BEGIN:VCALENDAR
VERSION:2.0
PRODID:researchseminars.org
CALSCALE:GREGORIAN
X-WR-CALNAME:researchseminars.org
BEGIN:VEVENT
SUMMARY:Manan Bhatia (Indian Institute of Science\, Bangalore)
DTSTART:20200729T090000Z
DTEND:20200729T104500Z
DTSTAMP:20260404T095121Z
UID:BPS/1
DESCRIPTION:Title: Moderate deviation estimates in stationary last passage percolation.\nby Manan Bhatia (Indian Institute of Science\, Bangalore) as part of Ba
ngalore Probability Seminar\n\n\nAbstract\nThe planar exponential last pas
sage percolation model is an important integrable model in the (1+1) dimen
sional KPZ universality class. In the first talk\, we introduce the model
and its connections to the TASEP and give some examples of the known resul
ts for the different initial conditions. In the second technical talk\, we
go through a proof of the optimal exponent right tail estimates for the e
xit time for the stationary initial condition\, as well as the upper and l
ower tail estimates for the stationary passage time.\n\nIt will be two tal
ks of 45 minutes each with a 10-15 minutes break.\n
LOCATION:https://stable.researchseminars.org/talk/BPS/1/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Antar Bandyopadhyay (Indian Statistical Institute\, Delhi)
DTSTART:20200812T090000Z
DTEND:20200812T095000Z
DTSTAMP:20260404T095121Z
UID:BPS/2
DESCRIPTION:Title: A Last Progeny Modified Branching Random Walk\nby Antar Bandyopadhy
ay (Indian Statistical Institute\, Delhi) as part of Bangalore Probability
Seminar\n\n\nAbstract\nIn this work\, we consider a modification of the u
sual Branching Random Walk (BRW)\, where at the n-th step we give certain
i.i.d. displacements to each individuals\, which may be different from th
e driving increment distribution. Depending on the value a parameter\, we
classify the model in three distinct cases\, namely\, the boundary case\,
below the boundary case and above the boundary case. Under very minimal a
ssumptions on the underlying point process of the increments\, we show tha
t at the boundary case\, the maximum displacement converges to a limit aft
er only an appropriate centering\, which is of the form c1 n - c2log n. We
give explicit formula for the constants c1 and c2 and show that c1 is exa
ctly same\, while c2 is 1/3 of the corresponding constants of the usual B
RW. We also characterize the limiting distribution. We further show that b
elow the boundary the logarithmic correction term is absent. For above the
boundary case\, we have only partial result which indicates a possible ex
istence of the logarithmic correction in the centering with exactly same
constant as that of the classical BRW. Our proofs are based on a novel met
hod of coupling with a more well studied process known as the smoothing t
ransformation\, which is used in various non-parametric statistical metho
ds.\n
LOCATION:https://stable.researchseminars.org/talk/BPS/2/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Parthanil Roy (Indian Statistical Institute\, Bangalore)
DTSTART:20200812T100000Z
DTEND:20200812T104500Z
DTSTAMP:20260404T095121Z
UID:BPS/3
DESCRIPTION:Title: Group measure space construction\, ergodicity and superrigidity for sta
ble random fields\nby Parthanil Roy (Indian Statistical Institute\, Ba
ngalore) as part of Bangalore Probability Seminar\n\n\nAbstract\nIn this w
ork\, it is established that the group measure space construction correspo
nding to a minimal representation is an invariant of a stationary symmetri
c $\\alpha$-stable (S$\\alpha$S) random field indexed by any countable gro
up $G$. When $G=\\mathbb{Z}^d$\, we characterize ergodicity of stationary
S$\\alpha$S fields in terms of the central decomposition of this crossed p
roduct von Neumann algebra coming from any (not necessarily minimal) Rosin
ski representation. This shows that ergodicity is a $W^*$-rigid property (
in a suitable sense) for this class of fields. All our results have analog
ues for stationary max-stable random fields as well.\n
LOCATION:https://stable.researchseminars.org/talk/BPS/3/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Hao Wu (Yau Mathematical Science Center\, Tsinghua University)
DTSTART:20200826T050000Z
DTEND:20200826T064500Z
DTSTAMP:20260404T095121Z
UID:BPS/4
DESCRIPTION:Title: Crossing Probabilities in 2D Critical Lattice Models\nby Hao Wu (Ya
u Mathematical Science Center\, Tsinghua University) as part of Bangalore
Probability Seminar\n\n\nAbstract\nThe planar Ising model is one of the mo
st studied lattice models in statistical physics. It was introduced in the
1920s by W. Lenz as a model for magnetic materials. R. Peierls showed in
1936\, in two (and higher) dimensions\, an order-disorder phase transition
in fact occurs at a certain critical temperature. Ever since\, there has
been active research to understand the 2D Ising model at criticality\, whe
re it enjoys conformal invariance in the scaling limit. In this talk\, we
give crossing probabilities of multiple interfaces in the critical planar
Ising model with alternating boundary conditions. Besides\, we also expla
in that a similar formula on the crossing probabilities also holds for cri
tical Percolation and level lines of Gaussian Free Field.\n
LOCATION:https://stable.researchseminars.org/talk/BPS/4/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Joseph Yukich (Lehigh University\, USA)
DTSTART:20200923T110000Z
DTEND:20200923T124500Z
DTSTAMP:20260404T095121Z
UID:BPS/5
DESCRIPTION:Title: Convex Hulls of Random Point Sets\nby Joseph Yukich (Lehigh Univers
ity\, USA) as part of Bangalore Probability Seminar\n\n\nAbstract\nThe con
vex hull of a point set is the smallest convex polytope containing the poi
nt set. Convex hulls of random points in Euclidean space arise naturall
y in a wide variety of disciplines\, including optimization\, computationa
l geometry\, statistics\, economics\, and ethology. This talk will surve
y old and new results describing statistics of the convex hull on large da
ta sets. We shall also discuss more recent results concerning the fluctuat
ions and the scaling limit of the convex hull boundary. This talk is based
on joint work with Pierre Calka.\n
LOCATION:https://stable.researchseminars.org/talk/BPS/5/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Rahul Roy (Indian Statistical Institute\, Delhi)
DTSTART:20201007T090000Z
DTEND:20201007T104500Z
DTSTAMP:20260404T095121Z
UID:BPS/6
DESCRIPTION:Title: On models of evolution of species\nby Rahul Roy (Indian Statistical
Institute\, Delhi) as part of Bangalore Probability Seminar\n\n\nAbstract
\nThis talk has two parts. In the first part we discuss the Bak-Sneppen an
d the\nGuiol-Machado-Schinazi (GMS) models of evolution.\n\nIn the second
part of the talk we study the variation of the GMS model introduced by Ben
-Ari and Schinazi (2016). This model is a birth and death model with an in
dividual at birth being either a mutant with a random fitness parameter in
[0\, 1] or having one of the existing fitness parameters with uniform pro
bability\; whereas a death event removes the entire population of the leas
t fit site. We change this to incorporate the notion of ‘survival of the
fittest’\, by requiring that a non-mutant individual\, at birth\, has a
fitness according to a preferential attachment mechanism\, i.e.\, it has
a fitness f with a probability proportional to the size of the population
of fitness f. Also death just removes one individual at the least fit site
. This preferential attachment rule leads to a power law behaviour in the
asymptotics\, unlike the exponential behaviour obtained by Ben-Ari and Sch
inazi (2016).\n
LOCATION:https://stable.researchseminars.org/talk/BPS/6/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Manjunath Krishnapur (Indian Institute of Science\, Bangalore)
DTSTART:20201021T090000Z
DTEND:20201021T110000Z
DTSTAMP:20260404T095121Z
UID:BPS/7
DESCRIPTION:Title: Two proofs of the KMT theorems\nby Manjunath Krishnapur (Indian Ins
titute of Science\, Bangalore) as part of Bangalore Probability Seminar\n\
n\nAbstract\nThe Komlós-Major-Tusnády theorem for simple symmetric rando
m walk asserts that up to n steps\, its path can be coupled to stay within
distance log(n) of a Brownian motion run for time n. A second KMT theorem
says that the empirical distribution function of n i.i.d. uniform random
variables on [0\,1] can be coupled to stay within log(n)/√n distance of
a Brownian bridge.\n\nAdding the idea of Cauchy criterion to existing proo
f architectures\, we obtain (perhaps) simpler proofs of the above theorems
. The first proof compares two Binomial distributions by combinatorial met
hods. The second proof compares Binomial and hypergeometric distributions
among themselves by coupling Markov chains with these as stationary distr
ibutions. This is based on Chatterjee's proof via a form of Stein's method
.\n\nThe first lecture will give an overview and the essence of the first
proof. The second lecture will give an account of the second proof. Despit
e the statement of the main results\, much of the lecture should be acces
sible (without knowing about Brownian motion) to those who know Markov cha
ins.\n
LOCATION:https://stable.researchseminars.org/talk/BPS/7/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Ryoki Fukushima (Univ. Tsukuba\, Japan)
DTSTART:20201104T050000Z
DTEND:20201104T070000Z
DTSTAMP:20260404T095121Z
UID:BPS/10
DESCRIPTION:Title: Annealed random walk conditioned on survival among Bernoulli obstacles
\nby Ryoki Fukushima (Univ. Tsukuba\, Japan) as part of Bangalore Prob
ability Seminar\n\n\nAbstract\nI will present two recent results on a disc
