arXiv subjects

Thomas Kruse

Publications and source records attributed to Thomas Kruse.

At least 19 recordsLinked to original sources

arXiv2026

Walk-on-Spheres Monte Carlo and deep neural network approximations of elliptic PDEs with drift and killing

In this paper we provide Monte Carlo and deep neural network approximations for stochastic representations of solutions to linear elliptic partial differential equations with constant diffusion, drift and killing. Building on the modified Walk-on-Spheres algorithm of Beznea et al. (arXiv:2209.01432), we introduce Monte Carlo estimators that explicitly incorporate sampled random times arising in the analyzed stochastic representations. We establish uniform error bounds for these estimators and show that, under suitable assumptions, a prescribed approximation accuracy is achieved with sample complexities growing at most polynomially in both the inverse accuracy and the problem dimension. Furthermore, we prove a deep neural network approximation result for the stochastic representations. Assuming suitable neural network representations of the boundary data and the distance function to the boundary, we use the constructed Monte Carlo to design deep neural networks that approximate the representation uniformly with a number of parameters growing at most polynomially in the inverse accuracy and the problem dimension. These results extend previous complexity analyses to a broader class of elliptic equations involving drift and killing.

math.NA↗

arXiv2026

Path-dependent Affine Processes

We extend the classical theory of affine processes to a path-dependent setting by introducing path-dependent coefficients and provide analytic formulas for their Fourier--Laplace transform in terms of generalized Riccati-type equations. In the proposed framework, we define path-dependent affine processes through their exponential-affine Fourier--Laplace transform on the path space and establish a characterization theorem. Conversely, for path-dependent stochastic differential equations with affine path-dependent coefficients, we also provide explicit exponential-affine representations of the Fourier--Laplace functional in terms of those Riccati equations. Moreover, we derive a condition ensuring non-negativity of the path-dependent diffusion coefficient, guaranteeing well-posedness of the model. Finally, we apply these results to a path-dependent volatility model and a path-dependent extension of the Heston model, including a delayed Heston model as a special case.

math.PR↗

arXiv2026

Optimal Asymptotic Rates for (Stochastic) Gradient Descent under the Local PL-Condition: A Geometric Approach

Stochastic gradient descent (SGD) has been studied extensively over the past decades due to its simplicity and broad applicability in machine learning. In this work, we analyze the local behavior of gradient descent and stochastic gradient descent for minimizing $C^2$-functions that satisfy the Polyak-Lojasiewicz (PL) inequality and under a multiplicative gradient noise model motivated by overparameterized neural networks. Using a geometric interpretation of the PL-condition, we prove a simple yet surprising fact: in this possibly non-convex setting, the asymptotic convergence rate of (S)GD matches the rate obtained for strongly convex quadratics.

math.OC↗

arXiv2026

Time discretization of BSDEs with singular terminal condition using asymptotic expansion

We consider a class of backward stochastic differential equations (BSDEs) with singular terminal condition and develop a numerical scheme to approximate their solution. To this end, we extend an asymptotic development of the BSDE solution known from the power case, which arises from optimal liquidation problems, to more general generators. This expansion allows to obtain a suitable approximation of the BSDE solution close to the terminal time. Using this as a terminal condition, we analyze the error of a backward Euler implicit scheme and detail its dependence on the terminal condition.

math.OC↗

arXiv2026

Matrix Riccati BSDEs with singular terminal condition and stochastic LQ control with linear terminal constraint

We analyze a class of multidimensional linear-quadratic stochastic control problems with random coefficients, motivated by multi-asset optimal trade execution. The problems feature non-diffusive controlled state dynamics and a terminal constraint that restricts the terminal state to a prescribed random linear subspace. We derive the associated Riccati backward stochastic differential equation (BSDE) and identify a suitable formalization of its singular terminal condition. Via a penalization approach, we establish existence of a minimal supersolution of the Riccati BSDE and use it to characterize both the value function and the optimal control. We analyze the asymptotic behavior of the supersolution near terminal time and discuss special cases where closed-form solutions can be obtained.

math.OC↗

arXiv2025

Stochastic Passivity in Stochastic Differential Equations: A Port-Hamiltonian Perspective

