Dr. Spencer Pitcher
Stellarex Group Ltd, CEO

Abstract

Fusion is the process that powers the Sun and the stars and is the primary source of energy in the Universe. It is an inexhaustible, CO2 free source of energy, which generates no long-lived radioactive waste and is intrinsically safe. Fusion energy is on the brink of commercialization worldwide. This lecture will discuss the historical development of fusion energy, give a status of worldwide activities, and describe plans for fusion energy development in Canada and the associated opportunities for universities, research institutes and industry.

Timbits, coffee, tea will be served in STI A before the colloquium.

Dr. Spencer Pitcher
Stellarex Group Ltd. - CEO

Fusion energy is the process that powers the Sun and stars. Scientists have been working on harnessing this power for the benefit of mankind since shortly after the Second World War. Incredible progress has been made over the intervening decades. These advances are to the point now that this technology is on the verge of commercialization.

The recent technical advances have given rise to what is effectively a gold-rush towards fusion energy. Advances include breakthroughs in supercomputers, computer simulations using Artificial Intelligence, advanced materials, and particularly the advent of high-temperature superconductors that are used to make the powerful magnets that contain the fusion fuel that is hotter than the centre of the Sun!

Fusion energy uses fuel that is virtually unlimited on Earth, does not produce greenhouse gas emissions, is intrinsically safe and produces no long-lived or high-level radioactive waste. Fusion is the ultimate energy source, and the original source of all energy in the Universe. Nothing comes after fusion energy.

n this talk, the global status of fusion energy research will be discussed, and the roadmap for Canada’s fusion energy development will be presented.

General admission is free.

Register here

Geoffrey Hinton
University of Toronto

Abstract

To train a neural net efficiently we need to compute the gradient of some measure of the performance of the net with respect to each of the connection weights. The standard way to do this is to use the chain rule to backpropagate gradients through layers of neurons. I shall  describe a very different way of getting the gradients that, for a while, seemed a lot more plausible as a model of how the brain gets gradients.

Consider a system composed of binary neurons that can be active or inactive with weighted pairwise couplings between pairs of neurons, including long range couplings. If the neurons represent pixels in a binary image, we can store a set of binary training images by adjusting the coupling weights so that the images are local minima of a Hopfield energy function which is minus the sum over all pairs of active neurons of their coupling weights. But this energy function can only capture pairwise correlations. It cannot represent the kinds of complicated higher-order correlations that occur in images. Now suppose that in addition to the "visible" neurons that represent the pixel intensities, we also have a large set of hidden neurons that have weighted couplings with each other and with the visible neurons. Suppose also that all of the neurons are asynchronous and stochastic: They adopt the active state with a log odds that is equal to the difference in the energy function when the neuron is inactive versus active. Given a set of training images, is there a simple way to set the weights on all of the couplings so that the training images are local minima of the free energy function obtained by integrating out the states of the hidden neurons?  The Boltzmann machine learning algorithm solved this problem in an elegant way. It  was proof of principle that learning in neural networks with hidden neurons was possible using only locally available information, contrary to what was generally believed at the time.

Timbits, coffee, tea will be served in STI A before the colloquium.

_{Chi-Kwan "CK" Chan,
Associate Astronomer, Steward Observatory}

Abstract

Black holes are the most powerful engines in the Universe, converting gravitational energy into radiation and, in some systems, relativistic jets that propagate across their host galaxies. By resolving these engines on event-horizon scales, the Event Horizon Telescope (EHT) has turned black holes into laboratories for testing gravity and extreme astrophysical models. In this talk, I will discuss my work on the Galactic Center black hole, Sgr A*, whose low luminosity and rapid variability make it a uniquely probe of spacetime and accretion physics. I will also discuss how observing and understanding black hole engines can drive innovation and advances in scientific computing, data science, machine learning, and even space technology

Timbits, coffee, tea will be served in STI A before the colloquium.

Mike McDonald
MIT

Abstract

In the past several decades it has become clear that mechanical (radio-mode) feedback from supermassive blackholes is necessary to moderate the growth of the most massive galaxies in which they reside. In particular, in the cores of galaxy clusters, the evolution of giant elliptical galaxies appears to be primarily governed by black hole feedback. Despite its apparent importance, our understanding of how feedback works is quite incomplete, particularly when it comes to mechanical or radio-mode feedback. In this talk I will discuss two directions that we are pursuing to understand the balance between cooling and feedback in galaxy cluster cores: (i) identifying systems for which feedback appears to not work, in an effort to understand the limitations and failure modes of the feedback/cooling cycle, and (ii) searching for both short- and long-term trends in the importance of AGN feedback, by considering large samples of clusters spanning 10 Gyr in cosmic time. These efforts are made possible by combining data from a variety of X-ray, optical, mm-wave, and radio telescopes, including Chandra, Hubble, James Webb, and the South Pole Telescope. I will conclude with a look towards the future of this field, and highlight some outstanding questions.

Timbits, coffee, tea will be served in STI A before the colloquium.

Prof. Geoffrey Hinton
University of Toronto

George & Maureen Ewan Lecture Series

Artificial Intelligence, or AI, is quickly becoming an integral part of our digital world. Whether it’s as simple as summarizing search results or as complex as generating photorealistic images and videos, there is no question that these tools are significantly impacting our digital experiences.

Prof. Geoffrey Hinton (UofT) has had a long and productive career in machine learning and AI. In 1986, Hinton co-authored a highly cited paper on multi-layer neural networks, and he is considered a leading figure in machine learning, to the point of being referred to as one of the "Godfathers of Deep Learning." In 2024, Hinton was awarded the Nobel Prize in Physics jointly with John Hopfield for "foundational discoveries and inventions that enable machine learning with artificial neural networks."

We are thrilled to host Prof. Hinton in Kingston for the 2026 George and Maureen Ewan Lecture, a lecture series made possible by a generous endowment from the late Ewans, of which George was one of the founding members of the Sudbury Neutrino Observatory.

In his talk, Prof. Hinton will explain the various forms of AI, their potentials, and their limits. He will discuss the advancements and integrations of AI into our digital world and how we should anticipate potential intellectual and security risks.

Register here

Hosted by the McDonald Institute