Title: Simple betting strategies in warped casinos 有偏向性赌场轮盘投注策略的算法研究　

     Speaker: Georgios Barmpalias (Inst. of Software, CAS) (barmpalias.net) 

     Time: 15:00, June 8th (Friday), 2018 　　

    Venue: Room 334, Building 5, State Key Laboratory of Computer Science, Institute of Software, Chinese Academy of Sciences 

　　Abstract: Suppose that you know the casino roulette is rigged and there is an imbalance of red/black outcomes, at least in the limit. Then there is a strategy which only bets on red or only bets on black, which guarantees you unbounded profit. More generally, suppose that you have the restriction that you cannot bet the dollars you earn by betting on red, to bet on black and vice-versa. In the same casino there is a successful strategy of this kind, which does not depend on where the bias is (red or black) or even the degree of the bias (ie how far from 1/2 each outcome frequency can get in the limit). Sometimes casinos are rigged in more subtle ways, while satisfying all commonly used laws of large numbers like the relative frequency limit of each outcome tending to 1/2. Then are there simple winning strategies? We study this question from an algorithmic perspective, which is a natural approach since it is reasonable to expect that a strategy is programmable in a computer. We show that in the case of programmable strategies the answer is positive while in the case of countable mixtures of programmable strategies the answer is negative.This is based on a paper just finished with Barmpalias' student Fang Nan, and Andy Lewis-Pye. 

    Biography: Georgios Barmpalias is a 1000 young talents scholar in the State Key laboratory of the Institute of Software, in the Chinese Academy of Sciences.  He has been with the institute since 2011, first as an international young scientist of the National Natural Science Foundation of China, and then as an Associate professor and 1000 young talents scholar.  He has a PhD in mathematics from the University of Leeds in 2004, and has accumulated an extensive publication record of 70+ articles in journals, conference proceedings and book chapters in mathematics, computer science, statistical physics and general science venues. His expertise lies from mathematical logic and computability theory, to aspects of information theory and Kolmogorov complexity, algorithmic learning, and more recently mathematical analysis of social networks and discrete dynamical systems that model real-life situations. Prior to joining the academy of sciences he held positions in the University of Leeds (U.K.), the University of Amsterdam (Netherlands) and Victoria University of Wellington (New Zealand), where he established various collaborations and supervised a number of student and postoc projects.  George is a regular and often invited speaker in various international meetings and workshops, from the ASL meeting in Logic, various Dagstuhl and Oberwolfach meetings, to various annual computability, complexity and randomness conferences. George also has an extensive teaching record in various institutions in mathematics and computer science curricula, both in the graduate and the undergraduate levels.