rete time random walk conditioned to avoid Bernoulli obstacles on the d-di
mensional integer lattice obtained in joint works with Jian Ding\, Rongfen
g Sun and Changji Xu. The first result on this model dates back to a famou
s work by Donsker and Varadhan on the Wiener sausage in 1975. Since then\,
it has been intensively studied and various localization results have bee
n proved. In particular\, the random walk is known to localize in a ball o
f sub-diffusive size under the annealed law. Our first result gives a more
detailed geometric description of the range of the random walk. More prec
isely\, we showed that it completely fills the ball where the walk is loca
lized\, and in addition we got a sharp estimate on the size of its boundar
y. \n\nOur second result is about the response to an external force. If we
give a bias to the random walk\, then the model is known to undergo a pha
se transition: for a large bias\, the walk is ballistic whereas for a smal
l bias\, it is sub-ballistic. This phase transition was proved by Sznitman
and later\, Ioffe and Velenik studied the ballistic phase in detail. In t
he sub-ballistic phase\, physicists conjectured that the walk is localized
in a sub-diffusive scale as in the unbiased case\, but it has not been pr
oved. We prove this conjecture with a precise information on the behavior
of whole path.\n
LOCATION:https://stable.researchseminars.org/talk/BPS/10/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Dmitry Belyaev (Univ. Oxford\, UK)
DTSTART:20200916T090000Z
DTEND:20200916T094500Z
DTSTAMP:20260404T095121Z
UID:BPS/11
DESCRIPTION:Title: On the number of level sets of smooth Gaussian fields - I\nby Dmit
ry Belyaev (Univ. Oxford\, UK) as part of Bangalore Probability Seminar\n\
n\nAbstract\nLevel sets of smooth Gaussian fields appear in many areas of
mathematics. They have numerous applications outside of mathematics from a
strophysics to oceanology to biology. Probably the first significant progr
ess in the study of level lines came in 1940s when Kac and Rice developed
formulas that allow to compute the expected number of roots of a random fu
nction in 1d. These formulas can be generalised to higher dimension where
they allow to compute the expected volume of level sets. Unfortunately\, t
hese formulas do not allow to study the number of level lines since it is
a non-local quantity which can not be written as a Kac-Rice-type integral
formula. In this talk we will discuss recent progress in the study of the
number of level lines for smooth Gaussian fields. \n\nIn the first part of
the talk D. Belyaev will give a gentle introduction to the area and give
a broad overview of recent results. In particular\, he will describe how t
he number of level lines depends on the level. This allows to show that th
e expected number of lines depends (almost) smoothly on the level and allo
ws to give a low bound on the fluctuation of the number of level lines. In
the second part S. Muirhead will explain in more details the ideas behind
these results and will sketch the proof of the variance bound.The talk is
based on a series of papers by D. Belyaev\, M. McAuley and S. Muirhead.\n
LOCATION:https://stable.researchseminars.org/talk/BPS/11/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Stephen Muirhead (Univ. Melbourne\, Australia)
DTSTART:20200916T100000Z
DTEND:20200916T104500Z
DTSTAMP:20260404T095121Z
UID:BPS/12
DESCRIPTION:Title: On the number of level sets of smooth Gaussian fields - II\nby Ste
phen Muirhead (Univ. Melbourne\, Australia) as part of Bangalore Probabili
ty Seminar\n\n\nAbstract\nLevel sets of smooth Gaussian fields appear in m
any areas of mathematics. They have numerous applications outside of mathe
matics from astrophysics to oceanology to biology. Probably the first sign
ificant progress in the study of level lines came in 1940s when Kac and Ri
ce developed formulas that allow to compute the expected number of roots o
f a random function in 1d. These formulas can be generalised to higher dim
ension where they allow to compute the expected volume of level sets. Unfo
rtunately\, these formulas do not allow to study the number of level lines
since it is a non-local quantity which can not be written as a Kac-Rice-t
ype integral formula. In this talk we will discuss recent progress in the
study of the number of level lines for smooth Gaussian fields. \n\nIn the
first part of the talk D. Belyaev will give a gentle introduction to the a
rea and give a broad overview of recent results. In particular\, he will d
escribe how the number of level lines depends on the level. This allows to
show that the expected number of lines depends (almost) smoothly on the l
evel and allows to give a low bound on the fluctuation of the number of le
vel lines. In the second part S. Muirhead will explain in more details the
ideas behind these results and will sketch the proof of the variance boun
d.The talk is based on a series of papers by D. Belyaev\, M. McAuley and S
. Muirhead.\n
LOCATION:https://stable.researchseminars.org/talk/BPS/12/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Radoslaw Adamczak (Univ. Warsaw\, Poland)
DTSTART:20201202T090000Z
DTEND:20201202T110000Z
DTSTAMP:20260404T095121Z
UID:BPS/13
DESCRIPTION:Title: Functional inequalities and moment estimates\nby Radoslaw Adamczak
(Univ. Warsaw\, Poland) as part of Bangalore Probability Seminar\n\n\nAbs
tract\nI will present moment estimates\, which can be obtained by means of
functional inequalities\, from the classical work due to Aida and Stroock
from the 1990s\, through generalizations of the Efron-Stein inequality fo
r functions of independent random variables by Boucheron-Bousquet-Lugosi-M
assart (2005) to more recent estimates obtained with various coauthors\n
LOCATION:https://stable.researchseminars.org/talk/BPS/13/
END:VEVENT
BEGIN:VEVENT
SUMMARY:D. Yogeshwaran (Indian Statistical Institute\, Bangalore)
DTSTART:20201125T090000Z
DTEND:20201125T110000Z
DTSTAMP:20260404T095121Z
UID:BPS/14
DESCRIPTION:Title: Sharp phase transition and noise sensitivity in continuum percolation
via continuous time decision trees.\nby D. Yogeshwaran (Indian Statist
ical Institute\, Bangalore) as part of Bangalore Probability Seminar\n\n\n
Abstract\nProofs of sharp phase transition and noise sensitivity in percol
ation \nhave been significantly simplified by the use of randomized algori
thms \nvia the OSSS (O'Donnell\, Saks\, Schramm and Servedio) variance ine
quality\nand the Schramm-Steif inequality. In a joint work with Giovanni
Peccati\nand Guenter Last\, we prove analogues of these inequalities for a
\nPoisson point process. This talk will be about these two inequalities \
nand their applications to continuum percolation. \n\nIn the first talk\,
we shall first introduce the Poisson Boolean \nmodel and continuum percola
tion. Then\, we will show how\nsharp phase transition in the Poisson Boole
an model is derived via \nthe Poisson OSSS inequality. We shall also indic
ate the proof of the \nPoisson OSSS inequality. Time-permitting\, we will
mention \napplications to other models such as k-percolation and confetti
\npercolation.\n\nIn the second talk\, we will see how noise sensitivity a
nd \nexistence of exceptional times at criticality for crossing events in
\nthe dynamical Poisson Boolean model are derived via the Schramm-Steif \n
inequality for Poisson processes. We shall also discuss the \nSchramm-Stei
f inequality very briefly. \n\nExcept some of the basics on Poisson proce
ss and continuum percolation\, \nthe two talks will be somewhat independen
t.\n
LOCATION:https://stable.researchseminars.org/talk/BPS/14/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Siva Athreya (Indian Statistical Institute\, Bangalore)
DTSTART:20201209T090000Z
DTEND:20201209T094500Z
DTSTAMP:20260404T095121Z
UID:BPS/15
DESCRIPTION:Title: Sero-Survey in Karnataka State: Summary\, Design\, and Statistical M
ethodology\nby Siva Athreya (Indian Statistical Institute\, Bangalore)
as part of Bangalore Probability Seminar\n\n\nAbstract\nThe state of Karn
ataka conducted a survey to estimate the total COVID-19 burden in the stat
e between September 03-16\, 2020. The survey was unique in two main aspec
ts\, namely\, it jointly estimated both current and past infection in the
state and secondly the survey covered \nthe entire geographical region of
the state\, a first in India.\n\nWe shall discuss the design of the survey
\, provide a brief summary of the findings\, and discuss in detail the pro
bability models/statistical methodology used to obtain the estimates.\n
LOCATION:https://stable.researchseminars.org/talk/BPS/15/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Rajesh Sundaresan ((Indian Institute of Science\, Bangalore)
DTSTART:20201209T100000Z
DTEND:20201209T104500Z
DTSTAMP:20260404T095121Z
UID:BPS/16
DESCRIPTION:Title: Sero-Survey in Karnataka State: Summary\, Design\, and Statistical M
ethodology\nby Rajesh Sundaresan ((Indian Institute of Science\, Banga
lore) as part of Bangalore Probability Seminar\n\n\nAbstract\nThe state of
Karnataka conducted a survey to estimate the total COVID-19 burden in the
state between September 03-16\, 2020. The survey was unique in two main
aspects\, namely\, it jointly estimated both current and past infection in
the state and secondly the survey covered \nthe entire geographical regio
n of the state\, a first in India.\n\nWe shall discuss the design of the s
urvey\, provide a brief summary of the findings\, and discuss in detail th
e probability models/statistical methodology used to obtain the estimates.
