Past

Recent developments on the black hole information paradox have shown that Euclidean wormholes — so called “replica wormholes’’  — can dominate the von Neumann entropy as computed by a gravitational path integral, and that inclusion of these wormholes results in a unitary Page curve. This development raises some puzzles from the perspective of factorization, and has raised questions regarding what the gravitational path integral is computing. In this talk, I will focus on understanding the relationship between the gravitational path integral and the partition function via the gravitational free energy (more generally the generating functional). A proper computation of the free energy requires a replica trick distinct from the usual one used to compute the entropy. I will show that in JT gravity there is a regime where the free energy computed without replica wormholes is pathological. Interestingly, the inclusion of replica wormholes is not quite sufficient to resolve the pathology: an alternative analytic continuation is required. I will discuss the implications of this for various interpretations of the gravitational path integral (e.g. as computing an ensemble average) and also mention some parallels with spin glasses. 

 I will be reviewing our recent article arXiv:2101.02218 where we propose a simple method for detecting global (a.k.a. non-perturbative) anomalies for generic quantum field theories. The basic idea is to study how the symmetries are realized on the Hilbert space of the theory. I will present several elementary examples where everything can be solved explicitly. After that, we will use these results to make non-trivial predictions about strongly interacting theories.

In this session, William Durham from the Bank of England will give a presentation about working as a mathematician in the BoE, and will give advice on interviewing for non-academic jobs. He has previously provided mock interviews in our department for jobs aimed at mathematicians with PhDs, and is happy to conduct some mock interviews (remotely, of course) for individuals as well.

Please email Helen McGregor (@email) by Monday 22 February if you might be interested in having a mock interview with William Durham on 5 March.
 

For $G$ a finite group, $p$ a prime and $(K, \mathcal{O}_K, k)$ a $p$-modular system the group ring $\mathcal{O}_K G$ is an $\mathcal{O}_k$-order in the $K$-algebra $KG.$ Graduated $\mathcal{O}_K$-orders are a particularly nice class of $\mathcal{O}_K$-orders first introduced by Zassenhaus. In this talk will see that an $\mathcal{O}_K$-order $\Lambda$ in a split $K$-algebra $A$ is graduated if the decomposition numbers for the regular $A$-module are no greater than $1$. Furthermore will see that graduated orders can be described (not uniquely) by a tuple $n$ and a matrix $M$ called the exponant matrix. Finding a suitable $n$ and $M$ for a graduated order $\Lambda$ in the $K$-algebra $A$ provides a parameterisation of the $\Lambda$-lattices inside the regular $A$-module. Understanding the $\mathcal{O}_K G$-lattices inside representations of certain groups $G$ is of interest to those involved in the Langlands programme as well as of independent interest to algebraists.

We propose a mathematical model that unifies the psychiatric concepts of drug-induced incentive salience (IST), reward prediction error

(RPE) and opponent process theory (OPT) to describe the emergence of addiction within substance abuse. The biphasic reward response (initially

positive, then negative) of the OPT is activated by a drug-induced dopamine release, and evolves according to neuro-adaptative brain

processes.  Successive drug intakes enhance the negative component of the reward response, which the user compensates for by increasing the

drug dose.  Further neuroadaptive processes ensue, creating a positive feedback between physiological changes and user-controlled drug

intake. Our drug response model can give rise to qualitatively different pathways for an initially naive user to become fully addicted.  The

path to addiction is represented by trajectories in parameter space that depend on the RPE, drug intake, and neuroadaptive changes.

We will discuss how our model can be used to guide detoxification protocols using auxiliary substances such as methadone, to mitigate withdrawal symptoms.

If this is useful here are my co-authors:
Davide Maestrini, Tom Chou, Maria R. D'Orsogna

Low rank orthogonal tensor approximation (LROTA) is an important problem in tensor computations and their applications. A classical and widely used algorithm is the alternating polar decomposition method (APD). In this talk, I will first give very a brief introduction to tensors and their decompositions. After that, an improved version named iAPD of the classical APD will be proposed and all the following four fundamental properties of iAPD will be discussed : (i) the algorithm converges globally and the whole sequence converges to a KKT point without any assumption; (ii) it exhibits an overall sublinear convergence with an explicit rate which is sharper than the usual O(1/k) for first order methods in optimization; (iii) more importantly, it converges R-linearly for a generic tensor without any assumption; (iv) for almost all LROTA problems, iAPD reduces to APD after finitely many iterations if it converges to a local minimizer. If time permits, I will also present some numerical experiments.

