Past Stochastic Analysis & Mathematical Finance Seminars

Given a finite graph, the arboreal gas is the measure on forests (subgraphs without cycles) in which each edge is weighted by a parameter β greater than 0. Equivalently this model is bond percolation conditioned to be a forest, the independent sets of the graphic matroid, or the q→0 limit of the random cluster representation of the q-state Potts model. Our results rely on the fact that this model is also the graphical representation of the nonlinear sigma model with target space the fermionic hyperbolic plane H^{0|2}, whose symmetry group is the supergroup OSp(1|2).

The main question we are interested in is whether the arboreal gas percolates, i.e., whether for a given β the forest has a connected component that includes a positive fraction of the total edges of the graph. We show that in two dimensions a Mermin-Wagner theorem associated with the OSp(1|2) symmetry of the nonlinear sigma model implies that the arboreal gas does not percolate for any β greater than 0. On the other hand, in three and higher dimensions, we show that percolation occurs for large β by proving that the OSp(1|2) symmetry of the non-linear sigma model is spontaneously broken. We also show that the broken symmetry is accompanied by massless fluctuations (Goldstone mode). This result is achieved by a renormalisation group analysis combined with Ward identities from the internal symmetry of the sigma model.

Using rough path techniques, we provide a priori estimates for the output of Deep Residual Neural Networks in terms of both the input data and the (trained) network weights. As trained network weights are typically very rough when seen as functions of the layer, we propose to derive stability bounds in terms of the total p-variation of trained weights for any p∈[1,3]. Unlike the C1-theory underlying the neural ODE literature, our estimates remain bounded even in the limiting case of weights behaving like Brownian motions, as suggested in [Cohen-Cont-Rossier-Xu, "Scaling Properties of Deep Residual Networks”, 2021]. Mathematically, we interpret residual neural network as solutions to (rough) difference equations, and analyse them based on recent results of discrete time signatures and rough path theory. Based on joint work with C. Bayer and P. K. Friz.
 

In this talk I will present several new stochastic representations for
solutions of the Navier-Stokes equations in a wall-bounded region,
in the spirit of mean field theory. These new representations are
obtained by using the duality of conditional laws associated with the Taylor diffusion family.
By using these representation, Monte-Carlo simulations for boundary fluid flows, including
boundary turbulence, may be implemented. Numerical experiments are given to demonstrate the usefulness of this approach.

We provide a framework for modelling risk and quantifying payment shortfalls in cleared markets with multiple central counterparties (CCPs). Building on the stylised fact that clearing membership is shared among CCPs, we show how this can transmit stress across markets through multiple CCPs. We provide stylised examples to lay out how such stress transmission can take place, as well as empirical evidence to illustrate that the mechanisms we study could be relevant in practice. Furthermore, we show how stress mitigation mechanisms such as variation margin gains haircutting by one CCP can have spillover effects on other CCPs. The framework can be used to enhance CCP stress-testing, which currently relies on the “Cover 2” standard requiring CCPs to be able to withstand the default of their two largest clearing members. We show that who these two clearing members are can be significantly affected by higher-order effects arising from interconnectedness through shared clearing membership. Looking at the full network of CCPs and shared clearing members is therefore important from a financial stability perspective.

This is joint work with Iñaki Aldasoro.

BIS Working Paper No 1052: https://www.bis.org/publ/work1052.pdf

In this talk I will present an exclusion process with different types of particles: A, B and C. This last type can be understood as holes. Two scaling limits will be discussed: hydrodynamic limits in the boundary driven setting; and equilibrium fluctuations for an evolution on the torus. In the later case, we distinguish several cases, that depend on the choice of the jump rates, for which we get in the limit either the stochastic Burgers equation or the Ornstein-Uhlenbeck equation. These results match with predictions from non-linear fluctuating hydrodynamics. 
(Joint work with G. Cannizzaro, A. Occelli, R. Misturini).

The signature is a non-commutative exponential that appeared in the foundational work of K-T Chen in the 1950s. It is also a fundamental object in the theory of rough paths (Lyons, 1998). More recently, it has been proposed, and used, as part of a practical methodology to give a way of summarising multimodal, possibly irregularly sampled, time-ordered data in a way that is insensitive to its parameterisation. A key property underpinning this approach is the ability of linear functionals of the signature to approximate arbitrarily any compactly supported and continuous function on (unparameterised) path space. We present some new results on the properties of a selection of topologies on the space of unparameterised paths. We discuss various applications in this context.
This is based on joint work with William Turner.
 

