Simons Society of Fellows Retreat 2024

Morris Ang
Columbia University

Random Curves and Surfaces
View Slides (PDF)

Morris Ang will give a high-level introduction to random geometries that arise in the contexts of statistical mechanics, conformal field theory and string theory. There will be many nice simulations and visualizations.
 

Paige Arnold
Rockefeller University

Elucidating a Role for Metabolism in the Bacteria Versus Phage Evolutionary Arms Race
View Slides (PDF)

Bacteria are in an ancient evolutionary arms race with prokaryotic viruses (known as phages), the most abundant biological entities on the planet. Given the alarming rise of antibiotic resistance globally, the ability of phages to selectively attack and destroy bacteria has emerged as a potential alternative to treating drug-resistant infections. However, the success of this therapy in contingent on our understanding of the dynamics of phage infection and the mechanisms by which pathogenic bacteria can circumvent elimination by phages. Phages are obligate intracellular parasites that hijack host metabolic machinery to survive and propagate. Bacteria, in turn, exploit phage dependence on host metabolism by employing anti-phage defense systems that disrupt host homeostasis to prevent phage production. However, very little is known about the specific host pathways that are critical for either supporting or preventing phage dissemination. The aim of Arnold’s postdoctoral research is to illuminate both how phages co-opt bacterial pathways to support their dissemination and how bacteria exploit phage dependence on host metabolism to avoid annihilation.
 

Eric Arsenault
Columbia University

Quantum Collusion: Directed Energy and Charge Flow Through Electronic-Vibrational Interactions

Electron behavior underlies the function of molecular, macromolecular and materials systems. However, electrons are not lone actors — their exhibited properties are influenced by the surrounding atomic framework (i.e., structure). In this way, the coupling between electronic and vibrational (atomic) degrees of freedom offers a means to realize selective and directed behavior. The ability to tailor these interactions to manipulate potential energy surfaces and ultimately dynamical processes, such as charge and energy flow, therefore provides an avenue to transform solar energy sources and device platforms.

Eric Arsenault will explore the tools typically employed in measuring relevant dynamical behavior on ultrafast timescales (10-15–10-12 s) and the recent advances in our understanding of natural and materials systems. Arsenault will discuss a variety of contexts in which natural photosynthetic systems leverage electronic and vibrational (el-vib) interactions to efficiently harvest solar energy. This work, while offering design principles for photosynthetic energy production distilled from billions of years of evolution, further offers inspiration for materials applications. Time permitting, Arsenault will introduce the emerging class of two-dimensional van der Waals materials which serve as a promising platform for novel device architectures. While these materials have displayed an incredible range of tuneability, especially in supporting quantum phases of matter, little is known about the role of interactions between the electrons and the surrounding (atomic) lattice. Current work aims to understand such interactions — a step towards dynamical control.
 

Abigail Bodner
New York University

Ocean Mixed Layer Dynamics: From Physics to Climate Models and Back Again

Climate simulations and future climate change projections are notoriously sensitive to unresolved physics involving complex air-sea interactions. This is particularly important in the ocean mixed layer, where small-scale mixing and turbulence modulate the transfer of properties — such as heat, momentum and carbon — between the atmosphere and ocean interior. These processes are on scales much smaller than the grid used in climate models, even at the highest possible resolution.

In this talk, Abigail Bodner will present work towards a comprehensive understanding of multi-scale turbulent interactions in the ocean mixed layer. Bodner will discuss insights learned from a combination of theory, numerical models and data driven methods in an attempt to isolate individual processes and improve their unresolved physical effects in climate models.
 

Floor Broekgaarden
Columbia University

Gravitational-Wave Paleontology: A New Frontier to Explore the Lives and Deaths of Massive Stars Across Cosmic History

Stars with more than ten times the mass of our sun are extremely rare, but play an outsized role in our universe: from shaping and enriching galaxies to triggering new star formation to producing all the oxygen we breathe here on earth. Despite their importance, much of the formation, lives and explosive deaths of these massive stars remains a mystery because it is extremely challenging to observe and study individual massive stars.

