A Neural Network Approach to High-Dimensional Optimal Switching Problems with Jumps in Energy Markets

Erhan Bayraktar; Asaf Cohen; April Nellis

doi:10.1137/22M1527246

A Neural Network Approach to High-Dimensional Optimal Switching Problems with Jumps in Energy Markets

Erhan Bayraktar, Asaf Cohen, April Nellis

Research output: Contribution to journal › Article › peer-review

2 Scopus citations

Abstract

We develop a backward-in-time machine learning algorithm that uses a sequence of neural networks to solve optimal switching problems in energy production, where electricity and fossil fuel prices are subject to stochastic jumps. We then apply this algorithm to a variety of energy scheduling problems, including novel high-dimensional energy production problems. Our experimental results demonstrate that the algorithm performs with accuracy and experiences linear to sublinear slowdowns as dimension increases, demonstrating the value of the algorithm for solving high-dimensional switching problems.

Original language	English
Pages (from-to)	1028-1061
Number of pages	34
Journal	SIAM Journal on Financial Mathematics
Volume	14
Issue number	4
DOIs	https://doi.org/10.1137/22M1527246
State	Published - 1 Dec 2023
Externally published	Yes

Keywords

deep neural networks
forward-backward systems of stochastic differential equations
Monte Carlo algorithm
optimal investment in power generation
optimal switching
planning problems

ASJC Scopus subject areas

Numerical Analysis
Finance
Applied Mathematics

Access to Document

10.1137/22M1527246

Cite this

@article{55a62b033747483ba9d1ae34668bd3f4,

title = "A Neural Network Approach to High-Dimensional Optimal Switching Problems with Jumps in Energy Markets",

abstract = "We develop a backward-in-time machine learning algorithm that uses a sequence of neural networks to solve optimal switching problems in energy production, where electricity and fossil fuel prices are subject to stochastic jumps. We then apply this algorithm to a variety of energy scheduling problems, including novel high-dimensional energy production problems. Our experimental results demonstrate that the algorithm performs with accuracy and experiences linear to sublinear slowdowns as dimension increases, demonstrating the value of the algorithm for solving high-dimensional switching problems.",

keywords = "deep neural networks, forward-backward systems of stochastic differential equations, Monte Carlo algorithm, optimal investment in power generation, optimal switching, planning problems",

author = "Erhan Bayraktar and Asaf Cohen and April Nellis",

note = "Publisher Copyright: {\textcopyright} 2023 Erhan Bayraktar, Asaf Cohen, April Nellis.",

year = "2023",

month = dec,

day = "1",

doi = "10.1137/22M1527246",

language = "English",

volume = "14",

pages = "1028--1061",

journal = "SIAM Journal on Financial Mathematics",

issn = "1945-497X",

publisher = "Society for Industrial and Applied Mathematics Publications",

number = "4",

}

TY - JOUR

T1 - A Neural Network Approach to High-Dimensional Optimal Switching Problems with Jumps in Energy Markets

AU - Bayraktar, Erhan

AU - Cohen, Asaf

AU - Nellis, April

PY - 2023/12/1

Y1 - 2023/12/1

N2 - We develop a backward-in-time machine learning algorithm that uses a sequence of neural networks to solve optimal switching problems in energy production, where electricity and fossil fuel prices are subject to stochastic jumps. We then apply this algorithm to a variety of energy scheduling problems, including novel high-dimensional energy production problems. Our experimental results demonstrate that the algorithm performs with accuracy and experiences linear to sublinear slowdowns as dimension increases, demonstrating the value of the algorithm for solving high-dimensional switching problems.

AB - We develop a backward-in-time machine learning algorithm that uses a sequence of neural networks to solve optimal switching problems in energy production, where electricity and fossil fuel prices are subject to stochastic jumps. We then apply this algorithm to a variety of energy scheduling problems, including novel high-dimensional energy production problems. Our experimental results demonstrate that the algorithm performs with accuracy and experiences linear to sublinear slowdowns as dimension increases, demonstrating the value of the algorithm for solving high-dimensional switching problems.

KW - deep neural networks

KW - forward-backward systems of stochastic differential equations

KW - Monte Carlo algorithm

KW - optimal investment in power generation

KW - optimal switching

KW - planning problems

UR - http://www.scopus.com/inward/record.url?scp=85175546242&partnerID=8YFLogxK

U2 - 10.1137/22M1527246

DO - 10.1137/22M1527246

M3 - Article

AN - SCOPUS:85175546242

SN - 1945-497X

VL - 14

SP - 1028

EP - 1061

JO - SIAM Journal on Financial Mathematics

JF - SIAM Journal on Financial Mathematics

IS - 4

ER -

A Neural Network Approach to High-Dimensional Optimal Switching Problems with Jumps in Energy Markets

Abstract

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this