From lma3 at uw.edu Tue Apr 19 08:21:23 2016
From: lma3 at uw.edu (Loyce M. Adams)
Date: Tue Jun 12 13:43:55 2018
Subject: [Amath-seminars] Fwd: Thursday Seminar, Savery 130
In-Reply-To:
References:
Message-ID:
FYI. See the announcement below.
---------- Forwarded message ----------
From: Loyce M. Adams
Date: Tue, Apr 19, 2016 at 8:01 AM
Subject: Thursday Seminar, Savery 130
To: "vfaculty@amath.washington.edu" ,
students-local@amath.washington.edu, affiliate@amath.washington.edu,
lma3@uw.edu
Dear Colleagues, Students, Affiliates:
We have a very interesting seminar this Thursday by two local
professors. Carl Bergstrom is from Biology department and Jevin West
from the I-School. Below is the title and abstract. See you there!
Loyce Adams
----
Title: Mapping citation networks for scientific discovery
Carl Bergstrom (Biology, UW) and Jevin West (I-School, UW)
Abstract: The scientific institution of scholarly citation creates a
latticework of citations from which we can in principle map the
geography of scientific thought and retrace the pathways along which
intellectual activity has proceeded. To uncover these patterns, we
need a way of revealing structure in massive network datasets with
weighted and directed links. To do this, we developed the
information-theoretic InfoMap algorithm, which takes advantage of the
duality between compressing data and finding patterns in that data.
With this structure in hand, we can build scholarly recommendation
engines, identify the classic papers and influencers within scientific
domains, and improve the ways in which scholars interact with the
literature. For the talk, we will first present an overview of the
InfoMap algorithm and how it works, and then explore some of the tools
we can create when InfoMap is applied to large citation networks.
From lma3 at uw.edu Sun May 1 08:18:02 2016
From: lma3 at uw.edu (Loyce M. Adams)
Date: Tue Jun 12 13:43:55 2018
Subject: [Amath-seminars] Thursday Seminar, 4-5pm, Savery 130
Message-ID:
Dear All:
Please join us for the seminar this Thursday. Information is below.
Loyce Adams
Prof. Panos Stinis
NIAC, UW
Applied Mathematics, UW
Title: "When big computers are not enough"
Despite the remarkable increase in computational power, most
real-world systems are still too complex to simulate in full detail.
In such cases, the hope is to construct numerical algorithms which can
retain the salient features of a system while reducing the complexity.
I will discuss two different ways of dealing with complexity, namely
mesh refinement and model reduction. The common concept underlying
these approaches is that of transfer of activity from larger scales to
smaller ones. Examples involving singularity detection and tracking as
well as uncertainty quantification for systems exhibiting bifurcations
will be presented for illustration purposes.
From lma3 at uw.edu Tue May 17 14:54:43 2016
From: lma3 at uw.edu (Loyce M. Adams)
Date: Tue Jun 12 13:43:55 2018
Subject: [Amath-seminars] Thursday Seminar, Savery 130, Hannah Choi, 4-5 pm
Message-ID:
Hi all,
This Thursday's seminar, Savery 130 from 4-5pm, will be given by
Hannah Choi. The title and abstract are below.
--- Loyce Adams
Dr. Hannah Choi
Department of Applied Mathematics
"Predictive coding explains discrimination of occluded shapes in an
intermediate visual cortical area"
The visual system recognizes partially occluded objects in natural
scenes without difficulty, and this capacity is yet to be matched in
computer vision. The neural basis of such robust object recognition,
however, is poorly understood. Recent experimental results from
primate area V4, an intermediate stage in the shape processing
pathway, suggests that feedback from higher cortical areas may be
important for maintaining robust shape selective signals. Here we
implement predictive coding to investigate possible underlying
mechanisms. Predictive coding interprets feedback signals as a carrier
of predictions made by the higher cortices about the activities of the
lower cortical areas, and thus has been hypothesized to be a method of
efficient coding. By applying predictive coding to a two-layer,
hierarchical Bayesian model of V4 and its efferent cortical area, we
show that feedback predictions combined with feed-forward sensory
inputs result in robust shape-selective responses under partial
occlusion.