Our understanding and ability to compute the solutions to nonlinear partial differential equations has been strongly curtailed by our inability to effectively parameterize the inertial manifold of their solutions.  I will discuss our ongoing efforts for using machine learning to advance the state of the art, both for developing a qualitative understanding of "turbulent" solutions and for efficient computational approaches.  We aim to learn parameterizations of the solutions that give more insight into the dynamics and/or increase computational efficiency. I will discuss our recent work using machine learning to develop models of the small scale behavior of spatio-temporal complex solutions, with the goal of maintaining accuracy albeit at a highly reduced computational cost relative to a full simulation.  References: https://www.pnas.org/content/116/31/15344 and https://arxiv.org/pdf/2102.01010.pdf 

We explore reinforcement learning methods for finding the optimal policy in the linear quadratic regulator (LQR) problem. In particular, we consider the convergence of policy gradient methods in the setting of known and unknown parameters. We are able to produce a global linear convergence guarantee for this approach in the setting of finite time horizon and stochastic state dynamics under weak assumptions. The convergence of a projected policy gradient method is also established in order to handle problems with constraints. We illustrate the performance of the algorithm with two examples. The first example is the optimal liquidation of a holding in an asset. We show results for the case where we assume a model for the underlying dynamics and where we apply the method to the data directly. The empirical evidence suggests that the policy gradient method can learn the global optimal solution for a larger class of stochastic systems containing the LQR framework and that it is more robust with respect to model mis-specification when compared to a model-based approach. The second example is an LQR system in a higher-dimensional setting with synthetic data.

This talk concerns applications of differential geometry in numerical optimization. They arise when the optimization problem can be formulated as finding an optimum of a real-valued cost function defined on a smooth nonlinear search space. Oftentimes, the search space is a "matrix manifold", in the sense that its points admit natural representations in the form of matrices. In most cases, the matrix manifold structure is due either to the presence of certain nonlinear constraints (such as orthogonality or rank constraints), or to invariance properties in the cost function that need to be factored out in order to obtain a nondegenerate optimization problem. Manifolds that come up in practical applications include the rotation group SO(3) (generation of rigid body motions from sample points), the set of fixed-rank matrices (low-rank models, e.g., in collaborative filtering), the set of 3x3 symmetric positive-definite matrices (interpolation of diffusion tensors), and the shape manifold (morphing).

In the recent years, the practical importance of optimization problems on manifolds has stimulated the development of geometric optimization algorithms that exploit the differential structure of the manifold search space. In this talk, we give an overview of geometric optimization algorithms and their applications, with an emphasis on the underlying geometric concepts and on the numerical efficiency of the algorithm implementations.

A link for this talk will be sent to our mailing list a day or two in advance.  If you are not on the list and wish to be sent a link, please contact @email.

The pandemic has had a deleterious influence on the Hollywood film
industry.  Fortunately,  however, the thin film industry continues to
flourish.  A host of effects are responsible for confined liquids
exhibiting properties that differ from their bulk counterparts. For
example, the dominant polarization and surface forces across a layered
system can control the material behavior on length scales vastly larger
than the film thickness.  This basic class of phenomena, wherein
volume-volume interactions create large pressures, are at play in,
amongst many other settings, wetting, biomaterials, ceramics, colloids,
and tribology.  When the films so created involve phase change and are
present in disequilibrium, the forces can be so large that they destroy
the setting that allowed them to form in the first place. I will
describe the connection between such films in a semi-traditional wetting
dynamics geometry and active brownian dynamics.  I then explore their
power to explain a wide range of processes from materials- to astro- to
geo-science.

Point counting on definable sets in non-archimedean settings has many faces. For sets living in Q_p^n, one can count actual rational points of bounded height, but for sets in C((t))^n, one rather "counts" the polynomials in t of bounded degree. What if the latter is of infinite cardinality? We treat three settings, each with completely different behaviour for point counting : 1) the setting of subanalytic sets, where we show finiteness of point counting but growth can be aribitrarily fast with the degree in t ; 2) the setting of Pfaffian sets, which is new in the non-archimedean world and for which we show an analogue of Wilkie's conjecture in all dimensions; 3) the Hensel minimal setting, which is most general and where finiteness starts to fail, even for definable transcendental curves! In this infinite case, one bounds the dimension rather than the (infinite) cardinality. This represents joint work with Binyamini, Novikov, with Halupczok, Rideau, Vermeulen, and separate work by Cantoral-Farfan, Nguyen, Vermeulen.

