Multivalued Backward Stochastic Differential Equations with Jumps and Moving Boundary

Basic Information

• Paper ID: 2511.21679
• Title: Multivalued backward stochastic differential equations with jumps and moving boundary
• Authors: Badr ELMANSOURI, Anas OUKNINE, Youssef OUKNINE
• Classification: math.PR (Probability Theory)
• Publication Date: November 27, 2025
• Paper Link: https://arxiv.org/abs/2511.21679

Abstract

This paper establishes the existence and uniqueness of solutions for one-dimensional multivalued backward stochastic differential equations (MBSDE) with jumps. The equation involves a family of maximal monotone operators $k_t(\cdot)$ indexed by time, which are associated with increasing functions $k(t,\cdot)$ taking values in $\mathbb{R}_-$, defined on intervals with time-dependent boundaries. Under the conditions that the driver satisfies Lipschitz continuity in $(y,z)$, the jump parameter $\psi$ satisfies monotonicity conditions, the terminal condition and driver are square-integrable, and $k(\cdot,y)$ satisfies local time integrability conditions, existence results are obtained via the penalty function method. The paper also extends to the case where the operator $k_t(\cdot)$ acts on unbounded intervals.

Research Background and Motivation

1. Research Problem

This paper investigates multivalued backward stochastic differential equations (MBSDE) with jumps, involving:

• Time-dependent moving boundaries: The state process $Y_t$ takes values in $[a_t, +\infty)$, where the lower boundary $a_t$ varies continuously with time
• Maximal monotone operators: Multivalued operators $k_t(\cdot)$ defined by increasing functions $k(t,\cdot)$
• Jump processes: Driven by independent Poisson random measures

2. Problem Importance

Backward stochastic differential equations (BSDE) have important applications in multiple fields:

• Mathematical Finance: Derivative pricing, risk measures, utility maximization
• Partial Differential Equations: Providing probabilistic representations of quasilinear PDEs
• Stochastic Control: Adjoint equations in Pontryagin's maximum principle
• Differential Game Theory

BSDEs with jumps can characterize jump-diffusion models, extending the classical Black-Scholes framework.

3. Limitations of Existing Methods

• Limitations of continuous case: N'zi and Ouknine 23,24 studied MBSDE driven by a single maximal monotone operator in the Brownian setting, but did not address the jump case
• Failure of comparison theorems: For general BSDEs with jumps, comparison principles may fail (counterexample by Barles et al. 1)
• Restriction to fixed boundaries: Existing literature primarily focuses on operators with fixed domains, while time-dependent moving boundaries are more general

4. Research Motivation

This paper aims to extend MBSDE theory to:

• Discontinuous cases with Poisson jumps
• Time-indexed families of multivalued operators with moving boundaries
• Recovery of comparison principles through monotonicity conditions

Core Contributions

1. Existence and Uniqueness Theorem: Under the jump environment, for MBSDE with time-dependent moving boundaries, existence and uniqueness of solutions is established (Theorem 4.1)
2. Penalty Function Method: Through Lipschitz approximations $k^n(t,\cdot)$, approximating BSDE sequences are constructed and convergence is proven
3. Comparison Principle: Under monotonicity conditions on the driver with respect to the jump parameter $\psi$, a comparison theorem is established (Theorem 3.2)
4. Extension to General Case: Through localization and patching techniques, results are extended from $k(t,\cdot)$ taking negative values to the general case of real-valued operators (Theorem 5.1)
5. Local Integrability Conditions: Two key local time integrability assumptions (B.1 and B.2) are introduced, ensuring the validity of the construction method

Methodology Details

Problem Formulation

We seek a quadruple $(Y, Z, \psi, K) = (Y_t, Z_t, \psi_t, K_t)_{t\leq T}$ satisfying:

BSDE Equation (Equation 3): $Y_t = \xi + \int_t^T f(s, Y_s, Z_s, \psi_s) ds - \int_t^T Z_s dW_s - \int_t^T \int_U \psi_s(e) \tilde{N}(ds, de) + (K_T - K_t)$

