Seminars

Galaxy images and spectra map large-scale cosmic structure and constrain fundamental physics. While galaxy-clustering bias models are well developed, forthcoming datasets can also trace the intergalactic medium and recover complementary cosmic-web information. Jamie Sullivan discusses models for both neutral and ionized intergalactic material. Neutral-hydrogen work includes the environments of Lyman-alpha emitters at redshifts 2.5–5, relevant to future DESI extensions and Spec-S5, together with the measured Lyman-alpha forest and an observationally practical compression of its bispectrum. Ionized-gas research studies momentum through an effective-field-theory kinetic Sunyaev–Zel’dovich bispectrum, and density through fast-radio-burst clustering and cross-correlations. Together, these approaches aim to map intergalactic modes alongside galaxy tracers and improve precision cosmological constraints.

Scientific computing has environmental impacts that extend beyond the growing use of artificial intelligence. This seminar examines how research computing affects the environment, how researchers can measure and reduce those effects, and how funders are responding. It also asks whether efficiency alone provides an adequate strategy. The speakers introduce Green DiSC, a certification scheme for sustainable computing, and the Environmentally Sustainable Computational Science community forum. The discussion connects practical choices in computational research with community approaches to sustainability and concludes with questions. It is suitable for researchers and software developers without prior knowledge of sustainable computing.

This seminar investigates Anderson localization in subwavelength elastic periodic systems with random perturbations. For the unperturbed structure, layer-potential methods recast the eigenvalue problem as boundary integral equations, yielding asymptotic expressions for subwavelength eigenvalues and establishing a band gap above the subwavelength band. A Floquet transform then brings the perturbed problem into a periodic formulation and produces equations for resonant frequencies under general perturbations. Numerical experiments on monomer and dimer structures test agreement with the analysis. Increasing the strength and number of random perturbations demonstrates localization, providing a mathematical basis for elastic metamaterial design.

Deep networks are often thought of as black boxes. Their ability to encompass vast swathes of knowledge indeed makes them hard to explain. Yet many of their behaviours — generalization under overparametrization, grokking, OOD failures, neural scaling laws — recur across architectures and scales, and each, however surprising, can be reproduced and explained in controlled settings. I will review these efforts to identify and explain the universal phenomena of deep learning, and suggest that an LLM may amount to a sum of such tractable sub-phenomena, interpolative in nature. Finally, leaving scientific rigor aside, I'll argue that what separates this prosaic picture from the apparent magic of LLMs may well be the industrial scale of compute and human labour behind it, and that AGI in its deeper extrapolative sense may be much further away than claimed.

Jie Yang presents BRIDGE, a multilingual evaluation benchmark built from real clinical tasks and more than one million samples derived from electronic health records. The benchmark addresses the gap between simplified examination-style tests and the complexity of clinical data, while enabling comparison of rapidly changing language models. Evaluation of 95 models involved over 24,000 experiments and 39 million predictions. Results vary substantially by model family, task and language, with some open-source systems matching proprietary models. The study also finds that chain-of-thought prompting frequently reduces accuracy on these tasks and examines stigmatizing language produced during model reasoning at large scale.

Matthias Vigl examines whether the computational scaling strategies behind modern machine learning can benefit particle physics. Empirical scaling laws relate model performance to computing resources and help balance model size against training-data volume. For a fixed dataset, increasing model capacity beyond the interpolation threshold can improve generalization through double descent, but gains eventually encounter limits imposed by the data. The talk then considers scaling data and model size together, deriving compute-optimal relations for transformer-based jet taggers trained on as many as billions of simulated jets. It studies how training hyperparameters and input representations change these relations. The observed behaviour suggests that larger computing budgets could yield useful gains in particle physics, especially with a move toward more general-purpose foundation models.