Since its introduction in 1978 the curve complex has become one of the most important objects to study surfaces and their homeomorphisms. The curve complex is defined only using data about curves and their disjointness: a stunning feature of it is the fact that this information is enough to give it a rigid structure, that is every symplicial automorphism is induced topologically. Ivanov conjectured that this rigidity is a feature of most objects naturally associated to surfaces, if their structure is rich enough.

During the talk we will introduce the curve complex, then we will focus on its rigidity, giving a sketch of the topological constructions behind the proof. At last we will talk about generalisations of the curve complex, and highlight some rigidity results which are clues that Ivanov's Metaconjecture, even if it is more of a philosophical statement than a mathematical one, could be "true".

A uniform spanning tree of $\mathbb{Z}^4$ can be thought of as the "uniform measure" on trees of $\mathbb{Z}^4$. The past of 0 in the uniform spanning tree is the finite component that is disconnected from infinity when 0 is deleted from the tree. We establish the logarithmic corrections to the probabilities that the past contains a path of length $n$, that it has volume at least $n$ and that it reaches the boundary of the box of side length $n$ around 0. Dimension 4 is the upper critical dimension for this model in the sense that in higher dimensions it exhibits "mean-field" critical behaviour. An important part of our proof is the study of the Newtonian capacity of a loop erased random walk in 4 dimensions. This is joint work with Tom Hutchcroft.

We combine the notions of graded Clifford algebras and Drinfeld algebras. This gives us a framework to study algebras with a PBW property and underlying vector space $\mathbb{C}[G] \# Cl(V) \otimes S(U) $ for $G$-modules $U$ and $V$. The class of graded Clifford-Drinfeld algebras contains the Hecke-Clifford algebras defined by Nazarov, Khongsap-Wang. We give a new example of a GCD algebra which plays a role in an Arakawa-Suzuki duality involving the Clifford algebra.

How can a random walker on a network be helpful for patients suffering from amyotrophic lateral sclerosis (ALS)? Clinical trials in ALS continue to rely on survival or clinical functional scales as endpoints, since anatomical patterns of disease spread in ALS are poorly characterized in vivo. In this study, we generated individual brain networks of patients and controls based on cerebral magnetic resonance imaging (MRI) data. Then, we applied a computational model with a random walker to the brain MRI scan of patients to simulate this progressive network degeneration. We observe that computer‐simulated aggregation levels of the random walker mimic true disease patterns in ALS patients. Our results demonstrate the utility of computational network models in ALS to predict disease progression and underscore their potential as a prognostic biomarker.

After presenting this study on characterizing the structural changes in neurodegenerative diseases with network science, I will give an outlook on my new work on characterizing the dynamic changes in brain networks for Parkinson’s disease and counteracting these with (simulated) deep brain stimulation using the neuroinformatics platform The Virtual Brain (www.thevirtualbrain.org) .

Article link: https://onlinelibrary.wiley.com/doi/full/10.1002/ana.25706

We prove that bounded-degree expanders with non-negative Ollivier-Ricci curvature do not exist, thereby solving a long-standing open problem suggested by Naor and Milman and publicized by Ollivier (2010). In fact, this remains true even if we allow for a vanishing proportion of large degrees, large eigenvalues, and negatively-curved edges. To establish this, we work directly at the level of Benjamini-Schramm limits. More precisely, we exploit the entropic characterization of the Liouville property on stationary random graphs to show that non-negative curvature and spectral expansion are incompatible 'at infinity'. We then transfer this result to finite graphs via local weak convergence and a relative compactness argument. We believe that this 'local weak limit' approach to mixing properties of Markov chains will have many other applications.

The conformal bootstrap program for CFTs in d>2 dimensions is
based on well-defined rules and in principle it could be easily included
into rigorous mathematical physics. I will explain some interesting
conjectures which emerged from the program, but which so far lack rigorous
proof. No prior knowledge of CFTs or conformal bootstrap will be assumed.