Constraints:

• (4) $Y$ is an RCLL process taking values in $\mathbb{R} \cap [a_t, +\infty)$
• (5) $K$ is continuous and increasing with $K_0 = 0$
• (6) Skorokhod condition: For any optional process $(α_t, β_t) \in \text{Gr}(k_t)$, the measure $(Y_t - α_t)(dK_t + β_t dt)$ is almost surely non-positive on $[0,T]$

Core Assumptions

Assumption (A):

• (A.1) Terminal condition: $\xi \in L^2(\mathcal{F}_T)$, taking values in $[a_T, +\infty)$
• (A.2) Driver $f$:
  • (i) Progressively measurable
  • (ii) $E[\int_0^T |f(t,0,0,0)|^2 dt] < \infty$
  • (iii) Lipschitz continuous in $(y,z)$ with constant $\tilde{C}$
  • (iv) Monotonicity in $\psi$: There exists a predictable process $\kappa^{y,z,\psi,\phi}_t(e)$ such that $f(t,y,z,\psi) - f(t,y,z,\phi) \leq \int_U (\psi(e) - \phi(e))\kappa^{y,z,\psi,\phi}_t(e) \pi(de)$ with $\kappa \geq -1$, $|\kappa| \leq \vartheta \in L^2_\pi$

Assumption (B) (Local Integrability):

• (B.1) For any $y \in ]\sup_{t\in I} a_t, +\infty[$, $\int_I |k(s,y)| ds < \infty$
• (B.2) There exists $z \in ]\sup_{t\in I} a_t, +\infty[$ such that $\int_I |k(s,z)|^2 ds < \infty$

Assumption (C) (Extension to Real-valued Case): There exists a function $\ell: [0,T] \times \mathbb{R} \to \mathbb{R}_+$, increasing in each variable, right-continuous, $\ell(T,x)=0$, $|\ell(t,x)| \leq C(1+|x|)$, and $(k(t,x))^+ \leq \ell(t,x)$

Method Architecture

Phase One: Uniqueness (Section 3)

Lemma 3.1 (Key Lemma): If $(Y^1, Z^1, \psi^1, K^1)$ and $(Y^2, Z^2, \psi^2, K^2)$ are two solutions to MBSDE$(\xi, f, k)$, then the measure $(Y^1_t - Y^2_t)(dK^1_t - dK^2_t)$ is almost surely non-positive on $[0,T]$.

Proof Strategy:

1. Construct optional process pairs $(α_t, β_t)$:
   • When $Y^1_t \neq Y^2_t$, take $α_t = \frac{Y^1_t + Y^2_t}{2}$, $β_t = k(t, α_t)$
   • When $Y^1_t = Y^2_t$, take $(\hat{α}_t, \hat{β}_t) = (a_t + \varepsilon, k(t, \hat{α}_t))$
2. Apply Skorokhod condition (6) to obtain two non-positive measures
3. Through measure decomposition prove $(Y^1_t - Y^2_t)(dK^1_t - dK^2_t) \leq 0$

Theorem 3.1 (Uniqueness): Applying Itô's formula to $e^{\zeta t}|\hat{Y}_t|^2$ (where $\hat{Y} = Y^1 - Y^2$), combined with Lemma 3.1 and Lipschitz conditions, choosing $\zeta = 2C(1+2C)$, we obtain: $E[e^{\zeta t}|\hat{Y}_t|^2] + E\left[\int_t^T e^{\zeta s}(|\hat{Z}_s|^2 + \|\hat{\psi}_s\|^2_\pi) ds\right] = 0$

Therefore $Y^1 = Y^2$, and consequently $(Z^1, \psi^1) = (Z^2, \psi^2)$, $K^1 = K^2$.