In this talk, Floor Broekgaarden will explain how her research aims to open a new frontier in astronomy called “gravitational wave paleontology” that would completely revolutionize this picture. Broekgaarden will explain how we can use observations of gravitational waves, ripples in space-time itself created by colliding black holes, as fossils to study the lives and deaths of massive stars. Broekgaarden will discuss the computational challenges in this effort and her work to overcome these in the pursuit of unraveling the mysteries of massive stars and black holes throughout our vast cosmos!
 

Michael Chapman
New York University

Are There Non-Approximable Objects?
View Slides (PDF)

It is a common theme in science and mathematics to use approximations. In this talk, Michael Chapman will discuss various types of (mathematical) approximations, specifically the notion of “convergence” — when a sequence of “simple” objects approaches a “complicated” limit object. After discussing various manifestations of convergence, we will ask a peculiar kind of question: Are there objects which are too complicated to be approximated? Namely, do the limits of “simple” objects cover everything?

Versions of the above question are commonly called “soficity problems”. Until a few years ago, the mathematical community essentially had no tools for tackling soficity problems — there were no known tools for showing these weak approximations do not exist. Recent breakthroughs in quantum information theory and theoretical computer science, quite miraculously, are breaching this front.
 

Sanchit Chaturvedi
New York University

The Shaky Past, Exciting Present and Bright Future of Theory of Gases

In this talk, Sanchit Chaturvedi will describe the history and the mathematical models used in the theory of gases. Chaturvedi will then talk about some recent developments and the direction the field is headed in.
 

Emanuele Galiffi
Advanced Science Research Center, CUNY

Scattering Light with Temporal Structure

Reflection, refraction, and diffraction are the basic building blocks of wave phenomena. They occur as the light scatters in the presence of spatial inhomogeneities, such as a mirror, a lens, or the grooves on a compact disk. Importantly, the static nature of these scatterers constrains our ability to control light with them. In practice, these limitations manifest themselves in e.g. fundamental bounds to the bandwidth of optical absorbers, the reciprocal nature of wave propagation, and the fundamental link between dissipation and dispersion (i.e. the frequency-dependence of the speed of light, responsible for e.g. rainbows!). But what if matter could be manipulated in time, as well as space? Can we overcome these limitations by engineering light scattering through temporal inhomogeneities?

In this talk, Emanuele Galiffi will discuss recent progress and opportunities for controlling light via temporally engineered matter, including theoretical and experimental achievements, the rise of extremely nonlinear materials, and the design of artificial “metamaterials”, whose electromagnetic properties can be switched within timescales faster than a single oscillation of the wave field.
 

Jane Hubbard
NYU Grossman School of Medicine

Diet, Aging and Stem Cells
View Slides (PDF)

We are currently investigating mechanisms by which diet and aging impact stem cells. To gain mechanistic molecular and cellular resolution, our studies leverage the nematode worm C. elegans. Several features of this research organism facilitate our work: it feeds on a bacterial diet, has a short lifespan and is transparent. It is also amenable to large-scale genetic screens and to genetic manipulation. Importantly, it bears a germline stem cell pool that shares features with mammalian stem cells and that is sensitive to both diet and age. We have identified evolutionarily conserved cellular and neuroendocrine signaling pathways that regulate the stem cell response to diet. Certain manipulations of these pathways allow the pool of stem cells to increase even when nutrients are scarce, underscoring the importance of inter-organ signaling for resource allocation. Our work in C. elegans has also led to the identification of microbial products that affect a pathogenic nematode parasite. On the aging front, our studies reveal separate tissue-specific mechanisms regulating lifespan, stem cell maintenance and the morphology of the stem cell niche. These studies may have implications for treating age-related loss of stem cells, a key hallmark of aging.
 

Ella King
New York University

Nonreciprocal Wave-Matter Interactions

Magical things happen when waves interact with matter. Sunlight helps to unfurl comet tails. Water waves reshape coastlines. Sound waves create flashes of light by cavitating water. Despite the centuries of study that have gone into untangling the magic of wave-matter interactions, we still observe behavior that defies all expectations.

Not only do we observe wave-matter interactions in nature, we can harness them to manipulate particles: by focusing a beam of sound, we can levitate particles in the air. Interestingly, when we levitate groups of particles in a sound field, these collections of particles begin to exhibit strange behavior, including effective interactions that seem to violate Newton’s fundamental laws.