Prof. Franco Rampazzo ‘Mathematical Control Theory’ (Department of Mathematics of the University of Padova, as part of Oxford Padova connection) TT 2021
Aimed at: Any DPhil students with interest in learning about Mathematical Control Theory
Course Length:     24 hours total (to be in English) 
Dates and Times:  starts 2 March 2021 

Nonlinear Fokker-Planck equations with measures as initial data and McKean-Vlasov equations This talk is about joint work with Viorel Barbu. We consider a class of nonlinear Fokker-Planck (- Kolmogorov) equations of type \begin{equation*} \frac{\partial}{\partial t} u(t,x) - \Delta_x\beta(u(t,x)) + \mathrm{div} \big(D(x)b(u(t,x))u(t,x)\big) = 0,\quad u(0,\cdot)=\mu, \end{equation*} where $(t,x) \in [0,\infty) \times \mathbb{R}^d$, $d \geq 3$ and $\mu$ is a signed Borel measure on $\mathbb{R}^d$ of bounded variation. In the first part of the talk we shall explain how to construct a solution to the above PDE based on classical nonlinear operator semigroup theory on $L^1(\mathbb{R}^d)$ and new results on $L^1- L^\infty$ regularization of the solution semigroups in our case. In the second part of the talk we shall present a general result about the correspondence of nonlinear Fokker-Planck equations (FPEs) and McKean-Vlasov type SDEs. In particular, it is shown that if one can solve the nonlinear FPE, then one can always construct a weak solution to the corresponding McKean-Vlasov SDE. We would like to emphasize that this, in particular, applies to the singular case, where the coefficients depend "Nemytski-type" on the time-marginal law of the solution process, hence the coefficients are not continuous in the measure-variable with respect to the weak topology on probability measures. This is in contrast to the literature in which the latter is standardly assumed. Hence we can cover nonlinear FPEs as the ones above, which are PDEs for the marginal law densities, realizing an old vision of McKean.

References V. Barbu, M. Röckner: From nonlinear Fokker-Planck equations to solutions of distribution dependent SDE, Ann. Prob. 48 (2020), no. 4, 1902-1920. V. Barbu, M. Röckner: Solutions for nonlinear Fokker-Planck equations with measures as initial data and McKean-Vlasov equations, J. Funct. Anal. 280 (2021), no. 7, 108926.

We will discuss the problem of deciding (algorithmically) whether a variety over a local field K has a K-rational point, surveying some known results. I will then allow K to be an infinite extension (of some arithmetic interest) of a local field and present some recent work.
 

One of the main questions in the theory of the linear transport equation is whether uniqueness of solutions to the Cauchy problem holds in the case the given vector field is not smooth. We will show that even for incompressible, Sobolev (thus quite “well-behaved”) vector fields, uniqueness of solutions can drastically fail. This result can be seen as a counterpart to DiPerna and Lions’ well-posedness theorem, and, more generally, it can be interpreted as an instance of the “flexibility vs. rigidity” duality, which is a common feature of PDEs appearing in completely different fields, such as differential geometry and fluid dynamics (joint with G. Sattig and L. Székelyhidi). 

Studying quasi-isometries between groups is a major theme in geometric group theory. Of particular interest are the situations where the existence of a quasi-isometry between two groups implies that the groups are equivalent in a stronger algebraic sense, such as being commensurable. I will survey some results of this type, and then talk about recent work with Daniel Woodhouse where we prove quasi-isometric rigidity for certain graphs of virtually free groups, which include "generic" cyclic HNN extensions of free groups.

We prove a homological mirror symmetry result for a one-parameter family of genus 2 curves (https://arxiv.org/abs/1908.04227), and then mention current joint work with H. Azam, H. Lee, and C.-C. M. Liu on generalizing this to the 6-parameter family of all genus 2 curves.

First we describe the B-model genus 2 curve in a 4-torus and the geometric construction of the generalized SYZ mirror. Then we set up the Fukaya-Seidel category on the mirror. Finally we will see the main algebraic HMS result on homogenous coordinate rings, which is at the level of cohomology. The method involves first considering mirror symmetry for the 4-torus, then restricting to the hypersurface genus 2 curve and extending to a mirror Landau-Ginzburg model with fiber the mirror 4-torus. 