How do humans stabilize an inverted pendulum, and why does a balanced pole eventually fall? Drawing on 25 years of fingertip pole-balancing research, this talk examines neural correction delays: longer poles move more slowly relative to the nervous system’s response time. Delay-differential models can stabilize the upright position, yet skilled people still experience falls. The proposed explanation is microchaos arising from interactions among delay, sensory dead zones and frequency-dependent force encoding. A region of transient falling solutions lies next to stable microchaotic dynamics. Such microchaos is absent in virtual frontal-plane balancing tasks, while models of standing postural sway lack the corresponding transient regime. The comparison suggests that human falls, unlike pole falls, are more plausibly associated with medical events or slips and trips.

Robot missions require both collision-free trajectories and logical task ordering, such as collecting a resource before entering a restricted region. Augmented graphs of convex sets combine continuous trajectory optimization and discrete task sequencing in one problem. A layered graph constructed from an exact convex partition of free space selects an optimal task order and trajectory through a shortest-path calculation, exact up to finite Bézier parameterization. Its structure corresponds to Bellman–Held–Karp dynamic programming for the travelling-salesperson problem, retaining singly exponential worst-case complexity and improving exponentially on general temporal-logic tools. A library of specifications covers ordered collection, alternative keys, combined prerequisites, timing and conditional logic with correctness guarantees. Numerical benchmarks demonstrate exponential speedups and near-global optimality. Possible extensions include moving environments, safety-aware planning with conformal prediction sets and a large galactic-survey benchmark.

Seminar by Morten Kjaergaard (Niels Bohr Institute, University of Copenhagen) on rapid FPGA-based feedback for superconducting qubits, including sparse-sampling techniques and on-FPGA inference enabling millisecond-timescale T1 estimation, ~100 ms readout optimization, and over 74,000 consecutive recalibrations in closed-loop operation.

To orchestrate complex behaviors in mammals, the cerebral cortex must continuously transmit its processing results to subcortical regions. While internal cortical processing relies on highly selective sparse codes, these descending cortical output streams employ an enigmatic dense code, characterized by high firing rates and low feature selectivity. This dense population code presents a profound paradox: how can downstream regions extract precise, cognitively relevant signals when most cortical output neurons can be active at any moment? We address this paradox by testing the hypothesis that cortex does not lose information when switching to a dense code. Instead, it utilizes a mechanism for sparse-to-dense coding transformations discovered by my laboratory: thalamocortical synapses target specifically the dendritic initiation zone for calcium action potentials (APs), which enables cortical output neurons to transmit multiple information streams simultaneously via a multiplexed 1-2-3 AP syntax. I will provide first evidence that this remarkable synaptic specificity and coding syntax generalize across long-range pathways. Information coupling via an anatomically and biophysically distinct dendritic nexus may hence be a ubiquitous mechanism for sparse-to-dense coding transformations in cortex. Dissecting these mechanistic origins of cortical output streams is only now possible due to a unique in vivo – in silico approach, perfected over two decades, that my laboratory developed for bridging the gaps between dendritic and population-level computations

Yixuan He (Arizona State University) considers graph neural networks for social, financial and biological networks in which edge signs and directions contain information needed for clustering and prediction. After an introduction to graph neural networks, the talk explains how signed and directed methods preserve relationships lost when every connection is treated as unsigned and undirected. The seminar then asks when a representation learned before the final task can transfer to a different graph or to a setting with few labels. Two theoretical perspectives address transfer across graph domains and the use of link prediction for linear community detection. TopoDIG and TopoSIGN combine encoders that preserve edge structure with persistent topological features, enabling pre-training and prompt-based adaptation on directed and signed networks.

Alok Puranik (Jane Street) examines data mixing for large language models: allocating model capacity and computation among training sources. A small model’s preferred data mix may differ from that of a larger model, yet the cost of large-model training forces many decisions to depend on smaller experiments. Scaling laws can predict performance as size and computation increase only when trends remain stable or change predictably. The talk uses experiments from Puranik’s work and frontier laboratories to examine data-mix questions that violate these assumptions, explain why small-scale results can mislead, and explore better methods for choosing training data. His research at Jane Street concerns scaling laws for sequence models in trading.