We extend deterministic port-Hamiltonian systems (PHS) to a stochastic framework by means of stochastic differential equations. As the dissipation inequality plays a crucial role for deterministic PHS, we develop several passivity concepts for stochastic input-state-output systems and characterize these in terms of the parameters of the system. Afterwards, we examine properties of a certain class of linear stochastic systems that can be regarded as an extension of linear deterministic PHS to a stochastic passivity framework.

math.OC↗

arXiv2025

Convexity and strict convexity for compositional neural networks in high-dimensional optimal control

Neural networks (NNs) have emerged as powerful tools for solving high-dimensional optimal control problems. In particular, their compositional structure has been shown to enable efficient approximation of high-dimensional functions, helping to mitigate the curse of dimensionality in optimal control problems. In this work, we build upon the theoretical framework developed by Kang & Gong (SIAM J. Control Optim. 60(2):786-813, 2022), particularly their results on NN approximations for compositional functions in optimal control. Theorem 6.2 in Kang & Gong (SIAM J. Control Optim. 60(2):786-813, 2022) establishes that, under suitable assumptions on the compositional structure and its associated features, optimal control problems with strictly convex cost functionals admit a curse-of-dimensionality-free approximation of the optimal control by NNs. We extend this result in two directions. First, we analyze the strict convexity requirement on the cost functional and demonstrate that reformulating a discrete-time optimal control problem with linear transitions and stage costs as a terminal cost problem ensures the necessary strict convexity. Second, we establish a generalization of Theorem 6.2 in Kang & Gong (SIAM J. Control Optim. 60(2):786-813, 2022) which provides weak error bounds for optimal control approximations by NNs when the cost functional is only convex rather than strictly convex.

math.OC↗

arXiv2025

Deep neural networks can provably solve Bellman equations for Markov decision processes without the curse of dimensionality

Discrete time stochastic optimal control problems and Markov decision processes (MDPs) are fundamental models for sequential decision-making under uncertainty and as such provide the mathematical framework underlying reinforcement learning theory. A central tool for solving MDPs is the Bellman equation and its solution, the so-called $Q$-function. In this article, we construct deep neural network (DNN) approximations for $Q$-functions associated to MDPs with infinite time horizon and finite control set $A$. More specifically, we show that if the the payoff function and the random transition dynamics of the MDP can be suitably approximated by DNNs with leaky rectified linear unit (ReLU) activation, then the solutions $Q_d\colon \mathbb R^d\to \mathbb R^{|A|}$, $d\in \mathbb{N}$, of the associated Bellman equations can also be approximated in the $L^2$-sense by DNNs with leaky ReLU activation whose numbers of parameters grow at most polynomially in both the dimension $d\in \mathbb{N}$ of the state space and the reciprocal $1/\varepsilon$ of the prescribed error $\varepsilon\in (0,1)$. Our proof relies on the recently introduced full-history recursive multilevel fixed-point (MLFP) approximation scheme.

math.OC↗

arXiv2025

Learning Brenier Potentials with Convex Generative Adversarial Neural Networks

Brenier proved that under certain conditions on a source and a target probability measure there exists a strictly convex function such that its gradient is a transport map from the source to the target distribution. This function is called the Brenier potential. Furthermore, detailed information on the H\"older regularity of the Brenier potential is available. In this work we develop the statistical learning theory of generative adversarial neural networks that learn the Brenier potential. As by the transformation of densities formula, the density of the generated measure depends on the second derivative of the Brenier potential, we develop the universal approximation theory of ReCU networks with cubic activation $\mathtt{ReCU}(x)=\max\{0,x\}^3$ that combines the favorable approximation properties of H\"older functions with a Lipschitz continuous density. In order to assure the convexity of such general networks, we introduce an adversarial training procedure for a potential function represented by the ReCU networks that combines the classical discriminator cross entropy loss with a penalty term that enforces (strict) convexity. We give a detailed decomposition of learning errors and show that for a suitable high penalty parameter all networks chosen in the adversarial min-max optimization problem are strictly convex. This is further exploited to prove the consistency of the learning procedure for (slowly) expanding network capacity. We also implement the described learning algorithm and apply it to a number of standard test cases from Gaussian mixture to image data as target distributions. As predicted in theory, we observe that the convexity loss becomes inactive during the training process and the potentials represented by the neural networks have learned convexity.

cs.LG↗

arXiv2025

Multi-asset optimal trade execution with stochastic cross-effects: An Obizhaeva-Wang-type framework