We introduce so-called functional input neural networks defined on infinite dimensional weighted spaces, where we use an additive family as hidden layer maps and a non-linear activation function applied to each hidden layer. Relying on approximation theory based on Stone-Weierstrass and Nachbin type theorems on weighted spaces, we can prove global universal approximation results for (differentiable and) continuous functions going beyond approximation on compact sets. This applies in particular to approximation of (non-anticipative) path space functionals via functional input neural networks but also via linear maps of the signature of the respective paths. We apply these results in the context of stochastic portfolio theory to generate path dependent portfolios that are trained to outperform the market portfolio. The talk is based on joint works with Philipp Schmocker and Josef Teichmann.

We introduce the mapping space signature, a generalization of the path signature for maps from higher dimensional cubical domains, which is motivated by the topological perspective of iterated integrals by K. T. Chen. We show that the mapping space signature shares many of the analytic and algebraic properties of the path signature; in particular it is universal and characteristic with respect to Jacobian equivalence classes of cubical maps. This is joint work with Chad Giusti, Vidit Nanda, and Harald Oberhauser.

What is the largest deterministic amount of time T∗ that a suitably normalized martingale X can be kept inside a convex body K in Rd? We show, in a viscosity framework, that T∗ equals the time it takes for the relative boundary of K to reach X(0) as it undergoes a geometric flow that we call (positive) minimum curvature flow. This result has close links to the literature on stochastic and game representations of geometric flows. Moreover, the minimum curvature flow can be viewed as an arrival time version of the Ambrosio–Soner codimension-(d − 1) mean curvature flow of the 1-skeleton of K. We present very preliminary sampling-based numerical approximations to the solution of the corresponding PDE. The numerical part is work in progress.

This work is based on a collaboration with Camilo Garcia Trillos, Martin Larsson, and Yufei Zhang.

In this talk, I will discuss the (non-)construction of the focusing Gibbs measures and the associated dynamical problems. This study was initiated by Lebowitz, Rose, and Speer (1988) and continued by Bourgain (1994), Brydges-Slade (1996), and Carlen-Fröhlich-Lebowitz (2016). In the one-dimensional setting, we consider the mass-critical case, where a critical mass threshold is given by the mass of the ground state on the real line. In this case, I will show that the Gibbs measure is indeed normalizable at the optimal mass threshold, thus answering an open question posed by Lebowitz, Rose, and Speer (1988).

In the three dimensional-setting, I will first discuss the construction of the $\Phi^3_3$-measure with a cubic interaction potential. This problem turns out to be critical, exhibiting a phase transition:normalizability in the weakly nonlinear regime and non-normalizability in the strongly nonlinear regime. Then, I will discuss the dynamical problem for the canonical stochastic quantization of the $\Phi^3_3$-measure, namely, the three-dimensional stochastic damped nonlinear wave equation with a quadratic nonlinearity forced by an additive space-time white noise (= the hyperbolic $\Phi^3_3$-model). As for the local theory, I will describe the paracontrolled approach to study stochastic nonlinear wave equations, introduced in my work with Gubinelli and Koch (2018). In the globalization part, I introduce a new, conceptually simple and straightforward approach, where we directly work with the (truncated) Gibbs measure, using the variational formula and ideas from theory of optimal transport.

The first part of the talk is based on a joint work with Philippe Sosoe (Cornell) and Leonardo Tolomeo (Bonn/Edinburgh), while the second part is based on a joint work with Mamoru Okamoto (Osaka) and Leonardo Tolomeo (Bonn/Edinburgh).

In this talk, I will present several results on the Anderson Hamiltonian with white noise potential in dimension 1. This operator formally writes « - Laplacian + white noise ». It arises as the scaling limit of various discrete models and its explicit potential allows for a detailed description of its spectrum. We will discuss localization of its eigenfunctions as well as the behavior of the local statistics of its eigenvalues. Around large energies, we will see that the eigenfunctions are localized and follow a universal shape given by the exponential of a Brownian motion plus a drift, a behavior already observed by Rifkind and Virag in tridiagonal matrix models. Based on joint works with Cyril Labbé.

In this talk, we present recent results on an anisotropic variant of the Kardar-Parisi-Zhang equation, the Anisotropic KPZ equation (AKPZ), in the critical spatial dimension d=2. This is a singular SPDE which is conjectured to capture the behaviour of the fluctuations of a large family of random surface growth phenomena but whose analysis falls outside of the scope not only of classical stochastic calculus but also of the theory of Regularity Structures and paracontrolled calculus. We first consider a regularised version of the AKPZ equation which preserves the invariant measure and prove the conjecture made in [Cannizzaro, Erhard, Toninelli, "The AKPZ equation at stationarity: logarithmic superdiffusivity"], i.e. we show that, at large scales, the correlation length grows like t1/2 (log t)1/4 up to lower order correction. Second, we prove that in the so-called weak coupling regime, i.e. the equation regularised at scale N and the coefficient of the nonlinearity tuned down by a factor (log N)-1/2, the AKPZ equation converges to a linear stochastic heat equation with renormalised coefficients. Time allowing, we will comment on how some of the techniques introduced can be applied to other SPDEs and physical systems at and above criticality. 