From rjl at uw.edu Sat May 21 12:47:38 2016
From: rjl at uw.edu (Randall J LeVeque)
Date: Tue Jun 12 13:43:55 2018
Subject: [Amath-seminars] Fwd: Clint Dawson's visit next Tuesday and 2 talks
In-Reply-To:
References:
Message-ID:
Clint Dawson from UT-Austin will
be visiting on Tuesday May 24 and giving two talks, one on modeling
hurricane storm surge and one on the DesignSafe
Cyberinfrastructure for hazards
researchers to run models, perform computational experiments, link models,
provide visualization support, archive data, etc.
More details in the abstracts below.
The first talk is in the M9 Project Seminar at 2:30pm in MOL 115
The second talk is in the eScience Institute Community Seminar at 4:30pm in
Physics/Astronomy C520
This will be followed by a reception in the Data Science Studio (6th floor
of Physics/Astronomy)
Clint will be available to talk with people during the day, please sign up
using the the Google Doc
https://docs.google.com/document/d/1jf_jEloWCDUzNdu3beLylZv4c9F6A6s_DS1v9uQjaKA/edit?usp=sharing
-----------------------------------------------------------------------
2:30pm, May 24, 2016 in Molecular Engineering MOL 115
M9 Seminar
Title: Modeling Hurricane Storm Surge and Proposed Mitigation Systems in
the Houston, TX Region
Abstract: Since the 2005-2008 hurricane seasons, there has been extensive
effort to understand and predict the impacts of hurricane-induced flooding
on low-lying regions in the Gulf of Mexico. After Hurricane Ike in 2008,
which made landfall at Galveston, TX, the Severe Storms Prediction,
Education and Evacuation from Disasters (SSPEED) Center was formed,
including a consortium of major universities in Texas. The focus of the
center was to study the socio-economic/environmental/structural impact and
possible mitigation of hurricane storm surge in the Houston-Galveston
region. As part of this effort, our group at the University of Texas
performed extensive studies using the Advanced Circulation (ADCIRC)
modeling framework. In this talk, we will discuss this framework, the
validation of this model for historical hurricanes, and the application of
the model to the study of potential storm mitigation systems, including
structural and non-structural options. We will also discuss a number of
other issues that arise in these types of studies, including political,
environmental and socio-economic considerations.
-----------------------------------------------------------------------
4:30pm, May 24, 2016 in Physics/Astronomy C520
eScience Institute Community Seminar
Title: The Natural Hazards Engineering Research Initiative (NHERI)
DesignSafe Cyberinfrastructure
Abstract: In 2015, the NSF awarded a consortium of universities, led by
the University of Texas at Austin, a grant to develop the DesignSafe-CI as
part of the NHERI. NHERI is the follow-on to NEES, and includes
wind-related hazards (e.g. hurricanes and storm surge) as well as
earthquake-related hazards (structural hazards, tsunamis, etc.). The
purpose of DesignSafe is to develop a new CI for hazards researchers to run
models, perform computational experiments, link models, provide
visualization support, archive data, etc. The CI will also provide support
for RAPID projects and experimental facilities, and will be closely linked
to the NHERI Simulation Center when it is awarded. In this talk, I will
describe the CI and discuss its functionality. I will also invite
discussion on how such a CI could be beneficial to researchers at UW.
-----------------------------------------------------------------------
Clint Dawson earned his Ph.D. in mathematical science from Rice University.
He is a professor of aerospace engineering and engineering mechanics,
leader of the ICES Computational Hydraulics Group and the graduate advisor
of the ICES Computational Sciences, Engineering, and Mathematics graduate
program. He holds the John McKetta Centennial Chair in Engineering.
Dawson?s research interests include numerical methods for partial
differential equations, specifically flow and transport problems in
computational fluid dynamics; scientific computing and parallel computing;
finite element analysis, discontinuous Galerkin methods; shallow water
systems, hurricane storm surge modeling, rainfall-induced flooding; ground
water systems, flow in porous media, geochemistry; data assimilation,
parameter estimation, uncertainty and error estimation.