\n
LOCATION:https://stable.researchseminars.org/talk/BPS/16/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Kerrie Mengersen (Queensland University of Technology\, Australia)
DTSTART:20210127T043000Z
DTEND:20210127T061500Z
DTSTAMP:20260404T095121Z
UID:BPS/17
DESCRIPTION:Title: Bayesian Modelling and Analysis of Challenging Data\nby Kerrie Men
gersen (Queensland University of Technology\, Australia) as part of Bangal
ore Probability Seminar\n\n\nAbstract\nPCM-2019-20\n\nhttps://www.isibang.
ac.in/~statmath/pcm2019/\n\nProfessor Kerrie Mengersen\nScience and Engine
ering Faculty\, School of Mathematical Sciences\nQueensland University of
Technology (QUT)\n(https://staff.qut.edu.au/staff/k.mengersen)\n\nwill del
iver the P.C. Mahalanobis Memorial Lectures 2019-20 on\n\nBayesian Modelli
ng and Analysis of Challenging Data\n\nDate: January 27th\, 2021\nTime: 11
:00-11:45am\n\nLecture 1: Bayesian Modelling and Analysis of Challenging D
ata\nIdentifying the Intrinsic Dimension of High-Dimensional Data\n\nAbstr
act: One of the challenges of high-dimensional data is\nidentifying the tr
ue information contained therein. In this\npresentation\, I will describe
some approaches that we have developed\nto address this challenge. These a
pproaches include Bayesian methods\nof matrix factorisation\, intrinsic di
mension and structured variable\nselection. This discussion will be set in
the context of substantive\ncase studies in image analysis\, sport and ge
nomics.\n\nDate: January 27th\, 2021\nTime: 12:00-12:45pm\n\nLecture 2: Ba
yesian Modelling and Analysis of Challenging Data Finding\nPatterns in Hig
hly Structured Spatio-Temporal Data\n\nAbstract: Many forms of current dat
a are indexed by space and/or\ntime. Often these space-time relationships
are highly\nstructured. Examples include data collected across networks of
\nstreams\, multi-state data collected over time\, and multivariate\nrespo
nses collected at small area level across a geographic space. In\nthis pre
sentation\, I will describe some of our approaches to Bayesian\nmodelling
and analysis of these types of data. Two particular issues\nwill be discus
sed. The first is the role of priors in spatial models\nand hidden Markov
models. The second is the detection of anomalies for\nevent identification
and to increase the trustworthiness of the data.\n\n\nThe lectures were o
riginally due to held in March 2020 but are now being\nheld online via Zoo
m platform.\n\nhttps://us02web.zoom.us/j/88118975864?pwd=cFRmWXMreWdiLzR5V
GlCYXBCbE1WZz09\n
LOCATION:https://stable.researchseminars.org/talk/BPS/17/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Kerrie Mengersen (Queensland University of Technology\, Australia)
DTSTART:20210129T043000Z
DTEND:20210129T061500Z
DTSTAMP:20260404T095121Z
UID:BPS/18
DESCRIPTION:Title: Bayesian Modelling and Analysis of Challenging Data\nby Kerrie Men
gersen (Queensland University of Technology\, Australia) as part of Bangal
ore Probability Seminar\n\n\nAbstract\nPCM-2019-20\n\nhttps://www.isibang.
ac.in/~statmath/pcm2019/\n\nProfessor Kerrie Mengersen\nScience and Engine
ering Faculty\, School of Mathematical Sciences\nQueensland University of
Technology (QUT)\n(https://staff.qut.edu.au/staff/k.mengersen)\n\nwill del
iver the P.C. Mahalanobis Memorial Lectures 2019-20 on\n\nBayesian Modelli
ng and Analysis of Challenging Data\n\nDate: January 29th\, 2021\nTime: 11
:00-11:45am\n\nLecture 3: Bayesian Modelling and Analysis of Challenging D
ata Describing\nSystems of Data\n\nAbstract: A common challenge is to anal
yse data as part of a\nsystem. In this presentation\, I describe a number
of Bayesian\napproaches to modelling such data\, using as motivation case
studies in\nneurology\, ecology and industry. These case studies focus on\
nunderstanding changes in the brain associated with a degenerative\ndiseas
e\, and suggesting optimal dredgingstrategies for conservation of\nseagras
s. The techniques employed include Bayesian wombling and\ndynamic Bayesian
networks.\n\nDate: January 29th\, 2021\nTime: 12:00-12:45pm\n\nLecture 4:
Bayesian Modelling and Analysis of Challenging Data Making New\nSources o
f Data Trustworthy\n\nAbstract: Evidence-based decisions depend critically
on trustworthy\ndata. Two forms of data that have been brought into quest
ion in recent\ntimes are sensor data and citizen science data. Sensors are
a key\ncomponent of IoT and have created a step-change in our ability to\
nmonitor systems. However\, they are often subject to technical\nanomalies
that raise concerns about the validity of their data and\nsignals. Citize
n science is also growing in utility and interest in\nmany areas\, but oft
en suffers from concerns about the credibility the\ninformation provided b
y community members. In this presentation\, I\nwill describe some new appr
oaches to resolving some of these\nconcerns. These include new methods for
anomaly detection in\nhigh-dimensional streaming time series\, and Bayesi
an models for\nestimating the latent ability of citizens taking into accou
nt the\ndifficulty of the tasks. This work has been developed in collabora
tion\nwith a number of teams working on challenges in ecology and industry
\;\nthese teams will be acknowledged and the associated challenges\ndiscus
sed during the presentation.\n\n\nThe lectures were originally due to held
in March 2020 but are now being\nheld online via Zoom platform.\n\nhttps:
//us02web.zoom.us/j/88118975864?pwd=cFRmWXMreWdiLzR5VGlCYXBCbE1WZz09\n
LOCATION:https://stable.researchseminars.org/talk/BPS/18/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Alon Nishry (Tel-Aviv University\, Israel)
DTSTART:20210201T090000Z
DTEND:20210201T094500Z
DTSTAMP:20260404T095121Z
UID:BPS/19
DESCRIPTION:Title: Gaussian complex zeros: conditional distribution on rare events\nb
y Alon Nishry (Tel-Aviv University\, Israel) as part of Bangalore Probabil
ity Seminar\n\n\nAbstract\nThe zero process of the Gaussian Entire Functio
n is a natural example of a two-dimensional random point configuration who
se distribution is invariant under rigid motions of the plane. Another wel
l-studied example is the infinite Ginibre ensemble. Due to non-trivial cor
relations\, the features of these two processes are quite different from t
he ones of the homogeneous Poisson point process. For this reason\, these
processes are of interest to analysts\, probabilists\, and mathematical ph
ysicists.\n\nSome particularly interesting statistical phenomena to study
are rare events\, when the number of points in a certain large domain is v
ery different from its expected value. An important example is the ‘hole
event’\, when there are no points at all. For the Ginibre ensemble\, th
e hole event can be studied using some classical tools from potential theo
ry. This is no longer true for complex zeros\, and in this talk I will men
tion some of the challenges and describe new results.\n\nBased on joint wo
rks with S. Ghosh (NUS) and A. Wennman (Tel Aviv)\n