In this talk, Ella King will uncover a novel effective driving force between particles in a sound field. The effective interaction force is nonreciprocal, a property that seemingly violates Newton’s third law. This nonreciprocity leads to emergent activity: collections of levitated particles will spontaneously rotate in some configurations and remain stationary in others. The forces we find can be used to design novel behavior in systems of particles interacting with waves, such as phononic wave guides and band gap materials.
 

Isabel Low
Columbia University

Investigating the Transformation of Memory into Action in Food-Caching Birds

Humans and other animals form one-shot episodic memories in an instant, which we can later use to guide our actions. For example, black-capped chickadees cache thousands of seeds and use memory to retrieve them. For chickadees, this process is a matter of life and death. To survive, chickadees must not only remember where they have hidden their seeds; they must also make the correct moment-to-moment decisions of whether to eat or cache them. This interplay between memory and behavior likely depends on precise coordination between the hippocampus, which supports memory formation and retrieval, and downstream regions that guide feeding behavior.

Isabel Low has identified a compact nucleus of hippocampal neurons that projects to one such feeding center, the lateral hypothalamus. Low is using lightweight, miniaturized neural recording devices to record from this hippocampal projection nucleus during food-caching behavior in an arena designed to engage this natural behavior in the lab. To identify which memory signals are relayed downstream, Low is tagging hippocampal neurons that project to the lateral hypothalamus. By relating these signals to transitions between feeding and caching, Low hopes to gain insight into the computations that transform an episodic memory of a cached seed into decisions to retrieve and eat or re-cache that seed.
 

Carol Mason
Columbia University

Encoding the Brain Circuit For Binocular Vision, Or, What’s Melanin in the Retinal Pigment Epithelium Got to do With It?
View Slides (PDF)

Binocular, or stereo, vision is a key feature of visual perception. The binocular circuit forms during brain development by the growth of retinal ganglion cells (RGCs) from each eye to the opposite brain hemisphere or contralaterally, and to the same hemisphere, or ipsilaterally. The ratio of ipsilateral : contralateral RGCs is tightly controlled during development by distinct programs of gene expression.

Closely apposed to the neural retina is a cell layer called the retinal pigment epithelium (RPE). RPE cells are connected to each other and to developing RGCs and have protective functions in the adult eye. In humans and mice with albinism, the RPE lacks melanin and the proportion of ipsi- to contralateral RGCs is altered, implicating a role for the RPE in determining RGC cell fate during development. Albino mouse retinae display a delay in the G1 phase of the cell cycle that can be rescued by activation of CyclinD2, a cell cycle regulator in the ciliary margin zone (CMZ) at the retina edge, where ipsilateral RGCs are born. This suggests a critical link between cell cycle timing and the production of ipsilateral RGCs.

We are now probing RPE interactions and communication during RGC cell fate specification: 1) Through ‘omics approaches, identifying factors that are missing or aberrantly upregulated in the albino RPE, and 2) analyzing transfer of RPE factors via junctions between the retina and RPE. This work relates to a relatively unstudied subject in developmental neurobiology, that of an epithelial cell population influencing neurogenesis, which may not be unique to the retina.
 

Francesca Mignacco
CUNY Graduate Center

Statistical Physics Approaches to Study Large Populations of Neurons

Recent experimental breakthroughs have paved the way for collecting “big” neural datasets through the simultaneous recording of the activity in thousands of neurons. However, our understanding of the fundamental principles governing neural activity at the population level remains sparse and requires the establishment of appropriate theoretical frameworks. In particular, understanding how neural systems process information through high-dimensional representations presents a fundamental open challenge. A parallel issue emerges in the context of artificial neural networks, that operate efficiently via the interaction of billions of artificial neurons. A major theoretical difficulty resides in solving complex inverse problems in high dimensions. Moreover, the multi-scale information propagation in large neural systems may involve highly non-linear transformations preserving only some relevant degrees of freedom. Statistical physics provides powerful tools to tackle precisely the inherent complexity of large-scale systems via prototypical models.

In this talk, Francesca Mignacco will propose two complementary approaches to extract interpretable low-dimensional characterizations of the neural representations. These methods — rooted in statistical mechanics and information theory — are data-driven and widely applicable across datasets and models.
 