Phase Two: Existence (Section 4)

Step 1: Construction of Approximating Sequence

By Lemma 4.1, there exists a sequence of functions $\{k^n(t,\cdot)\}_{n\geq 1}$:

• $k^n(t,\cdot)$ is increasing, Lipschitz continuous (constant $n$), taking negative values
• $\{k^n(t,\cdot)\}$ is decreasing
• For $x \in D_t$: $\lim_{n\to\infty} k^n(t,x) = k(t,x)$
• For $x \notin D_t$: $\lim_{n\to\infty} k^n(t,x) = -\infty$

Consider the approximating BSDE: $Y^n_t = \xi + \int_t^T \{f(s, Y^n_s, Z^n_s, \psi^n_s) - k^n(s, Y^n_s)\} ds - \int_t^T Z^n_s dW_s - \int_t^T \int_U \psi^n_s(e) \tilde{N}(ds, de)$

Define $K^n_t := -\int_0^t k^n(s, Y^n_s) ds$.

Step 2: Uniform Estimates

Applying Itô's formula to $e^{\zeta t}(Y^n_t - z)^2$, utilizing:

• Lipschitz property of $f$
• Monotonicity of $k^n(t,\cdot)$
• Choosing $z \in ]\sup_{t\in[0,T]} a_t, +\infty[$ satisfying assumption (B.2)

We obtain uniform estimates (Inequality 17): $\sup_{n\geq 1} \left\{E\left[\sup_{0\leq t\leq T} |Y^n_t|^2\right] + E\left[\int_0^T (|Z^n_s|^2 + \|\psi^n_s\|^2_\pi) ds\right]\right\} \leq C_{C,T}$

Step 3: Construction of State Variable $Y$

By the comparison theorem, $\{Y^n\}$ is increasing (since $k^{n+1} \leq k^n$), so there exists a right lower-semicontinuous process $Y$ such that: $Y_t = \lim_{n\to\infty} Y^n_t, \quad \text{P-a.s., } \forall t \in [0,T]$

By Fatou's lemma and dominated convergence theorem: $\lim_{n\to\infty} E\left[\int_0^T |Y^n_s - Y_s|^2 ds\right] = 0$

Step 4: Construction of Control Variables $(Z, \psi, K)$

For $n \geq m$, applying Itô's formula to $|Y^n_t - Y^m_t|^2$, we obtain: $E\left[\int_0^T (|Z^n_s - Z^m_s|^2 + \|\psi^n_s - \psi^m_s\|^2_\pi) ds\right] \leq C \cdot E\left[\int_0^T |Y^m_s - Y^n_s|^2 ds\right] + 2E\left[\int_t^T (Y^n_s - Y^m_s)(dK^n_s - dK^m_s)\right]$

Key Estimate (Inequalities 21-23): Using monotonicity and local integrability assumption (B): $(Y^n_s - Y^m_s)(dK^n_s - dK^m_s) \leq (Y^m_s - Y^n_s)k^n(s, Y^1_s) ds$

By carefully choosing $\varepsilon^z_s$ and estimating over different regions, we prove: $\lim_{n,m\to\infty} E\left[\int_t^T (Y^n_s - Y^m_s)(dK^n_s - dK^m_s)\right] = 0$

Therefore $\{Z^n, \psi^n\}$ is a Cauchy sequence, converging to $(Z, \psi)$.

Define: $K_T - K_t = Y_t - \xi - \int_t^T f(s, Y_s, Z_s, \psi_s) ds + \int_t^T Z_s dW_s + \int_t^T \int_U \psi_s(e) \tilde{N}(ds, de)$

Step 5: Verification of Properties

• Continuity of $K$: By $Y^n$ increasing convergence to $Y$ and Lemma 2.2 31, $Y$ and $K$ have RCLL paths. Analyzing jumps: $K_t - K_{t-} = -(Y_t - Y_{t-}) + \int_U \psi_t(e) N(\{t\}, de)$
  By $Y^{φ(n)}_t - Y^{φ(n)}_{t-} = \int_U \psi^{φ(n)}_t(e) N(\{t\}, de)$ and $Y_t \geq Y^{φ(n)}_{t-} + \int_U \psi^{φ(n)}_t(e) N(\{t\}, de)$, through UCP convergence we get $K_t - K_{t-} \leq 0$. Combined with $K$ increasing, $K$ is continuous.
• $Y_t \in [a_t, +\infty)$: By contradiction, if $Y_t < a_t$, then there exist $\varepsilon > 0$ and interval $[t,s]$ such that $Y_u < a_t - \varepsilon < a_u - \frac{\varepsilon}{2}$. By $k^{φ(n)}(u, Y^{φ(n)}_u) \leq k^{φ(n)}(u, a_u - \frac{\varepsilon}{2})$ and monotone convergence theorem: $K_s - K_t \geq \lim_{n\to\infty} \int_t^s (-k^{φ(n)}(u, a_u - \tfrac{\varepsilon}{2})) du = +\infty$ Contradiction.