LOCATION:https://stable.researchseminars.org/talk/BPS/19/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Aron Wenmann (Tel-Aviv University\, Israel)
DTSTART:20210201T100000Z
DTEND:20210201T104500Z
DTSTAMP:20260404T095121Z
UID:BPS/20
DESCRIPTION:Title: The forbidden region for random zeros: appearance of quadrature domain
s\nby Aron Wenmann (Tel-Aviv University\, Israel) as part of Bangalore
Probability Seminar\n\n\nAbstract\nWe consider the hole event for Gaussia
n complex zeros. As the size of the hole increases\, the density of zeros
vanishes\, not just inside the hole\, but also on a macroscopic region bey
ond the (rescaled) hole - a 'forbidden region' emerges. Surprisingly\, the
shape of this region is rigid under perturbations of the hole.\n\nI will
discuss the shape of the forbidden region for general simply connected hol
es. I plan to explain how to study the limiting zero density through a con
strained variational problem\, and touch upon a curious emergence of quadr
ature domains from potential theory.\n
LOCATION:https://stable.researchseminars.org/talk/BPS/20/
END:VEVENT
BEGIN:VEVENT
SUMMARY:M. E. Lakshmi Priya (Indian Institute of Science\, Bangalore)
DTSTART:20210215T090000Z
DTEND:20210215T094500Z
DTSTAMP:20260404T095121Z
UID:BPS/21
DESCRIPTION:Title: Overcrowding estimates for nodal volume of stationary Gaussian process
es (SGPs) on R^d\nby M. E. Lakshmi Priya (Indian Institute of Science\
, Bangalore) as part of Bangalore Probability Seminar\n\n\nAbstract\nWe co
nsider centered SGPs on Euclidean spaces R^d and study their nodal volume
in [0\,T]^d\, for T>0. From earlier studies\, we know the following statis
tics for nodal volume of SGPs under varying assumptions on their spectral
measures: expectation\, variance asymptotics\, CLT\, exponential concentra
tion (only for d=1)\, and finiteness of moments. \n\nWe study the unlikely
event of overcrowding of the nodal set in [0\,T]^d\; this is the event th
at the volume of the nodal set in [0\,T]^d is much larger than its expecte
d value. Under some mild assumptions on the spectral measure\, we obtain e
stimates for the overcrowding event's probability. We first get overcrowdi
ng estimates for the zero count of SGPs on R. In higher dimensions\, we co
nsider Crofton's formula\, which gives the volume of the nodal set in term
s of the number of intersections of the nodal set with all lines in R^d. W
e discretize this formula to get a more workable version of it and\, in a
sense\, reduce the overcrowding problem in higher dimensions to the one-di
mensional case.\n
LOCATION:https://stable.researchseminars.org/talk/BPS/21/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Subhajit Ghosh (Indian Institute of Science\, Bangalore)
DTSTART:20210215T100000Z
DTEND:20210215T104500Z
DTSTAMP:20260404T095121Z
UID:BPS/22
DESCRIPTION:Title: Total variation cutoff for the flip-transpose top with random shuffle<
/a>\nby Subhajit Ghosh (Indian Institute of Science\, Bangalore) as part o
f Bangalore Probability Seminar\n\n\nAbstract\nWe consider a random walk o
n the hyperoctahedral group $B_n$ generated by the signed permutations of
the form $(i\,n)$ and $(-i\,n)$ for $1\\leq i\\leq n$. We call this the \\
emph{flip-transpose top with random shuffle} on $B_n$. We find the spectru
m of the transition probability matrix for this shuffle. We prove that thi
s shuffle exhibits the total variation cutoff phenomenon with cutoff time
$n\\log n$. We also show that a biased variant of this shuffle exhibits th
e total variation cutoff with the same cutoff time.\n
LOCATION:https://stable.researchseminars.org/talk/BPS/22/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Vladimir Vovk (University of London\, U.K.)
DTSTART:20210315T093000Z
DTEND:20210315T101500Z
DTSTAMP:20260404T095121Z
UID:BPS/23
DESCRIPTION:Title: Martingales in the foundations of statistics\nby Vladimir Vovk (Un
iversity of London\, U.K.) as part of Bangalore Probability Seminar\n\n\nA
bstract\nhttps://www.isibang.ac.in/~statmath/pcm2020/\n\nMartingales in th
e foundations of statistics\nDate: March 15th\, 2021\nTime: 15:00-15:45 IS
T\n\n\nAbstract: Traditional methods of testing statistical hypotheses hav
e been developed for the batch setting\, in the terminology of machine lea
rning: given a batch of data\, statisticians typically compute measures of
disagreement\, such as p-values or Bayes factors\, between a null hypothe
sis and the data. An alternative that is popular in machine learning is th
e online setting\, in which the items of data (observations ) keep arrivin
g sequentially. In this introductory lecture I will explain the role of ma
rtingales\, in the form of test martingales\, in online hypothesis testing
and discuss their applications in the foundations of probability and stat
istics.\n\nMultiple hypothesis testing with e-values\nDate: March 15th\, 2
021\nTime: 16:00-16:45 IST\n\nAbstract: It is interesting that test martin
gales do not trivialize in the case of only one observation. In fact\, the
y provide a useful alternative\, sometimes called e-values\, to the standa
rd statistical notion of p-values. The most important mathematical advanta
ge of e-values over p-values is that the average of e-values is always an
e-value. This property is valuable in multiple hypothesis testing\, which
will be the topic of this lecture.\n\nConformal prediction\nDate: March 19
th\, 2021\nTime: 15:00-15:45 IST\n\nAbstract: Mainstream machine learning\
, despite its recent successes\, has a serious drawback: while its state-o
f-the-art algorithms often produce excellent predictions\, they do not pro
vide measures of their accuracy and reliability that would be both practic
ally useful and provably valid. On the other hand\, such measures are comm
onplace in statistics. Conformal prediction adapts rank tests\, popular in
nonparametric statistics\, to testing the IID assumption (the observation
s being independent and identically distributed)\, which is the standard a
ssumption made in machine learning. This gives us practical measures\, pro
vably valid under the IID assumption\, of the accuracy and reliability of
predictions produced by traditional and recent machine-learning algorithms
. In this lecture I will give a brief review of conformal prediction.\n\nC
onformal hypothesis testing\nDate: March 19th\, 2021\nTime: 16:00-16:45 IS
T\n\nAbstract: An interesting application of conformal prediction is the e
xistence of exchangeability martingales\, i.e.\, random processes that are
test martingales under any exchangeable probability measure. In particula
r\, they are martingales whenever the observations are IID. The topics of
this last lecture in this series will be the construction of exchangeabili
ty martingales and their use for different kinds of change detection\, inc
luding detecting a point at which the IID assumption becomes violated and
detecting concept shift.\n
LOCATION:https://stable.researchseminars.org/talk/BPS/23/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Vladimir Vovk (University of London\, U.K.)
