arXiv subjects

Marcus Olofsson

Publications and source records attributed to Marcus Olofsson.

15 recordsLinked to original sources

arXiv2025

On the Effectiveness of Classical Regression Methods for Optimal Switching Problems

Simple regression methods provide robust, near-optimal solutions for optimal switching problems, including high-dimensional ones (up to 50). While the theory requires solving intractable PDE systems, the Longstaff-Schwartz algorithm with classical regression methods achieves excellent switching decisions without extensive hyperparameter tuning. Testing linear models (OLS, Ridge, LASSO), tree-based methods (random forests, gradient boosting), $k$-nearest neighbors, and feedforward neural networks on four benchmark problems, we find that several simple methods maintain stable performance across diverse problem characteristics, outperforming the neural networks we tested against. In our comparison, $k$-NN regression performs consistently well, and with minimal hyperparameter tuning. We establish concentration bounds for this regressor and show that PCA enables $k$-NN to scale to high dimensions.

math.OC↗

arXiv2025

Estimates for viscosity solutions of fully nonlinear equations near smooth boundaries

We reduce the problem of proving decay estimates for viscosity solutions of fully nonlinear PDEs to proving analogous estimates for solutions of one-dimensional ordinary differential inequalities. Our machinery allow the ellipticity to vanish near the boundary and permits general, possibly unbounded, lower-order terms. A key consequence is the derivation of boundary Harnack inequalities for a broad class of fully nonlinear, nonhomogeneous equations near $C^{1,1}$-boundaries. In combination with $C^{1,\alpha}$-estimates, we also obtain that quotients of positive vanishing solutions are H\"older continuous near $C^{1,1}$-boundaries.This result applies to a wide family of fully nonlinear uniformly elliptic PDEs; and for $p(x)$-harmonic functions and planar $\infty$-harmonic functions near locally flat boundaries. We end by deriving some Phragm\'en-Lindel\"of-type corollaries in unbounded domains.

math.AP↗

arXiv2023

Management strategies for hydropower plants a simple dynamic programming approach

We use a dynamic programming approach to construct management strategies for a hydropower plant with a dam and a continuously adjustable unit. Along the way, we estimate unknown variables via simple models using historical data and forecasts. Our suggested scheme achieves on average 97.1 % of the theoretical maximum using small computational effort. We also apply our scheme to a Run-of-River hydropower plant and compare the strategies and results to the much more involved PDE-based optimal switching method studied earlier by the authors in (Optimization and Engineering (2021): 1-25); this comparison shows that our simple approach may be preferable if the underlying data is sufficiently rich.

math.OC↗

arXiv2023

Enhancing Precision with the Local Pivotal Method: A General Variance Reduction Approach

The local pivotal method (LPM) is a successful sampling method for taking well-spread samples from discrete populations. We show how the LPM can be utilized to sample from arbitrary continuous distributions and thereby give powerful variance reduction in general cases. The method creates an ``automatic stratification" on any continuous distribution, of any dimension, and selects a ``thin" well-spread sample. We demonstrate the simplicity, generality and effectiveness of the LPM with various examples, including Monte Carlo estimation of integrals, option pricing and stability estimation in non-linear dynamical systems. Additionally, we show how the LPM can be combined with other variance reduction techniques, such as importance sampling, to achieve even greater variance reduction. To facilitate the implementation of the LPM, we provide a quick start guide to using LPM in MATLAB and R, which includes sample code demonstrating how to achieve variance reduction with just a few lines of code.

stat.ME↗

arXiv2022

The maximality principle in singular control with absorption and its applications to the dividend problem

Motivated by a new formulation of the classical dividend problem, we show that Peskir's maximality principle can be transferred to singular stochastic control problems with 2-dimensional degenerate dynamics and absorption along the diagonal of the state space. We construct an optimal control as a Skorokhod reflection along a moving barrier, where the barrier can be computed analytically as the smallest solution to a certain non-linear ordinary differential equation. Contrarily to the classical 1-dimensional formulation of the dividend problem, our framework produces a non-trivial solution when the firm's (pre-dividend) equity capital evolves as a geometric Brownian motion. Such solution is also qualitatively different from the one traditionally obtained for the arithmetic Brownian motion.

