---------- Forwarded message --------- Da: Andreas Kyprianou ak257@bath.ac.uk Date: mer 1 lug 2020 alle ore 14:01 Subject: [owps] One World Probability Seminar Thursday 2nd July 2020: To: owps@lists.bath.ac.uk
One World Probability Seminar Thursday 2nd July 2020:
PLEASE NOTE WE ARE EXPERIMENTING WITH A NEW TIME: OWPS WILL START 1HR LATER THAN USUAL AT 1400 UTC
Tomorrow's speakers in the One World Probability Seminar are
------------------------------------------------------------
14.00 UTC: Ron Peled (Tel Aviv): Fluctuations of random surfaces and concentration inequalities for log-concave distributions
Abstract: Random surfaces in statistical physics are commonly modeled by a real-valued function on a d-dimensional lattice, whose probability density penalizes nearest-neighbor fluctuations according to an interaction potential U. The case U(x)=x^2 is the well-studied lattice Gaussian free field, while one-dimensional random surfaces are equivalent to random walks. Our focus is on dimensions d>=2 and general U. Brascamp-Lieb-Lebowitz conjectured in 1975 that such random surfaces are localized in dimensions d>=3 under mild assumptions on U. Their work establishes the conjecture when U is uniformly convex (its second derivative is uniformly bounded from zero), as a consequence of the Brascamp-Lieb concentration inequality. To date, this remains the main case for which the conjecture is verified, with the result missing even when U(x) = x^4. We establish new concentration inequalities for log-concave distributions, extending the Brascamp-Lieb inequality, and use them to prove localization in many new cases, including the family U(x) = |x| ^p with p>1. Further consequences regard the maximum height for a class of random surfaces discussed by Deuschel-Giacomin (2000).
The talk will be a gentle introduction to the model and the results. No prior knowledge of random surfaces or log-concave distributions will be assumed. Joint work with Alexander Magazinov.
15.00: UTC Omer Angel (Vancouver): Excited martingales
Abstract: We consider a random walk that moves in the Z^2 that moves vertically on the first visit to each site and horizontally on subsequent visits. We give new lower bounds on the growth of the range of such walks. Joint with Mark Holmes and Alejandro Ramirez.
------------------------------------------------------------
As always, the Zoom-room link will appear on the OWPS seminar webpage: https://www.owprobability.org/one-world-probability-seminar
But you can also link to it directly by clicking this link tomorrow: https://us02web.zoom.us/j/83512390999
Meeting-ID: 835 1239 0999
Please feel free to circulate this email.
We hope to see you all tomorrow! One World Probability Team
For the videos you can also subscribe to the Youtube or Bilibili channels:
https://www.youtube.com/channel/UCiLiEQGTp6bZEhuHDM-WNWQ
and
https://space.bilibili.com/151014650