DTSTART:20210315T103000Z
DTEND:20210315T111500Z
DTSTAMP:20260404T095121Z
UID:BPS/24
DESCRIPTION:Title: Multiple hypothesis testing with e-values\nby Vladimir Vovk (Unive
rsity of London\, U.K.) as part of Bangalore Probability Seminar\n\n\nAbst
ract\nhttps://www.isibang.ac.in/~statmath/pcm2020/\n\nMartingales in the f
oundations of statistics\nDate: March 15th\, 2021\nTime: 15:00-15:45 IST\n
\n\nAbstract: Traditional methods of testing statistical hypotheses have b
een developed for the batch setting\, in the terminology of machine learni
ng: given a batch of data\, statisticians typically compute measures of di
sagreement\, such as p-values or Bayes factors\, between a null hypothesis
and the data. An alternative that is popular in machine learning is the o
nline setting\, in which the items of data (observations ) keep arriving s
equentially. In this introductory lecture I will explain the role of marti
ngales\, in the form of test martingales\, in online hypothesis testing an
d discuss their applications in the foundations of probability and statist
ics.\n\nMultiple hypothesis testing with e-values\nDate: March 15th\, 2021
\nTime: 16:00-16:45 IST\n\nAbstract: It is interesting that test martingal
es do not trivialize in the case of only one observation. In fact\, they p
rovide a useful alternative\, sometimes called e-values\, to the standard
statistical notion of p-values. The most important mathematical advantage
of e-values over p-values is that the average of e-values is always an e-v
alue. This property is valuable in multiple hypothesis testing\, which wil
l be the topic of this lecture.\n\nConformal prediction\nDate: March 19th\
, 2021\nTime: 15:00-15:45 IST\n\nAbstract: Mainstream machine learning\, d
espite its recent successes\, has a serious drawback: while its state-of-t
he-art algorithms often produce excellent predictions\, they do not provid
e measures of their accuracy and reliability that would be both practicall
y useful and provably valid. On the other hand\, such measures are commonp
lace in statistics. Conformal prediction adapts rank tests\, popular in no
nparametric statistics\, to testing the IID assumption (the observations b
eing independent and identically distributed)\, which is the standard assu
mption made in machine learning. This gives us practical measures\, provab
ly valid under the IID assumption\, of the accuracy and reliability of pre
dictions produced by traditional and recent machine-learning algorithms. I
n this lecture I will give a brief review of conformal prediction.\nConfor
mal hypothesis testing\nDate: March 19th\, 2021\nTime: 16:00-16:45 IST\n\n
Abstract: An interesting application of conformal prediction is the existe
nce of exchangeability martingales\, i.e.\, random processes that are test
martingales under any exchangeable probability measure. In particular\, t
hey are martingales whenever the observations are IID. The topics of this
last lecture in this series will be the construction of exchangeability ma
rtingales and their use for different kinds of change detection\, includin
g detecting a point at which the IID assumption becomes violated and detec
ting concept shift.\n
LOCATION:https://stable.researchseminars.org/talk/BPS/24/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Vladimir Vovk (University of London\, U.K.)
DTSTART:20210319T093000Z
DTEND:20210319T101500Z
DTSTAMP:20260404T095121Z
UID:BPS/25
DESCRIPTION:Title: Conformal prediction\nby Vladimir Vovk (University of London\, U.K
.) as part of Bangalore Probability Seminar\n\n\nAbstract\nhttps://www.isi
bang.ac.in/~statmath/pcm2020/\n\nMartingales in the foundations of statist
ics\nDate: March 15th\, 2021\nTime: 15:00-15:45 IST\n\n\nAbstract: Traditi
onal methods of testing statistical hypotheses have been developed for the
batch setting\, in the terminology of machine learning: given a batch of
data\, statisticians typically compute measures of disagreement\, such as
p-values or Bayes factors\, between a null hypothesis and the data. An alt
ernative that is popular in machine learning is the online setting\, in wh
ich the items of data (observations ) keep arriving sequentially. In this
introductory lecture I will explain the role of martingales\, in the form
of test martingales\, in online hypothesis testing and discuss their appli
cations in the foundations of probability and statistics.\n\nMultiple hypo
thesis testing with e-values\nDate: March 15th\, 2021\nTime: 16:00-16:45 I
ST\n\nAbstract: It is interesting that test martingales do not trivialize
in the case of only one observation. In fact\, they provide a useful alter
native\, sometimes called e-values\, to the standard statistical notion of
p-values. The most important mathematical advantage of e-values over p-va
lues is that the average of e-values is always an e-value. This property i
s valuable in multiple hypothesis testing\, which will be the topic of thi
s lecture.\n\nConformal prediction\nDate: March 19th\, 2021\nTime: 15:00-1
5:45 IST\n\nAbstract: Mainstream machine learning\, despite its recent suc
cesses\, has a serious drawback: while its state-of-the-art algorithms oft
en produce excellent predictions\, they do not provide measures of their a
ccuracy and reliability that would be both practically useful and provably
valid. On the other hand\, such measures are commonplace in statistics. C
onformal prediction adapts rank tests\, popular in nonparametric statistic
s\, to testing the IID assumption (the observations being independent and
identically distributed)\, which is the standard assumption made in machin
e learning. This gives us practical measures\, provably valid under the II
D assumption\, of the accuracy and reliability of predictions produced by
traditional and recent machine-learning algorithms. In this lecture I will
give a brief review of conformal prediction.\nConformal hypothesis testin
g\nDate: March 19th\, 2021\nTime: 16:00-16:45 IST\n\nAbstract: An interest
ing application of conformal prediction is the existence of exchangeabilit
y martingales\, i.e.\, random processes that are test martingales under an
y exchangeable probability measure. In particular\, they are martingales w
henever the observations are IID. The topics of this last lecture in this
series will be the construction of exchangeability martingales and their u
se for different kinds of change detection\, including detecting a point a
t which the IID assumption becomes violated and detecting concept shift.\n
LOCATION:https://stable.researchseminars.org/talk/BPS/25/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Vladimir Vovk (University of London\, U.K.)
DTSTART:20210319T103000Z
DTEND:20210319T111500Z
DTSTAMP:20260404T095121Z
UID:BPS/26
DESCRIPTION:Title: Conformal hypothesis testing\nby Vladimir Vovk (University of Lond
on\, U.K.) as part of Bangalore Probability Seminar\n\n\nAbstract\nhttps:/
/www.isibang.ac.in/~statmath/pcm2020/\n\nMartingales in the foundations of
statistics\nDate: March 15th\, 2021\nTime: 15:00-15:45 IST\n\n\nAbstract:
Traditional methods of testing statistical hypotheses have been developed
for the batch setting\, in the terminology of machine learning: given a b
atch of data\, statisticians typically compute measures of disagreement\,
such as p-values or Bayes factors\, between a null hypothesis and the data
. An alternative that is popular in machine learning is the online setting
\, in which the items of data (observations ) keep arriving sequentially.