math.OC↗

arXiv2022

The de Finetti problem with unknown competition

We consider a resource extraction problem which extends the classical de Finetti problem for a Wiener process to include the case when a competitor, who is equipped with the possibility to extract all the remaining resources in one piece, may exist; we interpret this unknown competition as the agent being subject to possible fraud. This situation is modelled as a controller-and-stopper non-zero-sum stochastic game with incomplete information. In order to allow the fraudster to hide his existence, we consider strategies where his action time is randomised. Under these conditions, we provide a Nash equilibrium which is fully described in terms of the corresponding single-player de Finetti problem. In this equilibrium, the agent and the fraudster use singular strategies in such a way that a two-dimensional process, which represents available resources and the filtering estimate of active competition, reflects in a specific direction along a given boundary.

math.OC↗

arXiv2021

Systems of fully nonlinear parabolic obstacle problems with Neumann boundary conditions

We prove the existence of a unique viscosity solution to certain systems of fully nonlinear parabolic partial differential equations with interconnected obstacles in the setting of Neumann boundary conditions. The method of proof builds on the classical viscosity solution technique adapted to the setting of interconnected obstacles and construction of explicit viscosity sub- and supersolutions as bounds for Perron's method. Our motivation stems from so called optimal switching problems on bounded domains.

math.AP↗

arXiv2020

How to detect a salami slicer: a stochastic controller-stopper game with unknown competition

We consider a stochastic game of control and stopping specified in terms of a process $X_t=-θΛ_t+W_t$, representing the holdings of Player 1, where $W$ is a Brownian motion, $θ$ is a Bernoulli random variable indicating whether Player 2 is active or not, and $Λ$ is a non-decreasing process representing the accumulated "theft" or "fraud" performed by Player 2 (if active) against Player 1. Player 1 cannot observe $θ$ or $Λ$ directly, but can merely observe the path of the process $X$ and may choose a stopping rule $τ$ to deactivate Player 2 at a cost $M$. Player 1 thus does not know if she is the victim of fraud and operates in this sense under unknown competition. Player 2 can observe both $θ$ and $W$ and seeks to choose the fraud strategy $Λ$ that maximizes the expected discounted amount \[{\mathbb E} \left [θ\int _0^τ e^{-rs} dΛ_s \right ],\] whereas Player 1 seeks to choose the stopping strategy $τ$ so as to minimize the expected discounted cost \[{\mathbb E} \left [θ\int _0^τ e^{-rs} dΛ_s + e^{-rτ}M{\mathbb I}_{\{τ<\infty\}} \right ].\] This non-zero-sum game appears to be novel and is motivated by applications in fraud detection; it combines filtering (detection), non-singular control, stopping, strategic features (games) and asymmetric information. We derive Nash equilibria for this game; for some parameter values we find an equilibrium in pure strategies, and for other parameter values we find an equilibrium by allowing for randomized stopping strategies.

math.OC↗

arXiv2020

Management strategies for run-of-river hydropower plants -an optimal switching approach

The mathematical theory for optimal switching is by now relatively well developed, but the number of concrete applications of this theoretical framework remains few. In this paper, we bridge parts of this gap by applying optimal switching theory to a set of production planning problems related to hydropower plants. In particular, we study two different cases involving small run-of-river hydropower plants and show how optimal switching can be used to create fully automatic production schemes in these cases, with non-zero cost of switching between different states of production. Along the way of deriving these schemes, we also create a model for the random flow of water based on stochastic differential equations and fit this model to historical data. This stochastic flow model, which should be of independent interest, mimics the long term seasonal behaviour of the flow while still allowing for stochastic fluctuations and can incorporate a given forecast to damp the impact of such fluctuations in near time. We benchmark the performance of our model using actual flow data from a small river in Sweden and find that our production scheme lies close to the optimal, within 2 % and 5 %, respectively, in a long term investigation of the two plants considered.

math.OC↗

arXiv2020

Strong maximum principle and boundary estimates for nonhomogeneous elliptic equations