Step 6: Skorokhod Condition

Through monotone convergence theorem and properties of approximating sequences, we prove: $1_{\{Y_t \in D_t\}}(dK_t + k(t, Y_t)dt) \geq 0$$1_{\{Y_t \in D°_t\}}(dK_t + k^-(t, Y_t)dt) \leq 0$

For any $(α_t, β_t) \in \text{Gr}(k_t)$, decompose the measure $(Y_t - α_t)(dK_t + β_t dt)$ into three parts ($Y_t \in D°_t$, $Y_t \notin D_t$, $Y_t = a_t \in D_t$), verifying non-positivity separately.

Phase Three: Extension to Real-valued Case (Section 5)

For the general case where $k(t,\cdot)$ takes real values, under assumptions (A), (B), and (C):

1. Truncation: Define $k^n(t,\cdot) := k(t,\cdot) \wedge n$, with graph in $\mathbb{R} \times ]-\infty, n]$
2. Transformation: Let $\hat{k}^n(t,x) := k^n(t,x) - n$, with graph in $\mathbb{R} \times \mathbb{R}_-$
3. Local Solutions: For each $n$, by Theorem 4.1 obtain solution $(\hat{Y}^n, \hat{Z}^n, \hat{\psi}^n, \hat{K}^n)$ to MBSDE$(\xi, f-n, \hat{k}^n)$
4. Recovery: $(Y^n, Z^n, \psi^n, K^n) := (\hat{Y}^n, \hat{Z}^n, \hat{\psi}^n, \hat{K}^n - nt)$ is the solution to MBSDE$(\xi, f, k^n)$
5. Stopping Time Sequence: Define $\tau_n = \inf\{t \in [0,T]: \ell(t, Y^n_t) \leq n\}$
   By assumption (C), $\tau_n \downarrow 0$ a.s.
6. Patching: On $[\tau_{n-1}, T]$, $Y^n = Y^{n-1}$ (by uniqueness), therefore: $Y_t = \sum_{n=1}^{+\infty} 1_{[\tau_n, \tau_{n-1}]}(t) Y^n_t$
   Similarly define $Z$, $\psi$, $K$.

Technical Innovations

1. Moving Boundary Treatment: Through continuous function $t \mapsto a_t$ characterizing time-dependent domains, avoiding restrictions of fixed boundaries
2. Local Integrability Conditions: Assumptions (B.1) and (B.2) are key innovations, allowing $k(t,y)$ to be locally time-integrable rather than globally integrable, greatly enhancing theoretical applicability
3. Refined Convergence Analysis: In Step 4, through region-wise estimation ($\{Y^1_s > z\}$ and $\{Y^1_s \leq z\}$) and clever construction of $\varepsilon^z_s$, overcoming difficulties from potential unboundedness of $k^n$ at certain points
4. Utilization of Monotonicity Conditions: Assumption (A.2)-(iv) on monotonicity not only ensures comparison principles but also plays crucial roles in convergence proofs
5. Patching Technique: When extending to real-valued case, using assumption (C) to construct decreasing stopping time sequences, obtaining global solutions through patching of local solutions

Experimental Setup

This is a pure theoretical mathematics paper with no numerical experiments or datasets. Main results are obtained through rigorous mathematical proofs.