In this introductory lecture I will explain the role of martingales\, in t
he form of test martingales\, in online hypothesis testing and discuss the
ir applications in the foundations of probability and statistics.\n\nMulti
ple hypothesis testing with e-values\nDate: March 15th\, 2021\nTime: 16:00
-16:45 IST\n\nAbstract: It is interesting that test martingales do not tri
vialize in the case of only one observation. In fact\, they provide a usef
ul alternative\, sometimes called e-values\, to the standard statistical n
otion of p-values. The most important mathematical advantage of e-values o
ver p-values is that the average of e-values is always an e-value. This pr
operty is valuable in multiple hypothesis testing\, which will be the topi
c of this lecture.\n\nConformal prediction\nDate: March 19th\, 2021\nTime:
15:00-15:45 IST\n\nAbstract: Mainstream machine learning\, despite its re
cent successes\, has a serious drawback: while its state-of-the-art algori
thms often produce excellent predictions\, they do not provide measures of
their accuracy and reliability that would be both practically useful and
provably valid. On the other hand\, such measures are commonplace in stati
stics. Conformal prediction adapts rank tests\, popular in nonparametric s
tatistics\, to testing the IID assumption (the observations being independ
ent and identically distributed)\, which is the standard assumption made i
n machine learning. This gives us practical measures\, provably valid unde
r the IID assumption\, of the accuracy and reliability of predictions prod
uced by traditional and recent machine-learning algorithms. In this lectur
e I will give a brief review of conformal prediction.\nConformal hypothesi
s testing\nDate: March 19th\, 2021\nTime: 16:00-16:45 IST\n\nAbstract: An
interesting application of conformal prediction is the existence of exchan
geability martingales\, i.e.\, random processes that are test martingales
under any exchangeable probability measure. In particular\, they are marti
ngales whenever the observations are IID. The topics of this last lecture
in this series will be the construction of exchangeability martingales and
their use for different kinds of change detection\, including detecting a
point at which the IID assumption becomes violated and detecting concept
shift.\n
LOCATION:https://stable.researchseminars.org/talk/BPS/26/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Martin Barlow (University of British Columbia\, Canada)
DTSTART:20210316T093000Z
DTEND:20210316T111500Z
DTSTAMP:20260404T095121Z
UID:BPS/27
DESCRIPTION:Title: Harnack inequalities - from PDE to random graphs\nby Martin Barlow
(University of British Columbia\, Canada) as part of Bangalore Probabilit
y Seminar\n\n\nAbstract\nhttps://www.isibang.ac.in/~statmath/amml20/\n\nA
major problem in PDE\, solved independently in the late 1950s by de Gi
orgi\, Moser and Nash\, was the regularity of solutions to second order di
vergence form equations. Moser's approach was to use a Harnack inequal
ity\, which can be understood in probabilistic terms.a \n\n These PD
E methods are very versatile\, and have been extended to manifolds\, g
eneral metric spaces\, and graphs. They give continuity results for harmon
ic functions\, and lead to estimates of the transition density of Markov p
rocesses. Of particular importance is the fact that the PDE approach
is robust enough to able to handle small local perturbations of the space
. \n \n\n These lectures will review this progress\, and will present
applications of Harnack inequalities to random walks on random graphs.
\n\n \n\nDate: March 16th\, 2021\n\n Lecture 1 \n
Time: 15:00-15:45\, IST \n \n Lecture 2
\n Time: 16:00-16:45\, IST \n \nDate: March 18
th\, 2021\n\n Lecture 1 \n Time: 15:00-15:45\,
IST \n \n Lecture 2 \n Time: 16:
00-16:45\, IST\n
LOCATION:https://stable.researchseminars.org/talk/BPS/27/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Martin Barlow (University of British Columbia\, Canada)
DTSTART:20210318T093000Z
DTEND:20210318T101500Z
DTSTAMP:20260404T095121Z
UID:BPS/29
DESCRIPTION:Title: Harnack inequalities - from PDE to random graphs\nby Martin Barlow
(University of British Columbia\, Canada) as part of Bangalore Probabilit
y Seminar\n\n\nAbstract\nhttps://www.isibang.ac.in/~statmath/amml20/\n\nA
major problem in PDE\, solved independently in the late 1950s by de Gi
orgi\, Moser and Nash\, was the regularity of solutions to second order di
vergence form equations. Moser's approach was to use a Harnack inequal
ity\, which can be understood in probabilistic terms.a \n\n These PD
E methods are very versatile\, and have been extended to manifolds\, g
eneral metric spaces\, and graphs. They give continuity results for harmon
ic functions\, and lead to estimates of the transition density of Markov p
rocesses. Of particular importance is the fact that the PDE approach
is robust enough to able to handle small local perturbations of the space
. \n \n\n These lectures will review this progress\, and will present
applications of Harnack inequalities to random walks on random graphs.
\n\n \n\nDate: March 16th\, 2021\n\n Lecture 1 \n
Time: 15:00-15:45\, IST \n \n Lecture 2
\n Time: 16:00-16:45\, IST \n \nDate: March 18
th\, 2021\n\n Lecture 1 \n Time: 15:00-15:45\,
IST \n \n Lecture 2 \n Time: 16:
00-16:45\, IST\n
LOCATION:https://stable.researchseminars.org/talk/BPS/29/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Justin Salez (Univ. Paris-Dauphine\, Paris)
DTSTART:20210412T090000Z
DTEND:20210412T110000Z
DTSTAMP:20260404T095121Z
UID:BPS/30
DESCRIPTION:Title: An entropic interpretation of the cutoff phenomenon for finite Markov
chains\nby Justin Salez (Univ. Paris-Dauphine\, Paris) as part of Bang
alore Probability Seminar\n\n\nAbstract\nThe cutoff phenomenon is a sharp
phase transition in the convergence to equilibrium of certain Markov chain
s. Discovered by Aldous\, Diaconis and Shahshahani in the context of card
shuffling\, it has since then been established on a variety of examples.
However\, proving cutoff remains a delicate affair\, which requires a very
detailed knowledge of the chain. Identifying the general mechanisms under
lying this phase transition\, without having to pinpoint its precise locat
ion\, remains one of the most fundamental open problems in the area of mix
ing times.\n\nIn the first part of the lecture\, I will provide a self-con
tained introduction to this beautiful question\, and present some classica
l examples. In the second part\, I will discuss a recent interpretation of
the cutoff phenomenon in terms of concentration of entropy\, and show how
this can be used to deduce cutoff for a broad class of Markov chains with
non-negative curvature\, including random walks on abelian Cayley expande
rs.\n
LOCATION:https://stable.researchseminars.org/talk/BPS/30/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Mohammud Foondun (Univ. Strathclyde\, U. K.)
DTSTART:20210419T090000Z
DTEND:20210419T094500Z
DTSTAMP:20260404T095121Z
UID:BPS/31
DESCRIPTION:by Mohammud Foondun (Univ. Strathclyde\, U. K.) as part of Ban
galore Probability Seminar\n\nAbstract: TBA\n
LOCATION:https://stable.researchseminars.org/talk/BPS/31/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Kunwoo Kim (POSTECH\, Korea)
DTSTART:20210419T100000Z
DTEND:20210419T104500Z
DTSTAMP:20260404T095121Z
UID:BPS/32
DESCRIPTION:Title: Phase Analysis for a family of Stochastic Reaction-Diffusion Equations
\nby Kunwoo Kim (POSTECH\, Korea) as part of Bangalore Probability Sem
inar\n\n\nAbstract\nWe consider a family of stochastic reaction-diffusion
equations driven by space-time white noise. The reaction term belongs to a
large family of functions that includes Fisher-KPP nonlinearities [V (x)
= x(1 − x)] as well as Allen-Cahn potentials [V (x) = x(1 − x)(1 + x)]
. We show that (i) if the noise intensity is large\, our stochastic PDE h
as the unique invariant measure\; and (ii) if the noise intensity is small
\, the collection of all invariant measures is a non-trivial line segment\
, in particular infinite. Our methods also say a great deal about the stru
cture of these invariant measures. This is based on joint work with Davar
Khoshnevisan\, Carl Mueller and Shang-Yuan Shiu.\n
LOCATION:https://stable.researchseminars.org/talk/BPS/32/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Gugan Thoppe (Indian Institute of Science\, Bangalore)
DTSTART:20210503T090000Z
DTEND:20210503T094500Z
DTSTAMP:20260404T095121Z
UID:BPS/33
DESCRIPTION:Title: Active Cases and Disease Extinction in the Stochastic SIR Model: Estim
ates with Probabilistic Guarantees\nby Gugan Thoppe (Indian Institute
of Science\, Bangalore) as part of Bangalore Probability Seminar\n\n\nAbst
ract\nSIR models\, both deterministic and stochastic\, provide a viable se
tup for studying epidemics. While the deterministic ones have been around
for almost a century now\, it hasn't still been possible to obtain analyti
cal estimates for active infections in these setups. Also\, these are not
well-suited to answer questions relating to early termination. The stochas
tic variants\, on the other hand\, have indeed been amenable to such analy
ses. However\, the current approaches are too complex\; they involve using
different approximations (by branching processes\, ODEs\, etc.) for diffe
rent parts of the process.\n\nIn this work\, we consider a discrete-time s
tochastic SIR model and take a fundamentally different route to overcome t
he known challenges in analyzing SIR models. Namely\, our proofs rely on a
sequence of stopping times based on jumps in the susceptible population.
