact checking and when to do so to efficiently reduce the spread of fake news and misinformation with provable guarantees. Finally, I will show the effectiveness of our modeling framework and our algorithm using real-world data gathered from Wikipedia, Stack Overflow, Twitter and Weibo. This talk includes joint work with Behzad Tabibian, Jooyeon Kim, Isabel Valera, Mehrdad Farajtabar, Le Song, Alice Oh and Bernhard Schoelkopf.
In the first part of the talk, I will discuss the effects of short- range interactions in generalized Weyl semimetals with a monopole charge (n) greater than one. I will show that a strong enough short range interaction may lead to the onset of a translational symmetry breaking axion insulator or a rotational symmetry breaking gapless nematic state . To address this problem, I will use a new renormalization group scheme in which the monopole charge is an expansion parameter. The computed correlation length exponent ν=n/2, and therefore generalized interacting Weyl semimetals with n>1 provide rare examples of a non-Gaussian itinerant quantum criticality in three dimensions. I will discuss experimental signatures of the symmetry breaking phases in transport and thermodynamic responses. The second part of the talk is devoted to the role of the long-range Coulomb interaction in a simple (n=1) Weyl semimetal. In particular, I will show that this interaction leads to a universal enhancement of the zero-temperature optical conductivity that depends solely on the number of Weyl points at the Fermi level . This scaling is a remarkable consequence of an interplay between the quantum-critical nature of an interacting Weyl liquid, marginal irrelevance of the long-range Coulomb interaction and the violation of hyperscaling in three dimensions, and is directly measurable in recently discovered Weyl and Dirac materials.  B. Roy, P. Goswami, and V. Juricic, Phys. Rev. B 95, 201102 (R) (2017).  B. Roy and V. Juricic, Phys. Rev. B 96, 155117 (2017).
One hundred and fifty years ago, Maxwell first posed the thought experiment that become known as “Maxwell’s demon.” Designed to understand more deeply the nature of the newly formulated second law of thermodynamics, the demon was to play a long, controversial role in the development of statistical physics. Just two months later, Maxwell’s paper “On governors” gave the first analysis of a feedback system. These two foundational works reflect the fundamental and practical aspects of control. I will present an experiment that unites the two: using feedback to create “impossible” dynamics, we make a Maxwell demon that can reach the fundamental limits to control set by thermodynamics. We test—and then extend—Rolf Landauer’s 1961 prediction that information erasure requires at least as much work as can be extracted from a system by virtue of information. These fundamental thermodynamic limits are benchmarks for evaluating the performance of practical information engines, such as those active within cells and other complex systems.