Madison Chaos and Complex Systems Seminar

Fall 2017 Seminars

All seminars are Tuesday at 12:05 pm in 4274 Chamberlin Hall except as noted. Refreshments will be served.

Short List


September 12, 2017

The Fairsquare project: Countering programs that discriminate

Aws Albarghouthi, UW Department of Computer Science

Software has become a powerful arbitrator of a range of significant decisions with far-reaching societal impact---hiring, welfare allocation, prison sentencing, policing and, among many others. With the range and sensitivity of algorithmic decisions expanding by the day, the problem of understanding the nature of program discrimination and fairness is a pressing one. In this talk, I will describe our work on the FairSquare project, in which we are developing program verification and synthesis tools aimed at rigorously characterizing and reasoning about fairness of decision-making programs.

September 19, 2017

Nonlinear normal modes for analysis of geometrically nonlinear structures

Matthew Allen, UW Department of Engineering Physics

Geometric nonlinearity is an important consideration when designing many structures, for example the skin panels for future hypersonic cruise vehicles where intense pressures and aerodynamic heating can cause the panels to vibrate in and out of buckled states. Highly flexible joined-wing aircraft, which are being sought for station keeping at high altitude, can also exhibit nonlinear dynamic phenomena.  It may also be possible to add a nonlinear element to an otherwise linear structure in order to reduce vibration levels and increase its life, leading to quieter automobiles or more durable spacecraft.  All of these applications are challenging because numerical response predictions are expensive and these nonlinear systems exhibit a large range of phenomena, each of which may require a specialized analysis technique.  This work shows that tremendous insight can be gained into the dynamics of these types of nonlinear structures using undamped nonlinear modal analysis.

This presentation highlights advances in modeling for geometrically nonlinear structures and discusses how nonlinear modes can be used in analysis, design and testing.  While academics have used simplified Galerkin/Ritz models for years to qualitatively study the geometrically nonlinear response of plates and beams, those methods often do not scale to industrial practice where the geometry is far more complicated and many degrees of freedom must be considered.  The work focuses on structures that are modeled in commercial finite element software and uses a non-intrusive approach in which a series of static loads are applied to the structure and a nonlinear Reduced Order Model (ROM) is fit to the load-displacement behavior.  Nonlinear modes prove to be effective in discerning whether the reduced basis contains the fidelity needed to capture the dynamics of interest and in assuring that the loads are large enough to allow the ROM to be accurately computed.  Nonlinear modes are also found to be intimately connected to the response of the structure to random loading, such as the pressure fields experienced by many aircraft.  These concepts are demonstrated by applying them to a variety of finite element models, showing that the nonlinear modes provide tremendous insight into the dynamics of the structure.

September 26, 2017

Intelligent extraterrestrial life: Does it exist? and, if so, what are the prospects for discovery and communication?

Ed Churchwell, UW Department of Astronomy

I will explain what I mean by intelligent life, review the high points about what is known about the evolution of "intelligent" life on Earth and apply some of what are thought to be global principles that are likely to govern the origin and evolution of life in the universe. In particular, I will spend some time on limitations to our prospects for discovery and communication.  This is a very broad subject, and I certainly will not have time to cover all the issues associated with this subject, nor am I qualified to speak about many of them.

October 3, 2017

Systematic misperceptions of 3D motion explained by Bayesian inference

Bas Rokers, UW Department of Psychology

Over the years, a number of surprising, but seemingly unrelated errors in 3D motion perception have been reported. Given the relevance of accurate motion perception to our everyday life, it is important to understand the cause of these perceptual errors. 

We considered that these perceptual errors might arise as a natural consequence of estimating motion direction given sensory noise and the geometry of 3D viewing. We characterized the retinal motion signals produced by objects moving along arbitrary trajectories through three dimensions and developed a Bayesian model of perceptual inference. The model predicted a number of known errors, including a lateral bias in the perception of motion trajectories, and a dependency of this bias on stimulus contrast and viewing distance. The model also predicted a number of previously unknown errors, including a dependency of perceptual bias on eccentricity, and a surprising tendency to misreport approaching motion as receding and vice versa.