Their main advantage is that the number of recoveries between two successi
ve stopping times is then a truncated geometric random variable. Our main
results include probabilistic bounds for the number of active infections a
nd the disease extinction time. We also obtain an estimate for the expec
ted value of the largest epidemic size. This bound matches its analogue
in the deterministic case asymptotically. \n\nThis is ongoing work with D
r. Gal Dalal (nVidia\, Israel) and Dr. Balazs Szorenyi (Yahoo Research\, U
SA).\n
LOCATION:https://stable.researchseminars.org/talk/BPS/33/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Sarath Yasodharan (Indian Institute of Science\, Bangalore)
DTSTART:20210503T100000Z
DTEND:20210503T104500Z
DTSTAMP:20260404T095121Z
UID:BPS/34
DESCRIPTION:Title: Large deviations of mean-field interacting particle systems in a fast
varying environment\nby Sarath Yasodharan (Indian Institute of Science
\, Bangalore) as part of Bangalore Probability Seminar\n\n\nAbstract\nWe c
onsider a weakly interacting Markovian mean-field particle system in a fas
t varying environment. The particles evolve in the slow time scale and the
environment process evolves in the fast time scale. The system is ``fully
coupled”\, i.e.\, the evolution of the particles depend on the state of
the environment\, and the environment itself changes its state depending
on the empirical measure of the system of particles. For this two time sca
le mean-field model\, we prove a process-level large deviation principle f
or the joint law of the empirical measure process of the particles and the
occupation measure process of the fast environment. This extends previous
results known for two time scale diffusions to two time scale mean-field
models with jumps. Our proof is based on the method of stochastic exponent
ials. We characterise the rate function by studying a certain variational
problem associated with an exponential martingale.\n\nThis talk is based o
n joint work with Rajesh Sundaresan.\n
LOCATION:https://stable.researchseminars.org/talk/BPS/34/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Sayantan Chakraborty (TIFR\, Mumbai)
DTSTART:20210906T090000Z
DTEND:20210906T110000Z
DTSTAMP:20260404T095121Z
UID:BPS/35
DESCRIPTION:Title: Improved Bounds for Perfect Sampling of k-Colorings in Graphs.\nby
Sayantan Chakraborty (TIFR\, Mumbai) as part of Bangalore Probability Sem
inar\n\n\nAbstract\nI this talk we will present a randomized algorithm tha
t takes as input an undirected n-vertex graph G with maximum degree Δ and
an integer k>3Δ\, and returns a random proper k-coloring of G. The distr
ibution of the coloring is \\emph{perfectly} uniform over the set of all p
roper k-colorings\; the expected running time of the algorithm is poly(k\,
n)=O(nΔ^2⋅log(k)). This improves upon a result of Huber~(STOC 1998) who
obtained a polynomial time perfect sampling algorithm for k>Δ^2+2Δ. Pri
or to our work\, no algorithm with expected running time poly(k\,n) was kn
own to guarantee perfectly sampling with sub-quadratic number of colors in
general. Our algorithm (like several other perfect sampling algorithms in
cluding Huber's) is based on the Coupling from the Past method. Inspired b
y the \\emph{bounding chain} approach\, pioneered independently by Huber~(
STOC 1998) and Häggström & Nelander~(Scand. J. Statist.\, 1999)\, we emp
loy a novel bounding chain to derive our result for the graph coloring pro
blem.\n
LOCATION:https://stable.researchseminars.org/talk/BPS/35/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Sarah Penington (Univ. Bath\, U. K.)
DTSTART:20211018T090000Z
DTEND:20211018T110000Z
DTSTAMP:20260404T095121Z
UID:BPS/36
DESCRIPTION:Title: Genealogy of the N-particle branching random walk with polynomial tail
s\nby Sarah Penington (Univ. Bath\, U. K.) as part of Bangalore Probab
ility Seminar\n\n\nAbstract\nThe N-particle branching random walk is a dis
crete time branching particle system with selection consisting of N partic
les located on the real line. At every time step\, each particle is replac
ed by two offspring\, and each offspring particle makes a jump from its pa
rent's location\, independently from the other jumps\, according to a give
n jump distribution. Then only the N rightmost particles survive\; the oth
er particles are removed from the system to keep the population size const
ant.\nI will discuss recent results and open conjectures about the long-te
rm behaviour of this particle system when N\, the number of particles\, is
large. In the case where the jump distribution has regularly varying tail
s\, building on earlier work of J. Bérard and P. Maillard\, we prove that
at a typical large time the genealogy of the population is given by a sta
r-shaped coalescent\, and that almost the whole population is near the lef
tmost particle on the relevant space scale.\n\nBased on joint work with Ma
tt Roberts and Zsófia Talyigás.\n
LOCATION:https://stable.researchseminars.org/talk/BPS/36/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Riddhipratim Basu (ICTS\, Bengaluru)
DTSTART:20211108T090000Z
DTEND:20211108T094500Z
DTSTAMP:20260404T095121Z
UID:BPS/37
DESCRIPTION:Title: A non-technical introduction to the Liouville quantum gravity metric\nby Riddhipratim Basu (ICTS\, Bengaluru) as part of Bangalore Probabil
ity Seminar\n\n\nAbstract\nInformally speaking\, the Liouville quantum gra
vity (LQG) metric is a random Riemannian metric on the complex plane given
by the metric tensor $\\exp(2\\xi h) (dx^2+dy^2)$ where \\xi is a paramet
er and h is an appropriate version of the Gaussian free field on $\\mathbb
{C}$. In this talk\, we shall recall basic definitions and review some of
the recent developments on the construction of this metric and understandi
ng of its basic properties. I shall ignore most technical aspects and focu
s on geometric results that are particularly interesting from a first pass
age percolation perspective.\n
LOCATION:https://stable.researchseminars.org/talk/BPS/37/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Balazs Rath (Budapest University of Technology\, Hungary)
DTSTART:20211004T090000Z
DTEND:20211004T094500Z
DTSTAMP:20260404T095121Z
UID:BPS/38
DESCRIPTION:Title: Percolation of worms-I\nby Balazs Rath (Budapest University of Tec
hnology\, Hungary) as part of Bangalore Probability Seminar\n\nAbstract: T
BA\n
LOCATION:https://stable.researchseminars.org/talk/BPS/38/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Sándor Rokob (Budapest University of Technology\, Hungary)
DTSTART:20211004T100000Z
DTEND:20211004T104500Z
DTSTAMP:20260404T095121Z
UID:BPS/39
DESCRIPTION:Title: Percolation of worms-II\nby Sándor Rokob (Budapest University of
Technology\, Hungary) as part of Bangalore Probability Seminar\n\nAbstract
: TBA\n
LOCATION:https://stable.researchseminars.org/talk/BPS/39/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Sumit Mukerjee (Columbia University)
DTSTART:20211122T090000Z
DTEND:20211122T110000Z
DTSTAMP:20260404T095121Z
UID:BPS/40
DESCRIPTION:by Sumit Mukerjee (Columbia University) as part of Bangalore P
robability Seminar\n\nAbstract: TBA\n
LOCATION:https://stable.researchseminars.org/talk/BPS/40/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Riddhipratim Basu (ICTS\, Bengaluru)
DTSTART:20211108T100000Z
DTEND:20211108T104500Z
DTSTAMP:20260404T095121Z
UID:BPS/41
DESCRIPTION:Title: Environment seen from infinite geodesics in Liouville quantum gravity<
/a>\nby Riddhipratim Basu (ICTS\, Bengaluru) as part of Bangalore Probabi
lity Seminar\n\n\nAbstract\nIn this second part\, I shall talk about a rec
ent result (joint work with Manan Bhatia and Shirshendu Ganguly) on enviro
nments seen from infinite geodesics in the LQG metric. The question whethe
r the underlying noise seen from an appropriately chosen random point on a
n infinite geodesic originated in the context of planar first and last pas
sage percolation models. We shall briefly review some recent progress in t
hose models before turning our attention to a natural formulation of the q
uestion in the context of LQG. We shall discuss results establishing the e
xistence of a limiting environment and describing some basic properties of
the limiting environment.\n
LOCATION:https://stable.researchseminars.org/talk/BPS/41/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Elena Pulvirenti (TU Delft\, Netherlands)
DTSTART:20211206T090000Z
DTEND:20211206T094500Z
DTSTAMP:20260404T095121Z
UID:BPS/42
DESCRIPTION:Title: Metastability for the dilute Curie-Weiss model with Glauber dynamics\nby Elena Pulvirenti (TU Delft\, Netherlands) as part of Bangalore Prob
ability Seminar\n\n\nAbstract\nSystems subject to a random dynamics exhibi
t metastability when they persist for a very long time in a phase (called
metastable state) that is different from the one corresponding to the ther
modynamic equilibrium (called stable state). \n\nIn this talk I will analy
se the metastable behaviour of the dilute Curie–Weiss model\, a classica
l model of a disordered ferromagnet\, subject to a Glauber dynamics. The e
quilibrium model is a random version of a mean-field Ising model\, where t
he coupling coefficients are replaced by i.i.d. random coefficients\, e.g.