Oliver Philcox
Columbia University

Reflecting the Universe in a Mirror
View Slides (PDF)

From a human perspective, the world is asymmetric. As we turn our heads, move position and look in a mirror, our surroundings look different. Zooming out to cosmological scales, the situation changes considerably; the average properties of the universe are unchanged by rotations and translations. Does the Universe also look the same if you reflect it in a mirror?

In this talk, Oliver Philcox will discuss measurements of the mirror-asymmetric universe constructed from catalogs of galaxies measured by large telescopes, commenting both on the plausibility of these findings and their implications for our models of the universe. By performing such analyses, we can place bounds on some of the highest-energy (and most mysterious) processes occurring in physics.
 

Cynthia Steinhardt
Columbia University

Investigating How the Brain Interprets Artificial Electrical Signals via Cochlear Implants

Neural implants treat neurological disorders by injecting electrical current into the body in a variety of applications from the restoration of hearing and sight to treatment of pain, tremor or depression. Individual neurons are known to communicate through the opening and closing of voltage-gated channels that are naturally depolarized by endogenous chemical signaling. Neural implant-based treatments rely on the ability to similarly depolarize neurons to artificially induce signaling in a way that replicates how the brain naturally encodes information in targeted neurons. However, recent results indicate several differences in the identity of neurons activated and the signals they produce over time compared to natural activation. These differences may explain the ubiquitous limitations in the restoration of function across neural implant treatments. A major counter case is the cochlear implant in which fully electrical encoding can provide auditory input to the level at which children born deaf can learn to understand and speak English.

Our work uses an auditory speech recognition neural network as a model of the human auditory system to investigate how the auditory system may effectively use cochlear implant signals despite their mismatch to expectation. We compare how the model performs speech perception with normal auditory inputs and inputs transformed in the manner of cochlear implant stimulation. Our model shows auditory confusions that imitate those observed in cochlear implant patients when using cochlear implant inputs. We additionally find processing differences across the layers of the network in representation and timing when the model interprets cochlear implant inputs. Our findings suggest that there are signatures of poorly-represented auditory information and population-level mechanisms of information extraction in the presence of noise that could be utilized to guide design of improved neural implant-based interventions.
 

Andre Toussaint
Columbia University

Unraveling the Behavioral and Neural signatures of Chronic pain and Opioid abuse Vulnerability
View Slides (PDF)

The reason why chronic pain (CP) and opioid use disorder (OUD) co-occur is a significant public health concern. Despite advancements in pain management and opioid addiction treatment, the shared neural basis of CP and comorbid OUD are poorly understood, presenting challenges in treatment optimization. Heightened anxiety, a shared psychological risk factor between CP and OUD, offers an ideal approach to understand the cellular basis of both disorders. Using the SJLJ mouse strain, recognized for extreme aggression in males, our lab found that they display distinct sensory and affective pain features at baseline conditions. Sensory manifestations included elevated paw lifting responses, while affective components were evidenced by increased grimacing following stimulation.

Moreover, baseline assessments revealed heightened levels of anxiety in SJLJ males, as demonstrated by their behavior in open field tests and social interaction paradigms. These findings are intriguing because they parallel reports of the comorbidity of chronic pain and anxiety typically observed in clinical populations. Our next step involves recording neural activity in the basolateral amygdala (BLA), a critical brain region for both pain and emotional modulation, at single cell resolution using calcium imaging. The goal is to shed light on the interconnectedness of pain processing and emotional states in SJLJ mice, providing valuable insights into the neurobiological mechanisms underlying pain hypersensitivity and anxiety.
 

Lynn Yap
Columbia University

Smelling by Association: How Recollections Shape Our Perception of Odors

What we perceive is a function of incoming sensory inputs and our recollections of past experiences. Across sensory modalities, this interplay between externally and internally generated signals in the brain gives rise to our perceptual experience. It has long been appreciated that the emergence of olfactory percepts relies on learned associations with other stimuli — a verbal, visual or spatial cue for example.

In this talk, Lynn Yap will describe how such learned associations modulate neural activity in the primary olfactory cortex, using a series of neural recordings in mice as they learn a set of statistical regularities in their environment. This work sheds light on the defining role that prior experience has on our perception of odors, furthering our understanding of the close link between our sense of smell and our memories.