Talks

Through their observable properties, the first and smallest dark matter halos represent a rare probe of subkiloparsec-scale variations in the density of the early Universe. These density variations could hold clues to the nature of inflation, the postinflationary cosmic history, and the identity of dark matter. However, the dynamical complexity of these microhalos hinders their usage as cosmological probes. A theoretical understanding of the microhalo-cosmology connection demands numerical simulation, but microhalos are too small and dense to simulate up to the present day in full cosmological context. My research meets this challenge by using controlled numerical simulations to develop (semi)analytic models of dark matter structure. I will discuss these models, which describe the formation of the first halos and their subsequent evolution as they accrete onto larger systems. I will also explore two applications: breaking a degeneracy between the properties of thermal-relic dark matter and the postinflationary history, and probing inflation's late stages via the small-scale primordial power spectrum.

CMB lensing tomography has the potential to map the amplitude and growth of structure over cosmic time, provide some of the most stringent tests of gravity, and break important degeneracies between cosmological parameters. I use the unWISE photometric galaxy catalog to create three samples at median redshifts z~0.6, 1.1, and 1.5, and cross-correlate them with the most recent Planck CMB lensing maps. The resulting significance of ~80 at 100 < ell < 1000 is the highest significance detection to date of CMB lensing cross-correlation.  The redshift distribution of the two-band unWISE galaxies is a major source of systematic error. I primarily use cross-correlations with BOSS galaxies and quasars and eBOSS quasars to measure the redshift distribution, supplemented with cross-matching to deep COSMOS photometric redshifts.  I demonstrate how to propagate the uncertainty in the redshift distribution to the modeling of the signal, and perform a number of null tests. Finally, I discuss the cosmological implications of this measurement and lessons learned for CMB lensing cross-correlations with future photometric surveys.

Science fiction and science both inspire wonder and awe, albeit in very different ways.

At its best, science fiction asks profound questions about the human condition. In contrast, science asks – and often answers – even more profound questions about the very nature of matter, space, and time.

Both science fiction and science fact explore the concept of journeying to other stars and finding life on other worlds. When it comes to interstellar travel, the truth may soon become stranger (and more amazing) than fiction.

In his February 5 public lecture webcast at Perimeter Institute, astronomer Bryan Gaensler will provide an overview of the latest thinking on interstellar travel and on the search for alien life – including why he believes the frontiers of current research may be more exciting and visionary than any fictional stories we can imagine.

Gaensler is the Director of the Dunlap Institute for Astronomy and Astrophysics at the University of Toronto, where he holds a Canada Research Chair. He has authored 400 scientific papers on cosmic magnetism, neutron stars, supernova explosions, and interstellar gas, and his popular science book, Extreme Cosmos, has been translated into six languages.

In this talk I will review the state of the art in PN gravity, and in particular its significant advancement via the EFT of spinning gravitating objects. First, I will introduce the concept of a tower of EFTs for the binary inspiral problem. I will then present the intricate formulation of the EFT of spinning objects. Finally, I will present some advanced results accomplished within this framework.

I’ll talk about two independent works on classical and quantum neural networks connected by information theory. In the first part of the talk, I’ll treat sequence models as one-dimensional classical statistical mechanical systems and analyze the scaling behavior of mutual information. I'll provide a new perspective on why recurrent neural networks are not good at natural language processing. In the second part of the talk, I’ll study information scrambling dynamics when quantum neural networks are trained by classical gradient descent algorithm. For many problems, this hybrid quantum-classical training process consists of two stages where information scrambles very differently in the network.