Bernoulli random variables with fixed parameter p. This model can be also
viewed as an Ising model on the Erdos–Renyi random graph with edge prob
ability p.\n\nUnder the Glauber dynamics the system is a Markov chain wher
e spins flip according to a Metropolis dynamics at inverse temperature \\b
eta. \n\nI will show how to compute the average time the system takes to r
each the stable phase when it starts from a certain probability distributi
on on the metastable state (called the last-exit biased distribution)\, in
the regime where the system size goes to infinity\, \\beta is larger than
1 and the magnetic field is positive and small enough. I will explain how
to obtain asymptotic bounds on the probability of the event that the mean
metastable hitting time is approximated by that of the Curie–Weiss mode
l.\n\nThe proof uses the potential theoretic approach to metastability and
concentration\n\nof measure inequalities. This is a joint collaboration w
ith Anton Bovier (Bonn) and \n\nSaeda Marello (Bonn).\n
LOCATION:https://stable.researchseminars.org/talk/BPS/42/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Saeda Marello (Univ. Bonn\, Germany)
DTSTART:20211206T100000Z
DTEND:20211206T104500Z
DTSTAMP:20260404T095121Z
UID:BPS/43
DESCRIPTION:Title: Metastability for the Curie-Weiss model on inhomogeneous random graphs
: results and challenges\nby Saeda Marello (Univ. Bonn\, Germany) as p
art of Bangalore Probability Seminar\n\n\nAbstract\nWe are currently tryin
g to extend to inhomogeneous random graphs the results on metastability fo
r the Curie-Weiss model (CW) on the Erdős–Rényi random graph (namely t
he randomly dilute CW)\, obtained by Anton Bovier\, Elena Pulvirenti and m
yself\, and presented in the previous talk.\n\nThe idea is the same: obtai
ning information on a target model in terms of the correspondent “mean m
odel” quantities.\n\nAfter presenting few general results\, we will focu
s on a particular case: the CW on the Chung-Lu inhomogeneous random graph
and “its mean model”\, the so called “CW with disorder”. The latte
r model will be the core of the talk: we will present results\, techniques
and comparison with known models. This will allow us to see why many deta
ils can be obtained there and to point of the challenges we face in other
models.\n\nWe use extensively the potential-theoretic approach to metastab
ility.\n\nBased on ongoing joint work with Anton Bovier and Frank den Holl
ander.\n
LOCATION:https://stable.researchseminars.org/talk/BPS/43/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Domenico Marinucci (University of Rome “Tor Vergata”\, Italy)
DTSTART:20220124T090000Z
DTEND:20220124T094500Z
DTSTAMP:20260404T095121Z
UID:BPS/44
DESCRIPTION:by Domenico Marinucci (University of Rome “Tor Vergata”\,
Italy) as part of Bangalore Probability Seminar\n\nAbstract: TBA\n
LOCATION:https://stable.researchseminars.org/talk/BPS/44/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Maurizia Rossi (University of Milano-Bicocca\, Italy)
DTSTART:20220124T100000Z
DTEND:20220124T104500Z
DTSTAMP:20260404T095121Z
UID:BPS/45
DESCRIPTION:by Maurizia Rossi (University of Milano-Bicocca\, Italy) as pa
rt of Bangalore Probability Seminar\n\nAbstract: TBA\n
LOCATION:https://stable.researchseminars.org/talk/BPS/45/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Alison Etheridge (University of Oxford\, U. K.)
DTSTART:20220314T090000Z
DTEND:20220314T110000Z
DTSTAMP:20260404T095121Z
UID:BPS/46
DESCRIPTION:by Alison Etheridge (University of Oxford\, U. K.) as part of
Bangalore Probability Seminar\n\nAbstract: TBA\n
LOCATION:https://stable.researchseminars.org/talk/BPS/46/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Alison Etheridge (University of Oxford\, U. K.)
DTSTART:20220316T090000Z
DTEND:20220316T110000Z
DTSTAMP:20260404T095121Z
UID:BPS/47
DESCRIPTION:by Alison Etheridge (University of Oxford\, U. K.) as part of
Bangalore Probability Seminar\n\nAbstract: TBA\n
LOCATION:https://stable.researchseminars.org/talk/BPS/47/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Soumendu Sunder Mukherjee (Indian Statistical Institute\, Kolkata)
DTSTART:20220207T090000Z
DTEND:20220207T094500Z
DTSTAMP:20260404T095121Z
UID:BPS/48
DESCRIPTION:by Soumendu Sunder Mukherjee (Indian Statistical Institute\, K
olkata) as part of Bangalore Probability Seminar\n\nAbstract: TBA\n
LOCATION:https://stable.researchseminars.org/talk/BPS/48/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Eliran Subag (Weizmann Institute of Science)
DTSTART:20220221T090000Z
DTEND:20220221T110000Z
DTSTAMP:20260404T095121Z
UID:BPS/49
DESCRIPTION:Title: The free energy of pure spherical models: computation from the TAP app
roach\nby Eliran Subag (Weizmann Institute of Science) as part of Bang
alore Probability Seminar\n\nAbstract: TBA\n
LOCATION:https://stable.researchseminars.org/talk/BPS/49/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Kesav Krishnan (Univ. Illinois Urbana-Champaign)
DTSTART:20220207T100000Z
DTEND:20220207T104500Z
DTSTAMP:20260404T095121Z
UID:BPS/50
DESCRIPTION:Title: Disordered Monomer Dimer Models on Cylinder Graphs\nby Kesav Krish
nan (Univ. Illinois Urbana-Champaign) as part of Bangalore Probability Sem
inar\n\nAbstract: TBA\n
LOCATION:https://stable.researchseminars.org/talk/BPS/50/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Antti Knowles (University of Geneva)
DTSTART:20220411T090000Z
DTEND:20220411T110000Z
DTSTAMP:20260404T095121Z
UID:BPS/51
DESCRIPTION:Title: Spectral phases of Erdös-Rényi graphs\nby Antti Knowles (Univers
ity of Geneva) as part of Bangalore Probability Seminar\n\nAbstract: TBA\
n
LOCATION:https://stable.researchseminars.org/talk/BPS/51/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Ph.D. Students / Postdocs
DTSTART:20220425T090000Z
DTEND:20220425T110000Z
DTSTAMP:20260404T095121Z
UID:BPS/52
DESCRIPTION:Title: Short-talks\nby Ph.D. Students / Postdocs as part of Bangalore Pro
bability Seminar\n\nAbstract: TBA\n
LOCATION:https://stable.researchseminars.org/talk/BPS/52/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Ph.D. Students / Postdocs
DTSTART:20220502T090000Z
DTEND:20220502T110000Z
DTSTAMP:20260404T095121Z
UID:BPS/53
DESCRIPTION:Title: Short-talks\nby Ph.D. Students / Postdocs as part of Bangalore Pro
bability Seminar\n\nAbstract: TBA\n
LOCATION:https://stable.researchseminars.org/talk/BPS/53/
END:VEVENT
END:VCALENDAR