There has been a recent uptick in interest in fermionic CFTs. These mildly generalise the usual notion of CFT to allow dependence on a background spin structure. I will discuss how this generalisation manifests itself in the symmetries, anomalies, and boundary conditions of the theory, using the series of unitary Virasoro minimal models as an example.

The well known node2vec algorithm has been used to explore network structures and represent the nodes of a graph in a vector space in a way that reflects the structure of the graph. Random walks in node2vec have been used to study the local structure through pairwise interactions. Our motivation for this project comes from a desire to understand higher-order relationships by a similar approach. To this end, we propose an extension of node2vec to a method for representing the k-simplices of a simplicial complex into Euclidean space. 

In this talk I outline a way to do this by performing random walks on simplicial complexes, which have a greater variety of adjacency relations to take into account than in the case of graphs. The walks on simplices are then used to obtain a representation of the simplices. We will show cases in which this method can uncover the roles of higher order simplices in a network and help understand structures in graphs that cannot be seen by using just the random walks on the nodes. 

Saturated fusion systems capture and abstract conjugacy in $p$-subgroups of finite groups and have recently found application in finite group theory, representation theory and algebraic topology. In this talk, we describe a situation in which we may identify a rank $2$ amalgam within $\mathcal{F}$ and, using some local group theoretic techniques, completely determine $\mathcal{F}$ up to isomorphism.

Magnitude is an isometric invariant of metric spaces that was introduced by Tom Leinster in 2010, and is currently the object of intense research, since it has been shown to encode many invariants of a metric space such as volume, dimension, and capacity.

Magnitude homology is a homology theory for metric spaces that has been introduced by Hepworth-Willerton and Leinster-Shulman, and categorifies magnitude in a similar way as the singular homology of a topological space categorifies its Euler characteristic.

In this talk I will first introduce magnitude and magnitude homology. I will then give an overview of existing results and current research in this area, explain how magnitude homology is related to persistent homology, and finally discuss new stability results for magnitude and how it can be used to study point cloud data.

This talk is based on  joint work in progress with Miguel O’Malley and Sara Kalisnik, as well as the preprint https://arxiv.org/abs/1807.01540.

Zilber's Restricted Trichotomy Conjecture predicts that every sufficiently rich strongly minimal structure which can be interpreted from an algebraically closed field K, must itself interpret K. Progress toward this conjecture began in 1993 with the work of Rabinovich, and recently Hasson and Sustretov gave a full proof for structures with universe of dimension 1. In this talk I will discuss a partial result in characteristic zero for universes of dimension greater than 1: namely, the conjecture holds in this case under certain geometric restrictions on definable sets. Time permitting, I will discuss how this result implies the full conjecture for expansions of abelian varieties.

A new explanation of geometric nature of the reservoir computing phenomenon is presented. Reservoir computing is understood in the literature as the possibility of approximating input/output systems with randomly chosen recurrent neural systems and a trained linear readout layer. Light is shed on this phenomenon by constructing what is called strongly universal reservoir systems as random projections of a family of state-space systems that generate Volterra series expansions. This procedure yields a state-affine reservoir system with randomly generated coefficients in a dimension that is logarithmically reduced with respect to the original system. This reservoir system is able to approximate any element in the fading memory filters class just by training a different linear readout for each different filter. Explicit expressions for the probability distributions needed in the generation of the projected reservoir system are stated and bounds for the committed approximation error are provided.

In this talk, we will introduce two problems of contract theory, in continuous–time, with a multitude of agents. First, we will study a model of optimal contracting in a hierarchy, which generalises the one–period framework of Sung (2015). The hierarchy is modeled by a series of interlinked principal–agent problems, leading to a sequence of Stackelberg equilibria. More precisely, the principal (she) can contract with a manager (he), to incentivise him to act in her best interest, despite only observing the net benefits of the total hierarchy. The manager in turn subcontracts the agents below him. Both agents and the manager each independently control a stochastic process representing their outcome. We will see through a simple example that even if the agents only control the drift of their outcome, the manager controls the volatility of the Agents’ continuation utility. Even this first simple example justifies the use of recent results on optimal contracting for drift and volatility control, and therefore the theory on 2BSDEs. We will also discuss some possible extensions of this model. In particular, one extension consists in the elaboration of more general contracts, indexing the compensation of one worker on the result of the others. This increase in the complexity of contracts is beneficial for the principal, and constitutes a first approach to even more complex contracts, in the case, for example, of a continuum of workers with mean–field interactions. This will lead us to introduce the second problem, namely optimal contracting for demand–response management, which consists in extending the model by Aïd, Possamaï, and Touzi (2019) to a mean–field of consumers. Finally, we will conclude by mentioning that this principal-agent approach with a multitude of agents can be used to address many situations, for example to model incentives for
lockdown in the current epidemic context.
 