We then used standard 3D displays as well as a virtual reality (VR) headsets to test these predictions in naturalistic settings, and established that people make the predicted errors.

In sum, we developed a quantitative model of 3D motion perception and provided a parsimonious account for a range of systematic perceptual errors in naturalistic environments.

October 10, 2017

Understanding complexities of human development in terms of space, time, and energy realities
Part 1: Growing up and growing old

Bernard Z. Friedlander, Department of Psychology, University of Hartford (emeritus)

These presentations ask three questions:
  1. Are there fundamental principles in physical science that can be useful in helping us comprehend fundamental concepts in the organization of human behavior?
  2. Can these principles help us answer the constant question–How do people get the way we are?
  3. Does this approach, in the light of current laboratory research findings, help us clarify persistent scientific and philosophical issues in Western thought?
The two-presentation series touches upon basic concepts in these four areas of serious study: human development, basic physical science, industrial manufacturing technology, and the fine arts.

See expanded written pdf version.

Part 2: Individual differences, volition, and "consciousness" will be given in spring 2018.

October 17, 2017

Chaos theory from a topological perspective

Anda Xiong, UW Department of Physics

I will talk about some phenomena in chaos theory and how can they be viewed by topology, such as the connection between fractals and topological subdivision, and calculating the self-linking number of attractors. Furtherly I will give an example for how persistence homology can be helpful to research in chaos theory.

October 24, 2017

How an important Supreme Court case was created

Bill Whitford, UW Law School

Whitford is the lead plaintiff in Gill v. Whitford, which will be heard by the U.S. Supreme Court on Oct. 3, 2017  The case raises the question whether the federal courts should interpret the U.S. Constitution to place some limits on the extent of partisan bias in legislative districting. If Whitford and the other plaintiffs win, as they did in the trial court, it will be the first time the federal courts have placed any limits of the extent of partisan bias, and hence create a new precedent. Whitford will explain the context in which case arose and how he and others were able to obtain the services of a team of lawyers and experts that have gotten the case to its present position.

See a detailed outline of talk.

October 31, 2017

Partisan redistricting in Wisconsin

David Canon, UW Department of Political Science

Professor David Canon will discuss the potential landmark Supreme Court case, Gill v. Whitford, concerning partisan redistricting. Canon will provide some background on the law and the basic principles of redistricting, explore the history of partisan redistricting and then discuss the Wisconsin case. He will speculate about the likely outcome, with a focus on the role of Justice Kennedy and the new measure of partisan bias (the “efficiency gap”). Canon will conclude by talking about possible reforms, with a focus on Iowa’s nonpartisan model of redistricting.

November 7, 2017

Multidimensional spectroscopy of complex chemical systems: Using nonlinearity to isolate signals

Blaise Thompson, UW Department of Chemistry

Chemical systems are typically composed of many individual components. Each component may be unique. Furthermore, each component may experience a different chemical environment. At room temperature, these environments evolve on ultrafast time scales. Scientists need specialized techniques to understand what is happening in these complex, coupled systems.

Spectroscopy uses the interaction of light and matter to measure chemical energies. Multidimensional spectroscopy (MDS) capitalizes on nonlinearities in this interaction to peer into higher-order properties of the chemical system. These higher-order signals reveal coupling parameters of the system. In this way, scientists can use MDS to isolate unique properties of chemical systems that cannot be measured through other means. Ultrafast dynamics can also be tracked.

This presentation will introduce the basic concepts of MDS. An intuitive description of the technique will be presented. Practical advantages will be highlighted.

