Paper: arXiv 2501.09727
Authors: Alessandro Gnoatto, Katharina Oberpriller, Athena Picarelli
Abstract
We study the error arising in the numerical approximation of FBSDEs and related PIDEs by means of a deep learning-based method. Our results focus on decoupled FBSDEs with jumps and extend the seminal work of HAn and Long (2020) analyzing the numerical error of the deep BSDE solver proposed in E et al. (2017). We provide a priori and a posteriori error estimates for the finite and infinite activity case.
Complexity vs Empirical Score
- Math Complexity: 9.2/10
- Empirical Rigor: 3.5/10
- Quadrant: Lab Rats — high math complexity, limited empirical testing
Why this score: The paper presents highly advanced stochastic analysis, PIDE theory, and deep learning convergence proofs, reflecting deep mathematical complexity. It lacks any reported backtests, numerical experiments, or implementation details, focusing solely on theoretical error estimates.
Research Flowchart
flowchart TD
A["<b>Research Goal:</b><br>Estimation of numerical errors in<br>Deep BSDE solver for FBSDEs with jumps"] --> B["<b>Methodology:</b><br>Extended analysis of HAn & Long (2020)<br>Applied to decoupled FBSDEs"]
B --> C["<b>Inputs/Targets:</b><br>Finite & Infinite Activity Jump Processes<br>Related Parabolic Integro-Differential Eqs (PIDEs)"]
C --> D["<b>Computational Process:</b><br>Deep Learning-based BSDE Solver<br>(E et al. 2017 framework)"]
D --> E
subgraph E ["Outcomes"]
F["<b>A Priori Error Estimates</b>"]
G["<b>A Posteriori Error Estimates</b>"]
end
Evidence Card
What the authors claim, on what data, and who has checked it. Read from the paper's own text by a language model: every quote is verbatim, nothing was re-run. How to read this card
Proposes
The paper proposes a deep learning‑based algorithm for solving decoupled forward‑backward SDEs with jumps and derives a priori and a posteriori error estimates for both finite and infinite jump activity cases.
Source sentence
We study the error arising in the numerical approximation of FBSDEs and related PIDEs by means of a deep learning-based method. Our results focus on decoupled FBSDEs with jumps and extend the seminal work of HAn and Long (2020) analyzing the numerical error of the deep BSDE solver proposed in E et al. (2017).
Assumes 3
- The underlying probability space is filtered, supports a standard Rd‑valued Brownian motion and a Poisson random measure with Lévy measure ν satisfying ν({0})=0 and integrability conditions on |z|². model
Source sentence
Setting and Preliminaries Given a fixed time horizon T < ∞, let (Ω, F, F, P) be a filtered probability space with the filtration F = (Ft)t∈[0,T] satisfying the usual conditions. We assume that the filtered probability space supports an Rd-valued standard Brownian motion W = (Wt)t∈[0,T] and a Poisson random measure N with associated L´evy measure ν, such that ν({0}) = 0, Z Rd(1 ∧|z|2)ν(dz) < ∞, and Z |z|≥1 |z|2ν(dz) < ∞.
- The coefficients b, σ, Γ, f, g satisfy the regularity and growth conditions collected in Assumption 2.1 (A1) and (A3). model
Source sentence
Note that if the function Γ : Rd × Rd →Rd satisfies Assumption 2.1 (A3) the following estimates hold for x ∈Rd |γ(x)| ≤|γ(0)| + Kγ|x| ≤eCγ(1 + |x|), |γ(x)|2 ≤2|γ(0)|2 + 2K2 γ|x|2 ≤Cγ(1 + |x|2) with eCγ := 2(|γ(0)| ∨Kγ) and Cγ := 2(|γ(0)|2 ∨K2 γ). Let Assumptions 2.1 (A1), (A3), 3.1 and 3.3 hold.
- Neural network approximations Uρn satisfy measurability, growth and Lipschitz‑type gradient bounds (Assumption 3.3). other
Source sentence
We assume that for each n = 0, ..., M −1 the function Uρn n : Rd →Rd is measurable and satisfies the following growth condition (3.6) Uρn n (x) 2 ≤An + Bn|x|2 for constants An, Bn≥0. Moreover, for each n = 0, ..., M −1 the function Uρn n : Rd →Rd is differentiable and the gradient DxUρn n (x) satisfies the following growth condition (3.7) DxUρn n (x) 2 ≤Cn for a constant Cn≥0. Remark 3.4.
Data
No dataset named in the text.
Costs
Not applicable
Finds 3
- A priori error estimates are obtained for the deep BSDE solver in both finite and infinite jump activity regimes. theoretical
Source sentence
We provide a priori and a posteriori error estimates for the finite and infinite activity case. Section 4 provides a priori estimates both for the finite and infinite activity case. 2.
- A posteriori error estimates are derived, extending previous results and covering infinite activity jumps. theoretical
Source sentence
Our analysis of a posteriori estimates (which excludes weak coupling) is in line with that of [30] but we also consider infinite activity jumps and, more importantly, we obtain a priori estimates which are missing in [30]. In Section 3 we provide a self contained derivation of a posteriori estimates that is in line with the results of [30] but is extended to consider the infinite activity case in Section 3.2.
- Path‑regularity of the discretized solution is established, showing an overall L²‑error of order Δt. theoretical
Source sentence
Then the following path regularity holds max n=0,...,M−1E " sup t∈[tn,tn+1] Xt −ee Xt 2 # + max n=0,...,M−1 E " sup t∈[tn,tn+1] Yt −eeY t 2 # + E Z T 0 Zt −eeZt 2dt + E Z T 0 Ψt −eeΨt 2dt ≤C∆t, (4.1) where C is a constant only depending on the data of the FBSDE in (2.3)-(2.4) and the processes ee X, eeY , eeZ, eeΨ are defined on [tn, tn+1) by ee Xt := Xn, eeY t := Yn, eeZt := 1 ∆tE Z tn+1 tn Zsds Fn ,
Checked by others
No other paper in the archive has been found to support or contradict these findings (compared with 12 related papers). Coverage is partial, so this is neither replication nor refutation.