Data scientists are often faced with the challenge of understanding a high dimensional data set organized as a table. These tables may have columns of different (sometimes, non-numeric) types, and often have many missing entries. In this talk, we discuss how to use low rank models to analyze these big messy data sets. Low rank models perform well --- indeed, suspiciously well — across a wide range of data science applications, including applications in social science, medicine, and machine learning. In this talk, we introduce the mathematics of low rank models, demonstrate a few surprising applications of low rank models in data science, and present a simple mathematical explanation for their effectiveness.

--

A link for this talk will be sent to our mailing list a day or two in advance.  If you are not on the list and wish to be sent a link, please contact @email.

We consider the homogenization of a Stokes system in a domain having many small random holes. This model mainly arises from problems of solid-fluid interaction (e.g. the flow of a viscous and incompressible fluid through a porous medium). We aim at the rigorous derivation of the homogenization limit both in the Brinkmann regime and in the one of Darcy’s law. In particular, we focus on holes that are distributed according to probability measures that allow for overlapping and clustering phenomena.

We study the evolution of solid surfaces and pattern formation by
surface diffusion. Phase field models with degenerate mobilities are
frequently used to model such phenomena, and are validated by
investigating their sharp interface limits. We demonstrate by a careful
asymptotic analysis involving the matching of exponential terms that a
certain combination of degenerate mobility and a double well potential
leads to a combination of both surface and non-linear bulk diffusion to
leading order. If time permits, we will discuss implications and extensions.

We characterize which Borel functions are decomposable into
a countable union of functions which are piecewise continuous on
$\Pi^0_n$ domains, assuming projective determinacy. One ingredient of
our proof is a new characterization of what Borel sets are $\Sigma^0_n$
complete. Another important ingredient is a theorem of Harrington that
there is no projective sequence of length $\omega_1$ of distinct Borel
sets of bounded rank, assuming projective determinacy. This is joint
work with Adam Day.

The course covers the standard material on nonlinear wave equations, including local existence, breakdown criterion, global existence for small data for semi-linear equations, and Strichartz estimate if time allows.

A 3-manifold fibers over the circle if it can be identified with the mapping torus of a surface homeomorphism. If the surface is compact with non-empty boundary then the corresponding 3-manifold group is free-by-cyclic, and the action of the cyclic group on the free group is induced by the surface homeomorphism. Although most free-by-cyclic groups do not arise as fundamental groups of 3-manifolds which fiber over the circle, there is a strong analogy between the two families.

In this talk I will discuss how dynamical properties of the monodromy affect the geometry/algebra of the corresponding mapping torus. We will see how the same 3-manifold or group can admit multiple fiberings and what properties of the monodromy are known to be preserved under different fiberings.

In this talk I review the recent discovery of Yangian symmetry for massive Feynman integrals and how it can be used to set up a Yangian Bootstrap. I will provide elementary proofs of the symmetry at one and two loops, whereas at generic loop order I conjecture that all graphs cut from regular tilings of the plane with massive propagators on the boundary enjoy the symmetry. After demonstrating how the symmetry may be used to constrain the functional form of Feynman integrals on explicit examples, I comment on how a subset of the diagrams for which the symmetry is conjectured to hold is naturally embedded in a Massive Fishnet theory that descends from gamma-deformed Coulomb branch N=4 Super-Yang-Mills theory in a particular double scaling limit.

Since the seminal work of Keating and Snaith, the characteristic polynomial of a random (Haar-distributed) unitary matrix has seen several of its functional studied in relation with the probabilistic study of the Riemann Zeta function. We will recall the history of the topic starting with the Montgommery-Dyson correspondance and will revisit the last twenty years of computations of integer moments of some functionals, with a particular focus on the mid-secular coefficients recently studied by Najnudel-PaquetteSimm. The new method here introduced will be compared with one of the classical ways to deal with such functionals, the Conrey-Farmer-Keating-Rubinstein-Snaith (CFKRS) formula.