We analyze a continuous-time optimal trade execution problem in multiple assets where the price impact and the resilience can be matrix-valued stochastic processes that incorporate cross-impact effects. In addition, we allow for stochastic terminal and running targets. Initially, we formulate the optimal trade execution task as a stochastic control problem with a finite-variation control process that acts as an integrator both in the state dynamics and in the cost functional. We then extend this problem continuously to a stochastic control problem with progressively measurable controls. By identifying this extended problem as equivalent to a certain linear-quadratic stochastic control problem, we can use established results in linear-quadratic stochastic control to solve the extended problem. This work generalizes [Ackermann, Kruse, Urusov; FinancStoch'24] from the single-asset setting to the multi-asset case. In particular, we reveal cross-hedging effects, showing that it can be optimal to trade in an asset despite having no initial position. Moreover, as a subsetting we discuss a multi-asset variant of the model in [Obizhaeva, Wang; JFinancMark'13].

math.OC↗

arXiv2025

Modelling car-following dynamics with stochastic input-state-output port-Hamiltonian systems

In this contribution, we introduce a general class of car-following models with an input-state-output port-Hamiltonian structure. We derive stability conditions and long-term behavior of the finite system with periodic boundaries and quadratic interaction potential by spectral analysis and using asymptotic properties of multivariate Ornstein-Uhlenbeck processes. The uncontrolled dynamics exhibit instability and random collective behavior under stochastic perturbations. By implementing an open-loop speed control, the system stabilizes and weakly converges to Gaussian limit distributions. The convergence is unconditional for constant speed control. However, a stability condition arises for the closed-loop system where the speed control acts as a dynamic feedback depending on the distance ahead. The results are illustrated by numerical simulations. Interestingly, only the closed-loop system is able to reproduce, at least transiently, realistic stop-and-go behavior that can be resolved using the Hamiltonian component of the model.

math.DS↗

arXiv2024

Stabilisation of stochastic single-file dynamics using port-Hamiltonian systems

This study revisits a recently proposed symmetric port-Hamiltonian single-file model in one dimension. The uniform streaming solutions are stable in the deterministic model. However, the introduction of white noise into the dynamics causes the model to exhibit divergence. In response, we introduce a control term in a port-Hamiltonian framework. Our results show that this control term effectively stabilises the dynamics even in the presence of noise, providing valuable insights for the control of road traffic flows.

math.DS↗

arXiv2023

Reducing Obizhaeva-Wang type trade execution problems to LQ stochastic control problems

We start with a stochastic control problem where the control process is of finite variation (possibly with jumps) and acts as integrator both in the state dynamics and in the target functional. Problems of such type arise in the stream of literature on optimal trade execution pioneered by Obizhaeva and Wang (models with finite resilience). We consider a general framework where the price impact and the resilience are stochastic processes. Both are allowed to have diffusive components. First we continuously extend the problem from processes of finite variation to progressively measurable processes. Then we reduce the extended problem to a linear quadratic (LQ) stochastic control problem. Using the well developed theory on LQ problems we describe the solution to the obtained LQ one and trace it back up to the solution to the (extended) initial trade execution problem. Finally, we illustrate our results by several examples. Among other things the examples show the Obizhaeva-Wang model with random (terminal and moving) targets, the necessity to extend the initial trade execution problem to a reasonably large class of progressively measurable processes (even going beyond semimartingales) and the effects of diffusive components in the price impact process and/or in the resilience process.

math.OC↗

arXiv2023

Deep neural networks with ReLU, leaky ReLU, and softplus activation provably overcome the curse of dimensionality for Kolmogorov partial differential equations with Lipschitz nonlinearities in the $L^p$-sense