In this talk, we consider differential equations perturbed by multiplicative fractional Brownian noise. Depending on the value of the Hurst parameter $H$, the resulting equation is pathwise viewed as an ordinary ($H>1$), Young  ($H \in (1/2, 1)$) or rough  ($H \in (1/3, 1/2)$) differential equation. In all three regimes we show regularisation by noise phenomena by proving the strongest kind of well-posedness  for equations with irregular drifts: strong existence and path-by-path uniqueness. In the Young and smooth regime $H>1/2$ the condition on the drift coefficient is optimal in the sense that it agrees with the one known for the additive case.

In the rough regime $H\in(1/3,1/2)$ we assume positive but arbitrarily small drift regularity for strong 
well-posedness, while for distributional drift we obtain weak existence. 

This is a joint work with Máté Gerencsér.

We study elastic manifolds with self-repelling terms and estimate their effective radius. This class of manifolds is modelled by a self-repelling vector-valued Gaussian free field with Neumann boundary conditions over the domain [−N,N]^d∩Z^d, that takes values in R^D. Our main results state that for two dimensional domain and range (D=2 and d=2), the effective radius R_N​ of the manifold is approximately N. When the dimension of the domain is d=2 and the dimension of the range is D=1, the effective radius of the manifold is approximately N^{4/3}. This verifies the conjecture of Kantor, Kardar and Nelson (Phys. Rev. Lett. ’86). We also provide results for the case where d≥3 and D≤d. These results imply that self-repelling elastic manifolds with a low dimensional range undergo a significantly stronger stretching than in the case where d=D. 

This is a joint work with Carl Mueller.

Noether’s theorem plays a fundamental role in modern physics by relating symmetries of a Lagrangian to conserved quantities of the Euler-Lagrange equations. In ideal fluid dynamics, the theorem relates the particle labeling symmetry to a Kelvin circulation law. Circulation is conserved for incompressible flows and, otherwise, is generated by advected variables through the momentum map due to a broken symmetry. We will introduce variational principles for fluid dynamics that constrain advection to be the sum of a smooth and geometric rough-in-time vector field. The corresponding rough Euler-Poincare equations satisfy a Kelvin circulation theorem and lead to a natural framework to develop parsimonious non-Markovian parameterizations of subgrid-scale dynamics.

 "In this talk, we will discuss invariant measures techniques to establish probabilistic global well-posedness for PDEs. We will go over the limitations that the Gibbs measures and the so-called fluctuation-dissipation measures encounter in the context of energy-supercritical PDEs. Then, we will present a new approach combining the two aforementioned methods and apply it to the energy supercritical Schrödinger equations. We will point out other applications as well."

I will introduce a class of mean-field games under forward performance and for general risk preferences. Players interact through competition in fund management, driven by relative performance concerns in an asset diversification setting. This results in a common-noise mean field game. I will present the value and the optimal policies of such games, as well as some concrete examples. I will also discuss the partial information case, i.e.. when the risk premium is not directly observed. 

 We develop a probabilistic framework for analysing model-based reinforcement learning in the episodic setting. We then apply it to study finite-time horizon stochastic control problems with linear dynamics but unknown coefficients and convex, but possibly irregular, objective function. Using probabilistic representations, we study regularity of the associated cost functions and establish precise estimates for the performance gap between applying optimal feedback control derived from estimated and true model parameters. We identify conditions under which this performance gap is quadratic, improving the linear performance gap in recent work [X. Guo, A. Hu, and Y. Zhang, arXiv preprint, arXiv:2104.09311, (2021)], which matches the results obtained for stochastic linear-quadratic problems. Next, we propose a phase-based learning algorithm for which we show how to optimise exploration-exploitation trade-off and achieve sublinear regrets in high probability and expectation. When assumptions needed for the quadratic performance gap hold, the algorithm achieves an order (N‾‾√lnN) high probability regret, in the general case, and an order ((lnN)2) expected regret, in self-exploration case, over N episodes, matching the best possible results from the literature. The analysis requires novel concentration inequalities for correlated continuous-time observations, which we derive.

-----------------------------------------------------------------------
Dr Lukasz Szpruch

We consider, in a framework of iterated random conformal maps, a two-parameteraggregation model of Hastings-Levitov type, in which the size and intensity of new particles are each chosen to vary as a power of the density of harmonic measure. Then we consider a limit
in which the overall intensity of particles become large, while the particles themselves become
small. For a certain `sub-critical' range of parameter values, we can show a law of large numbers and fluctuation central limit theorem. The admissible range of parameters includes an off-lattice version of the Eden model, for which we can show that disk-shaped clusters are stable.
Many open problem remain, not least because the limit PDE does not yet have a satisfactory mathematical theory.