For high order finite elements (continuous piecewise polynomials) the conditioning of the basis is important. However, so far there seems no generally accepted concept of "a well-conditioned basis”,  or a general strategy for how to obtain such representations. In this presentation, we use the $L^2$ condition number as a measure of the conditioning, and construct representations by frames such that the associated $L^2$ condition number is bounded independently of the polynomial degree. The main tools include the bubble transform, which is a stable decomposition of functions into local modes, and orthogonal polynomials on simplexes.  We also include a brief discussion on potential applications in preconditioning. This is a joint work with Ragnar Winther. 

--

A link for this talk will be sent to our mailing list a day or two in advance.  If you are not on the list and wish to be sent a link, please contact @email.

We explain how group analogues of Slodowy slices arise by interpreting certain Weyl group elements as braids. Such slices originate from classical work by Steinberg on regular conjugacy classes, and different generalisations recently appeared in work by Sevostyanov on quantum group analogues of W-algebras and in work by He-Lusztig on Deligne-Lusztig varieties.

Our perspective furnishes a common generalisation, essentially solving the problem. We also give a geometric criterion for Weyl group elements to yield strictly transverse slices.

The study of motifs in networks can help researchers uncover links between structure and function of networks in biology, the sociology, economics, and many other areas. Empirical studies of networks have identified feedback loops, feedforward loops, and several other small structures as "motifs" that occur frequently in real-world networks and may contribute by various mechanisms to important functions these systems. However, the mechanisms are unknown for many of these motifs. We propose to distinguish between "structure motifs" (i.e., graphlets) in networks and "process motifs" (which we define as structured sets of walks) on networks and consider process motifs as building blocks of processes on networks. Using the covariances and correlations in a multivariate Ornstein--Uhlenbeck process on a network as examples, we demonstrate that the distinction between structure motifs and process motifs makes it possible to gain quantitative insights into mechanisms that contribute to important functions of dynamical systems on networks.

That neural networks may be pruned to high sparsities and retain high accuracy is well established. Recent research efforts focus on pruning immediately after initialization so as to allow the computational savings afforded by sparsity to extend to the training process. In this work, we introduce a new `DCT plus Sparse' layer architecture, which maintains information propagation and trainability even with as little as 0.01% trainable kernel parameters remaining. We show that standard training of networks built with these layers, and pruned at initialization, achieves state-of-the-art accuracy for extreme sparsities on a variety of benchmark network architectures and datasets. Moreover, these results are achieved using only simple heuristics to determine the locations of the trainable parameters in the network, and thus without having to initially store or compute with the full, unpruned network, as is required by competing prune-at-initialization algorithms. Switching from standard sparse layers to DCT plus Sparse layers does not increase the storage footprint of a network and incurs only a small additional computational overhead.

--

A link for this talk will be sent to our mailing list a day or two in advance.  If you are not on the list and wish to be sent a link, please contact @email.

I will connect approaches to classical integrable systems via 4d Chern-Simons theory and via symmetry reductions of the anti-self-dual Yang-Mills equations. In particular, I will consider holomorphic Chern-Simons theory on twistor space, defined using a range of meromorphic (3,0)-forms. On shell these are, in most cases, found to agree with actions for anti-self-dual Yang-Mills theory on space-time. Under symmetry reduction, these space-time actions yield actions for 2d integrable systems. On the other hand, performing the symmetry reduction directly on twistor space reduces the holomorphic Chern-Simons action to 4d Chern-Simons theory.

Interacting particle systems have found diverse applications in mathematics and several related fields, including statistical physics, population dynamics, and machine learning.  We will focus, in particular, on the zero range process and the symmetric simple exclusion process.  The large-scale behavior of these systems is essentially deterministic, and is described in terms of a hydrodynamic limit.  However, the particle process does exhibit large fluctuations away from its mean.  Such deviations, though rare, can have significant consequences---such as a concentration of energy or the appearance of a vacuum---which make them important to understand and simulate.

In this talk, which is based on joint work with Benjamin Gess, I will introduce a continuum model for simulating rare events in the zero range and symmetric simple exclusion process.  The model is based on an approximating sequence of stochastic partial differential equations with nonlinear, conservative noise.  The solutions capture to first-order the central limit fluctuations of the particle system, and they correctly simulate rare events in terms of a large deviations principle.