Navid Azizan presents methods that make neural models obey physical, safety and operational constraints at deployment. Hard-constrained neural networks, or HardNets, enforce input-dependent constraints by construction while preserving universal approximation within the feasible function class. Applications include models of chaotic dynamics with bounded trajectories, energy-constrained operator learning, safe reinforcement learning and control with formal guarantees. The talk then turns to enforcing constraints during sampling from pretrained diffusion and flow-matching models. Formulating generation as trajectory optimization allows receding-horizon control to guide outputs toward feasibility without retraining or imposing excessive restrictions on the sampling process. Examples from fluid dynamics, robot planning and control, PDE control and language-guided image editing illustrate how constraints can define admissible behavior while retaining expressive learning and generation.

Lenka Zdeborová introduces the Rules-and-Facts model, in which some observations follow a shared rule while others are isolated exceptions requiring memorization. The framework studies when a learner can acquire the rule and retain those exceptions simultaneously. Its results emphasize how capacity is organized and deployed, rather than capacity alone. The talk examines how regularization and the geometry of kernels or learned feature maps can reserve resources for memorization without undermining rule learning. It connects the balance between abstraction and memory to architectural and algorithmic choices, providing a theoretical account of neural networks that both generalize and recall specific facts.

MIT Biology Colloquium Series talk by Peter Koo (Cold Spring Harbor Laboratory), hosted by Yunha Hwang, on using AI to learn the gene regulatory code.

Srini Devadas examines how hardware and cryptography can strengthen each other. Hardware can protect keys, monitor attacks in real time and reduce the trusted computing base, while hardware vulnerabilities can undermine software security. The talk presents research on accelerating cryptographic computation and using those capabilities to design more efficient protocols. It also asks how cryptography and new security definitions can guide the construction of secure hardware. Devadas is MIT's Webster Professor of Electrical Engineering and Computer Science. Hybrid: MIT Stata Center, Room 32-G449 (Kiva), 32 Vassar Street, Cambridge, MA 02139; online via Zoom. Thursday 24 September 2026, talk at 19:00 EDT (UTC−4); in-person refreshments begin at 18:30. This public event is organized by the Boston Chapter of the IEEE Computer Society and GBC/ACM. Register through the organizer-linked Zoom form for either mode and indicate online or in-person attendance. For physical access, use the Stata Center entrance nearest Main Street, then the fourth-floor elevators.

Leonid Berlyand discusses mathematical models for the onset of cell motion driven by myosin contraction. A two-dimensional free-boundary PDE model links cell-shape evolution to diffusion and Keller–Segel-type transport. The talk examines linear stability, a stability-determining eigenvalue and the way nonlinear diffusion changes the bifurcation from supercritical to subcritical. It also considers the curvature of the bifurcation curve, connections to bistability, and the role of non-self-adjoint operators. An example illustrates why a spectral gap alone need not guarantee stability. This is an in-person Analysis of Fluids and Related Topics seminar at Princeton.

Florian Marquardt surveys approaches to reducing the energy and computational cost of neural networks by implementing their functions in physical devices. The central question is how experiments can efficiently determine updates to adjustable parameters, allowing the hardware itself to learn. Examples include learning through time reversal, obtaining nonlinear computation from linear optical scattering, and training analog quantum simulators as neuromorphic systems. The seminar is available in person and through the registration link on the organizer’s event page.

Salvatore Torquato gives a public lecture in the Simons Foundation's 2026 Randomness series, connecting the geometry of complex systems with random structures and applications across mathematics, computer science, physics and chemistry.

This seminar discusses a proof-of-concept gravitational-waveform model described in arXiv:2511.10522. It augments post-Newtonian equations using physics-informed neural-network methods and universal differential equations. The talk also covers ongoing development of a precessing surrogate model designed to evaluate in less than one millisecond while covering a broader portion of the LVK frequency band.