We give a simple proof of the strong maximum principle for viscosity subsolutions of fully nonlinear elliptic PDEs on the form $$ F(x,u,Du,D^2u) = 0 $$ under suitable structure conditions on the equation allowing for non-Lipschitz growth in the gradient terms. In case of smooth boundaries, we also prove the Hopf lemma, the boundary Harnack inequality and that positive viscosity solutions vanishing on a portion of the boundary are comparable with the distance function near the boundary. Our results apply to weak solutions of an eigenvalue problem for the variable exponent $p$-Laplacian.

math.AP↗

arXiv2017

Existence, uniqueness and regularity of solutions to systems of nonlocal obstacle problems related to optimal switching

We study viscosity solutions to a system of nonlinear degenerate parabolic partial integro-differential equations with interconnected obstacles. This type of problem occurs in the context of optimal switching problems when the dynamics of the underlying state variable is described by an $n$-dimensional Levy process. We first establish a continuous dependence estimate for viscosity sub- and supersolutions to the system under mild regularity, growth and structural assumptions on the partial integro-differential operator and on the obstacles and terminal conditions. Using the continuous dependence estimate, we obtain the comparison principle and uniqueness of viscosity solutions as well as Lipschitz regularity in the spatial variables. Our main contribution is construction of suitable families of viscosity sub- and supersolutions which we use as barrier functions to prove Hölder continuity in the time variable, and, through Perron's method, existence of a unique viscosity solution. This paper generalizes parts of the results of Biswas, Jakobsen and Karlsen (2010) and of Lundström, Nyström and Olofsson (2014) to hold for more general systems of equations.

math.AP↗

arXiv2015

A Brownian optimal switching problem under incomplete information

In this paper we study an incomplete information optimal switching problem in which the manager only has access to noisy observations of the underlying Brownian motion $\{W_t\}_{t \geq 0}$. The manager can, at a fixed cost, switch between having the production facility open or closed and must find the optimal management strategy using only the noisy observations. Using the theory of linear stochastic filtering, we reduce the incomplete information problem to a full information problem, show that the value function is non-decreasing with the amount of information available, and that the value function of the incomplete information problem converges to the value function of the corresponding full information problem as the noise in the observed process tends to $0$.

math.OC↗

arXiv2015

Reflected BSDE of Wiener-Poisson type in Time-dependent Domains

In this paper we study multi-dimensional reflected backward stochastic differential equations driven by Wiener-Poisson type processes. We prove existence and uniqueness of solutions, with reflection in the inward spatial normal direction, in the setting of certain time-dependent domains.

math.PR↗

arXiv2014

Optimal Switching Problems under Partial Information

In this paper we formulate and study an optimal switching problem under partial information. In our model the agent/manager/investor attempts to maximize the expected reward by switching between different states/investments. However, he is not fully aware of his environment and only an observation process, which contains partial information about the environment/underlying, is accessible. It is based on the partial information carried by this observation process that all decisions must be made. We propose a probabilistic numerical algorithm based on dynamic programming, regression Monte Carlo methods, and stochastic filtering theory to compute the value function. In this paper, the approximation of the value function and the corresponding convergence result are obtained when the underlying and observation processes satisfy the linear Kalman-Bucy setting. A numerical example is included to show some specific features of partial information.

math.OC↗

arXiv2013

Systems of variational inequalities for non-local operators related to optimal switching problems: Existence and uniqueness

In this paper we study a system of variational inequalities where the operator is non-local, possibly degenerate and of second order. A special case of this type of problem occurs in the context of optimal switching problems when the dynamics of the underlying state variables is described by an N-dimensional Levy process. We establish a general comparison principle for viscosity sub- and supersolutions to the system under mild regularity, growth and structural assumptions on the data. Using the comparison principle we then prove the existence of a unique viscosity solution to the system by Perron's method. Our main contribution is that we establish existence and uniqueness of viscosity solutions, in the setting of Levy processes and non-local operators, with no sign assumption on the switching costs and allowing them to depend on x as well as t.

math.OC↗