This is joint work with Vittoria Silvestri and Amanda Turner.

TBC

Most of SDE models in epidemics, ecology, biology, finance etc. are highly nonlinear and do not have explicit solutions. Monte Carlo simulations have played a more and more important role. This talk will point out several well-known numerical schemes may fail to preserve the positivity or moment of the solutions to SDE models. Reliable numerical schemes are therefore required to be designed so that the corresponding Monte Carlo simulations can be trusted. The talk will then concentrate on new numerical schemes for the well-known stochastic Lotka--Volterra model for interacting multi-species. This model has some typical features: highly nonlinear, positive solution and multi-dimensional. The known numerical methods including the tamed/truncated Euler-Maruyama (EM) applied to it do not preserve its positivity. The aim of this talk is to modify the truncated EM to establish a new positive preserving truncated EM (PPTEM).

Modern atmospheric and ocean science require sophisticated geophysical fluid dynamics models. Among them, stochastic partial

differential equations (SPDEs) have become increasingly relevant. The stochasticity in such models can account for the effect

of the unresolved scales (stochastic parametrizations), model uncertainty, unspecified boundary condition, etc. Whilst there is an

extensive SPDE literature, most of it covers models with unrealistic noise terms, making them un-applicable to

geophysical fluid dynamics modelling. There are nevertheless notable exceptions: a number of individual SPDEs with specific forms

and noise structure have been introduced and analysed, each of which with bespoke methodology and painstakingly hard arguments.

In this talk I will present a criterion for the existence of a unique maximal strong solution for nonlinear SPDEs. The work

is inspired by the abstract criterion of Kato and Lai [1984] valid for nonlinear PDEs. The criterion is designed to fit viscous fluid

dynamics models with Stochastic Advection by Lie Transport (SALT) as introduced in Holm [2015]. As an immediate application, I show that 

the incompressible SALT 3D Navier-Stokes equation on a bounded domain has a unique maximal solution.

This is joint work with Oana Lang, Daniel Goodair and Romeo Mensah and it is partially supported by European Research Council (ERC)

Synergy project Stochastic Transport in the Upper Ocean Dynamics (https://www.imperial.ac.uk/ocean-dynamics-synergy/

Wasserstein distance induces a natural Riemannian structure for the probabilities on the Euclidean space. This insight of classical transport theory is fundamental for tremendous applications in various fields of pure and applied mathematics. We believe that an appropriate probabilistic variant, the adapted Wasserstein distance $AW$, can play a similar role for the class $FP$ of filtered processes, i.e. stochastic processes together with a filtration. In contrast to other topologies for stochastic processes, probabilistic operations such as the Doob-decomposition, optimal stopping and stochastic control are continuous w.r.t. $AW$. We also show that $(FP, AW)$ is a geodesic space, isometric to a classical Wasserstein space, and that martingales form a closed geodesically convex subspace. Finally we consider computational aspects and provide a novel method based on the Sinkhorn algorithm.

The talk is based on articles with Daniel Bartl, Mathias Beiglböck and Stephan Eckstein.

I will present a simple abstract quantitative method for proving the hydrodynamic limit of interacting particle systems on a lattice, both in the hyperbolic and parabolic scaling. In the latter case, the convergence rate is uniform in time. This "consistency-stability" approach combines a modulated Wasserstein-distance estimate comparing the law of the stochastic process to the local Gibbs measure, together with stability estimates à la Kruzhkov in weak distance, and consistency estimates exploiting the regularity of the limit solution. It avoids the use of “block estimates” and is self-contained. We apply it to the simple exclusion process, the zero range process, and the Ginzburg-Landau process with Kawasaki dynamics. This is a joint work with Daniel Marahrens and Angeliki Menegaki (IHES).

Distribution dependent equations (or McKean—Vlasov equations) have found many applications to problems in physics, biology, economics, finance and computer science. Historically, equations with either Brownian noise or zero noise have received the most attention; many well known results can be found in the monographs by A. Sznitman and F. Golse. More recently, attention has been paid to distribution dependent equations driven by random continuous noise, in particular the recent works by M. Coghi, J-D. Deuschel, P. Friz & M. Maurelli, with applications to battery modelling. Furthermore, the phenomenon of regularisation by noise has received new attention following the works of D. Davie and M. Gubinelli & R. Catellier using techniques of averaging along rough trajectories. Building on these ideas I will present recent joint work with L. Galeati and F. Harang concerning well-posedness and stability results for distribution dependent equations driven first by merely continuous noise and secondly driven by fractional Brownian motion.