Recently, several deep learning (DL) methods for approximating high-dimensional partial differential equations (PDEs) have been proposed. The interest that these methods have generated in the literature is in large part due to simulations which appear to demonstrate that such DL methods have the capacity to overcome the curse of dimensionality (COD) for PDEs in the sense that the number of computational operations they require to achieve a certain approximation accuracy $\varepsilon\in(0,\infty)$ grows at most polynomially in the PDE dimension $d\in\mathbb N$ and the reciprocal of $\varepsilon$. While there is thus far no mathematical result that proves that one of such methods is indeed capable of overcoming the COD, there are now a number of rigorous results in the literature that show that deep neural networks (DNNs) have the expressive power to approximate PDE solutions without the COD in the sense that the number of parameters used to describe the approximating DNN grows at most polynomially in both the PDE dimension $d\in\mathbb N$ and the reciprocal of the approximation accuracy $\varepsilon>0$. Roughly speaking, in the literature it is has been proved for every $T>0$ that solutions $u_d\colon [0,T]\times\mathbb R^d\to \mathbb R$, $d\in\mathbb N$, of semilinear heat PDEs with Lipschitz continuous nonlinearities can be approximated by DNNs with ReLU activation at the terminal time in the $L^2$-sense without the COD provided that the initial value functions $\mathbb R^d\ni x\mapsto u_d(0,x)\in\mathbb R$, $d\in\mathbb N$, can be approximated by ReLU DNNs without the COD. It is the key contribution of this work to generalize this result by establishing this statement in the $L^p$-sense with $p\in(0,\infty)$ and by allowing the activation function to be more general covering the ReLU, the leaky ReLU, and the softplus activation functions as special cases.

math.NA↗

arXiv2023

The Collective Dynamics of a Stochastic Port-Hamiltonian Self-Driven Agent Model in One Dimension

The collective motion of self-driven agents is a phenomenon of great interest in interacting particle systems. In this paper, we develop and analyze a model of agent motion in one dimension with periodic boundaries using a stochastic port-Hamiltonian system (PHS). The interaction model is symmetric and based on nearest neighbors. The distance-based terms and kinematic relations correspond to the skew-symmetric Hamiltonian structure of the PHS, while the velocity difference terms make the system dissipative. The originality of the approaches lies in the stochastic noise that plays the role of an external input. It turns out that the stochastic PHS with a quadratic potential is an Ornstein-Uhlenbeck process, for which we explicitly determine the distribution for any time $t\ge 0$ and in the limit $t\to\infty$. We characterize the collective motion by showing that the agents' mean velocity is Brownian noise whose fluctuations increase with time, while the variance of the agents' velocities and distances, which quantify the coordination of the agents' motion, converge. The motion model does not specify a preferred direction of motion. However, assuming a equilibrium uniform starting configuration, the results show that the noise triggers rapidly coordinated agent motion determined by the Brownian behavior of the mean velocity. Interestingly, simulation results show that some theoretical properties obtained with the Ornstein-Uhlenbeck process also hold for the nonlinear model with general interaction potential.

math.DS↗

arXiv2023

Nonlinear Monte Carlo methods with polynomial runtime for Bellman equations of discrete time high-dimensional stochastic optimal control problems

Discrete time stochastic optimal control problems and Markov decision processes (MDPs), respectively, serve as fundamental models for problems that involve sequential decision making under uncertainty and as such constitute the theoretical foundation of reinforcement learning. In this article we study the numerical approximation of MDPs with infinite time horizon, finite control set, and general state spaces. Our set-up in particular covers infinite-horizon optimal stopping problems of discrete time Markov processes. A key tool to solve MDPs are Bellman equations which characterize the value functions of the MDPs and determine the optimal control strategies. By combining ideas from the full-history recursive multilevel Picard approximation method, which was recently introduced to solve certain nonlinear partial differential equations, and ideas from $Q$-learning we introduce a class of suitable nonlinear Monte Carlo methods and prove that the proposed methods do overcome the curse of dimensionality in the numerical approximation of the solutions of Bellman equations and the associated discrete time stochastic optimal control problems.

math.OC↗

arXiv2022

Self-exciting price impact via negative resilience in stochastic order books

Most of the existing literature on optimal trade execution in limit order book models assumes that resilience is positive. But negative resilience also has a natural interpretation, as it models self-exciting behaviour of the price impact, where trading activities of the large investor stimulate other market participants to trade in the same direction. In the paper we discuss several new qualitative effects on optimal trade execution that arise when we allow resilience to take negative values. We do this in a framework where both market depth and resilience are stochastic processes.

q-fin.TR↗

arXiv2022

On the speed of convergence of Picard iterations of backward stochastic differential equations

It is a well-established fact in the scientific literature that Picard iterations of backward stochastic differential equations with globally Lipschitz continuous nonlinearity converge at least exponentially fast to the solution. In this paper we prove that this convergence is in fact at least square-root factorially fast. We show for one example that no higher convergence speed is possible in general. Moreover, if the nonlinearity is $z$-independent, then the convergence is even factorially fast. Thus we reveal a phase transition in the speed of convergence of Picard iterations of backward stochastic differential equations.

math.PR↗