Seminars in Fall 2017

All seminars are held at 4:10 PM in Physics 123, unless otherwise noted.
Refreshments will be served at 4:00 PM.

Upcoming Seminar

November 29, 2017
Student Thesis Talks

Ruoyu Liu

Lizzy Arellano

Malcolm McCarthy

Chris Orita

Aug 30

John Essick
Thesis Start-Up Talk

Sept 6

Aja Procita, Beadle Beadle, Patrick Bedard, Nate MacFadden, Noah Shofer
Summer student presentations

Sept 13

Farhan Hasan, Amanda Swanson, Yunjia Bao, Sarah Racz, Andrew Ryder
Summer student presentations

Sept 20

Chuck Adler, St. Mary’s College
Where is the Science in all that Fiction?

When you hear the words "science fiction", you feel that there should be some science in the fiction.  However, a lot of books, TV shows and movies throw in "bolognium" which looks and sounds "scientific", but is really nonsense.  How can we sort through the baloney to find the good stuff?  I'm going to talk about some basic scientific principles, especially the law of conservation of mass-energy, and review some fictional works that have good science, some that have bad, and one or two which are truly awful.

Sept 27

Lauren Tompkins, Stanford University
The World’s Most Complicated Game of Bingo: Pattern Recognition and the Physics of Subatomic Particles at the Large Hadron Collider

The Large Hadron Collider produces the world’s highest energy particle collisions in order to study the smallest constituents of matter: subatomic particles and their interactions.   These fleetingly rare particles are produced in proton-proton collisions which occur every 25 nanoseconds, requiring fast decisions on which data might hold the clues to the fundamental structure of matter.  I will describe a novel system for observing these collisions by playing Bingo with the detector data, and describe what questions we are trying to answer with these data.

Oct 4

Ira Globus-Harris, Kenji Arai, Ali Cox
Summer student presentations

Oct 11

Caitlin Whalen '08, University of Washington
A Global View of Mixing from Oceanic Internal Waves

Winds blowing over the ocean’s surface and tides flowing over rough bathymetry can produce oceanic internal waves that can propagate horizontally as well as vertically until they break and turbulently mix the water. Even though this mixing occurs on centimeter scales, it has global consequence for the ocean’s density structure, circulation, and surface properties. I will discuss the  complex global geography of this mixing and the underlying mechanisms that govern its distribution in space and time.

Oct 25

Inna Vishik, UC Davis
Shedding light on unconventional superconductors: adventures in momentum space

Superconductivity—the remarkable property of some materials to lose all resistance to current flow below a critical temperature—is a phenomenon that has been observed for over 100 years, but continues to surprise physicists via discoveries of ‘unconventional superconductors’ whose propensity towards superconductivity is unexplained by existing theories.  Some of these unconventional superconductors also have high transition temperatures and confounding electronic phenomena outside of superconductivity.  I will discuss recent experimental progress in understanding, using, and finding more unconventional and high temperature superconductors. 

Nov 1

Rory Donovan-Maiye '04, Allen Institute
Deep Learning the Integrated Cell

Location Change: Biology 19

The modeling team at the Allen Institute for Cell Science has been developing some interesting statistical / machine learning tools for image-based learning, with the ultimate aim of a data-driven understanding of how cells are organized. I’ll present work by our team on two related projects that apply deep neural networks to our 3D images of human induced pluripotent stem cells to accomplish this goal.
First, I’ll discuss a relatively simple tool for predicting images of fluorescently labeled subcellular structures from unlabeled bright-field input images using a U-Net architecture.  This tool produces startlingly good predictions, allowing us to infer many of the features of fluorescent labeling “for free” without any of the associated toxic effects of fluorescent imaging.
Second, I will present our work on our use of generative adversarial networks (GANs) to learn conditional relationships between the morphology of a cell and the localization of its subcellular structures.  Our generative model produces photo-realistic images of subcellular structure localization, and the latent space of the generative model provides an opportunity to explore cellular state transitions — in particular, the cell cycle.
Note: no expertise in either deep learning or cell biology is necessary to understand the majority of this talk.

Nov 8

Peter Collings, Swarthmore College
Water-Based Liquid Crystals: Ordered Fluids with Unusual Properties and Significant Potential for Applications

The liquid crystals used in displays are oily fluids in which the molecules possess orientational order.  Another class of liquid crystals relies on the spontaneous formation of molecular assemblies when certain dyes and drugs are dissolved in water.  These aqueous systems are the subject of significant scientific research, due to the possibility of applications in biology and medicine.  This research reveals that water-based liquid crystals behave quite differently from their oily counterparts, thus creating the understanding necessary to develop new techniques and devices in an area where liquid crystals have had little impact.

Nov 15

Jamie Lomax, University of Washington
A Detailed View of Mass Loss in Massive Stars

Mass loss is important for our understanding of the late evolutionary phases of massive stars, including the formation of Wolf-Rayet (WR) stars and determining what type of supernovae various systems will become. However, small changes in mass loss rates cause very different theoretical predictions for the evolutionary path and supernovae end state for a massive star. For example, massive star winds are expected to be clumped, but this lowers overall mass loss rate estimates to a point where theoretical models cannot explain the observed number of WR stars. Additionally, massive binaries further complicate the mass loss picture because they lose material and interact with their local environments in several ways that single stars cannot: through mass streams, accretion, and mass loss via outflows in Roche-lobe overflow systems; and through stellar winds in colliding-wind systems. In this talk I will focus on presenting results from studies of a small number of systems that provide an in depth look at the circumstellar structures that can form during periods of heavy mass loss. In particular, I will show data from many observing techniques and wavelength regimes that, when viewed all together, present a three dimensional view of where circumstellar material is located in massive systems.

Nov 29

Student Thesis Talks

Ruoyu Liu

Lizzy Arellano

Malcolm McCarthy

Chris Orita