November 14, 2017

Low algebraic dimension matrix completion

Becca Willett, UW Department of Electrical and Computer Engineering

The past decade of research on matrix completion has shown it is possible to leverage linear dependencies to impute missing values in a low-rank matrix. However, the corresponding assumption that the data lies in or near a low-dimensional linear subspace is not always met in practice. Extending matrix completion theory and algorithms to exploit low-dimensional nonlinear structure in data will allow missing data imputation in a far richer class of problems. In this talk, I will describe how models of low-dimensional nonlinear structure can be used for matrix completion. In particular, we will explore matrix completion in the context of unions of subspaces, in which data points lie in or near one of several subspaces, and nonlinear algebraic varieties, a polynomial generalization of classical linear subspaces. Low Algebraic-Dimension Matrix Completion (LADMC) is a novel and efficient method for imputing missing values and admits new bounds on the amount of missing data that can be accurately imputed. The proposed algorithms are able to recover synthetically generated data up to predicted sample complexity bounds and outperform standard low-rank matrix completion in experiments with real recommender system and motion capture data

November 21, 2017

Artificial intelligence: Evolution and agency

Terry Allard, Office of Naval Research and NASA (retired)

One fear often expressed in the mass media and popular culture is that artificially intelligent machines will become fully autonomous and self-improving to the point that their capabilities will exceed that of human beings. As a consequence, intelligent machines could become a more dominant life form on the planet, superseding human primacy in culture, science, technology, competitiveness and value. A key concept underlying this fear is the notion that machines can have their own agency independent of human control or influence. Today’s discussion will explore the nature of agency and the future of artificial intelligence from narrow single-function capability to super-intelligence exceeding current human potential.

Listen to an audio version of this talk.
See the slides in PDF format for this talk.

November 28, 2017

Activation versus information in visual working memory

Brad Postle, UW Department of Psychology

Working memory refers to the ability to hold a small amount of information in mind, to manipulate it, and to use it to guide behavior. Individual differences in working memory capacity predict a wide range of psychometric and real-world outcomes, from general fluid intelligence to standardized testing performance to lifetime earning potential. "Working memory" is also often used, particularly by cautious psychologists and neuroscientists, as a proxy for the 'contents of consciousness.'  This talk will address recent work -- using brain imaging (fMRI and EEG), brain stimulation (TMS), and computational modeling -- that challenges the longstanding assumption that for information to be held in working memory, it must be held in an active state.

December 5, 2017

Rewards are worth the risk: Working in direct dark matter detection

Kimberly Palladino, UW Department of Physics

For particle physicists, determining the nature of Dark Matter is one of the greatest open mysteries. An abundance of astrophysical evidence indicates that the matter density of the universe is dominated by a new form of matter, which played a key role in growth of large scale structure. One candidate for Dark Matter is the Weakly Interacting Massive Particle (WIMP). We hope to detect WIMPS by seeing them scattering off of the target materials in our detectors. Liquid xenon has proved itself an excellent target, and LZ is a dual-phase TPC that will begin taking science data in 2020.  Much of the originally proposed parameter space for WIMPS has been excluded over the past few decades, so I will also delve into the sociology of working on direct dark matter searches.

December 12, 2017

Fractal occupancy of human landscapes: The concept of profit in Evansville, WI

Tim Allen, UW Department of Botany

Systems are predictable on two criteria: the thermodynamics of process; rate-independent constraints.  The economics of this distinction is high gain, straight consumption of quality fuel, versus low gain where low quality material is processed to make quality fuel.  A quality resource might sit on a hot spot.  Evansville, WI, is a low gain system that depended on rail road (which is planned).   Janesville and Madison are high gain depending or roads and fossil fuel.  Roads are not planned, they simply straighten and widen in response to the flux of traffic.  Railroads and roads end up with roughly the same fractal dimension, but railroads branch out from the main line, while roads emerge up scale from small roads to turnpikes.  Evansville depends on a diffuse low quality landscape, while amassing capital by concentration.  Janesville and Madison depend on high quality, locally focused resources. The argument turns on my satellite repairman based in Evansville, servicing a diffuse landscape with information and constraint structure.