3 papers in the archive : INEUS: Iterative Neural Solver for High-Dimensional PIDEs
Code
No repository by the authors found in the paper.
Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.
Cited References (67)
- Arbitrage-Free Pricing of XVA – Part I: Framework and Explicit Examples (2015) — link
- Malliavin calculus in the Canonical L 'e vy process: White noise theory and financial applications (2015) — link
- Canonical L'evy process and Malliavin calculus (2007) — link
- Backward Stochastic Differential Equations (2017)
- Adapted solution of a backward stochastic differential equation (1990) — link
- Option pricing when underlying stock returns are discontinuous (1976)
- Transform Analysis and Option Pricing for Affine Jump-Diffusions (2000)
- The Fine Structure of Asset Returns: An Empirical Investigation (2002) — link
and 59 more
- Computational methods for quantitative finance (2013)
- Financial Modelling with Jump Processes (2003)
- The C omprehensive T ex A rchive N etwork ( CTAN ) (1993)
- Pricing interest rate derivatives under volatility uncertainty (2020)
- Term Structure Modeling under Volatility Uncertainty (2022)
- The H ull- W hite Model under volatility uncertainty (2021)
- Default Ambiguity (2019)
- Affine processes under parameter uncertainty (2019)
- Nonlinear L ' e vy processes and their characteristics (2016)
- Measurability of semimartingale characteristics with respect to the probability law (2014)
- Towards a general theory of bond markets (1997)
- The equivalent martingale measure conditions in a general model for interest rates (2005)
- Consistency P roblems for H eath- J arrow- M orton I nterest R ate M odels (2001)
- Stochastic P D E s and term structure models (1993)
- Interest- R ate M odels: an I nfinite- D imensional S tochastic A nalysis P erspective (2006)
- Non-linear A ffine P rocesses with J umps (2023)
- Non-Linear Continuous Semimartingales (2022) — link
- Classical and deep pricing for path- dependent options in non-linear generalized affine models (2022) — link
- Signature-based models: theory and calibration (2023)
- Constructing sublinear expectations on path space (2013)
- Joint callibration to SPX and VIX options with signature-based models — → 2301.13235
- Conversation of the ump-resp. maxmin-property of statistical tests under extensions of probability measures (1984)
- Time-consistent and market-consistent evaluations (2014)
- A BENCHMARK APPROACH TO FINANCE (2006)
- Local risk-minimization under the benchmark appraoch (2014)
- BENCHMARKED RISK MINIMIZATION (2016)
- Actuarial-consistency and two-step actuarial valuations: a new paradigm to insurance valuation (2023)
- Fair valuation of insurance liability cash-flow streams in continuous time: Theory (2019)
- Max-Min optimization problem for variable annuities pricing (2015)
- Indifference fee rate for variable annuities (2016)
- Risk-minimizing hedging strategies for insurance payment streams (2001)
- Market-consistent actuarial valuation (2007)
- Insurance-Finance arbitrage (2022) — link
- A Unified Approach to xVA with CSA Discounting and Initial Margin (2021)
- Convergence of the deep B S D E method for coupled F B S D E s (2020)
- Time discretization and M arkovian iteration for coupled F B S D E s (2008)
- Backward S tochastic D ifferential E quations with J umps and T heir A ctuarial and F inancial A pplications (2017)
- Discrete-time approximation of decoupled F orward- B ackward S D E with jumps (2008)
- Contr^ o le stochastique et m'ethods num'eriques en finance math'ematique (2006)
- A deep solver for B S D E s with jumps (2024) — link
- Limit Theorems for S tochastic P rocesses (2003)
- Measures, I ntegrals and M artingales (2005)
- A deep B S D E approximation of nonlinear I ntegro- P D E s with unbounded nonlocal operators (2024) — link
- Viscosity solutions of second order integral-partial differential equations without monotonicity condition: A new result (2016)
- Backward stochastic differential equations and integral-partial differential equatioins (1997)
- Appoximation of small jumps of L'evy processes with a view towards simulation (2001)
- Gaussian approximation of multivariate L'evy processes with applications to simulation of tempered stable processes (2007)
- Numerical A pproximation of S tochastic D ifferential E quations D riven by L evy M otion with I nfinitely M any J umps (2015)
- L'evy P rocesses and S tochastic C alculus (2011)
- Deep learning-based numerical methods for high-dimensional parabolic partial differential equations and backward stochastic differential equations (2017)
- Integro-differential equations for option prices in exponential L'evy prices (2005)
- Deep backward schemes for high-dimensional nonlinear P D E s (2020)
- Deep Splitting Method for Parabolic P D E s (2021) — link
- D G M : A deep learning algorithm for solving partial differential equations (2018) — link
- Deep learning numerical methods for high-dimensional fully non-linear P I D E s and coupled F B S D E s with jumps (2023) — link
- Deep learning schemes for parabolic nonlocal integro-differential equations (2022) — link
- Convergence Analysis of the Deep Splitting Scheme: the Case of Partial Integro-Differential Equations and the associated FBSDEs with Jumps (2022) — link
- Deep Neural Network Algorithms for Parabolic PIDEs and Applications in Insurance Mathematics
- Deep learning algorithms for FBSDE s with jumps: applications to option pricing and a MFG model for smart grids (2024) — link
1 of 67 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2596 linked papers.