Dawson has authored or co-authored more than 125 technical articles in the
areas of numerical analysis, numerical methods and parallel computing with
applications to flow and transport in porous media, and shallow water
systems. He has chaired the Society for Industrial and Applied Mathematics
Activity Group on Geosciences, and has served on several conference
organizing committees and review panels. In addition, he has served on
numerous editorial boards, and is currently co-managing editor of
?Computational Geosciences.?
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From lma3 at uw.edu Thu May 26 08:10:07 2016
From: lma3 at uw.edu (Loyce M. Adams)
Date: Tue Jun 12 13:43:55 2018
Subject: [Amath-seminars] Today's Thursday Seminar, Savery 130, 4pm-5pm
Message-ID:
Dear All,
Prof. Michael Buice of the Allen Institute will be giving the seminar
today from 4pm-5pm in Savery 130. His title and abstract are below.
---- Loyce Adams
Dr. Michael Buice
Allen Institute
Title: Correlations and Coding in Sensory Networks
It is well known that there are neurons that code for stimulus
information through spiking. It remains an open question whether the
joint activity of neurons conveys meaningful information about stimuli
over and above that contained in the individual spiking, and how
coding depends upon cell type, layer, and brain region. The Allen
Institute for Brain Science is on the verge of releasing the Allen
Brain Observatory, a large scale data set collected via two photon
optical imaging of activity in several visual areas of the mouse
cortex in response to an array of visual stimuli (static and drifting
gratings, natural images and movies, locally sparse noise) during
passive but awake viewing in which the mice are free to run. I will
give a preview of this data set along with some preliminary results
regarding population coding across the visual areas in the mouse
cortex. I will assume no neuroscience background. Afterwards, I will
describe a theoretical approach to understanding the relationship
between the connectivity within a network and the correlations
observed in the activity, itself related to the computation and coding
capacity of the network. I will also describe how various UW students
and postdocs are contributing to the Allen Institute effort to
understand the computations and dynamics of cortex.
From lma3 at uw.edu Wed Jun 1 20:47:10 2016
From: lma3 at uw.edu (Loyce M. Adams)
Date: Tue Jun 12 13:43:55 2018
Subject: [Amath-seminars] Thursday Seminar, June 2, 4-5pm Savery 130,
Dr. Travis Askham
Message-ID:
Dear All,
Details for tomorrow's seminar, 4-5pm Savery 130 are below. This
is the last Thursday seminar this quarter.
Cheers,
Loyce Adams
Dr. Travis Askham
Department of Applied Mathematics
Integral-equation methods for inhomogeneous elliptic partial differential
equations in complex geometry
Abstract:
In the first part of the talk, we will discuss some of the theory
and numerical algorithms at the heart of the so-called integral-equation
methods, which have emerged as powerful tools for the solution
of the homogeneous partial differential equations of mathematical physics.
By using the methods of potential theory, boundary
value problems can be recast as boundary integral equations, reducing the
dimensionality of the problem by one and permitting geometrically flexible
discretization. Combined with suitable fast algorithms, this has led to
optimal or nearly optimal complexity solvers in a variety of important
application areas.
For inhomogeneous, variable coefficient and nonlinear equations, there
is no corresponding reduction in dimensionality, since the interior of the
domain needs to be discretized and integral-equation methods appear
to be less natural. They lead to dense linear systems of equations,
while direct discretization of the partial differential equation leads to a
sparse system of the same size.
In the second part of the talk, we present a new collection of
integral-equation methods for inhomogeneous partial differential equations
including the Poisson, modified Helmholtz, and modified Stokes equations
that are of optimal computational complexity, while fully adaptive, high-order
accurate and easy to use even in complex geometry. For this, we have
combined fast methods for computing the volume integral operators
of potential theory, with new quadrature methods for layer potentials, and
a new method for smooth function extension from arbitrarily-shaped
domains. Unlike finite difference and finite element discretization of the
governing partial differential equation, the only unknowns in the formulation
lie on the domain boundary, so that there is still an effective reduction in
dimensionality in terms of the size of the linear system which needs to be
solved. The interior degrees of freedom are accounted for using an
explicit integral transform, for which we have developed specialized
versions of the fast multipole method.