Theoretical Verification Methods

• Constructive proofs: Through approximating sequences and limit processes
• Uniform estimates: Using Itô's formula, BDG inequality, Gronwall's lemma
• Convergence analysis: Fatou's lemma, dominated convergence theorem, monotone convergence theorem

Key Technical Tools

• Itô's Formula: For computing evolution of processes like $e^{\zeta t}|Y_t|^2$
• BDG Inequality: Controlling supremum of stochastic integrals
• Monotone Convergence Theorem: Handling limits of $k^n$ sequences
• Skorokhod Lemma: Characterizing reflected boundary conditions

Experimental Results

Main Theorems

Theorem 3.1 (Uniqueness): Under assumption (A), MBSDE$(\xi, f, k)$ has at most one solution.

Theorem 3.2 (Comparison Principle): If $(Y^1, Z^1, \psi^1, K^1)$ and $(Y^2, Z^2, \psi^2, K^2)$ are solutions to MBSDE$(\xi^1, f^1, k^1)$ and MBSDE$(\xi^2, f^2, k^2)$ respectively, and satisfy:

• $\xi^1 \leq \xi^2$ a.s.
• $f^1(t,y,z,\psi) \leq f^2(t,y,z,\psi)$ a.s.
• $a^1_t \leq a^2_t$, $k^1(t,x) \geq k^2(t,x)$ on $]a^2_t, +\infty[$

Then $Y^1_t \leq Y^2_t$ a.s. for all $t \in [0,T]$.

Theorem 4.1 (Existence - Negative-valued Case): Under assumptions (A) and (B), when $k(t,\cdot)$ takes negative values, there exists a unique solution $(Y, Z, \psi, K)$ satisfying: $E\left[\sup_{0\leq t\leq T} |Y_t|^2 + \int_0^T (|Z_s|^2 + \|\psi_s\|^2_\pi) ds + |K_T|^2\right] < \infty$

Theorem 5.1 (Existence - General Case): Under assumptions (A), (B), and (C), for general $k(t,\cdot)$ taking real values, there exists a unique solution.

Key Inequalities

Uniform Estimates (Inequality 17): $\sup_{n\geq 1} E\left[\sup_{0\leq t\leq T} |Y^n_t|^2 + \int_0^T |Y^n_s|^2 ds + \int_0^T (|Z^n_s|^2 + \|\psi^n_s\|^2_\pi) ds\right] \leq C_{C,T}$

Convergence (Inequalities 26-27): $\lim_{n\to\infty} E\left[\int_0^T (|Z^n_s - Z_s|^2 + \|\psi^n_s - \psi_s\|^2_\pi) ds\right] = 0$$\lim_{n\to\infty} E\left[\sup_{0\leq t\leq T} |Y^n_t - Y_t|^2\right] = 0$

Important Corollaries

Corollary 3.1: Solutions under different operators satisfy partial comparison properties.

Remark 3.1: When $k(t,\cdot)$ is Lipschitz continuous on $\mathbb{R}$, MBSDE degenerates to standard BSDE: $K_t = -\int_0^t k(s, Y_s) ds$

Related Work

Development History of Backward Stochastic Differential Equations

1. Linear BSDE: Bismut 3 (1973) introduced linear form in stochastic control
2. Nonlinear BSDE: Pardoux-Peng 29 (1990) established general theory, proving existence and uniqueness for Lipschitz drivers
3. BSDE with Jumps: Tang-Li 37, Rong 33, Situ 36, Royer 34 and others studied Poisson jump cases
4. Reflected BSDE: El Karoui et al. 6 studied continuous case, Essaky 8, Hamadène-Ouknine 14,15 studied jump cases

Research on Multivalued BSDE

1. Continuous Case:
   • N'zi-Ouknine 23 (1997): Single maximal monotone operator, square-integrable data, Lipschitz driver
   • N'zi-Ouknine 24 (1997): Continuous driver, linear growth
   • N'zi 21,22 (1997): Locally Lipschitz driver, bounded terminal condition
2. Discontinuous Case:
   • Lépingle-Marois 18 (2006): Multivalued SDE in Brownian environment
   • Marois 19 (1990): Discontinuous multivalued SDE with moving boundaries
3. Contribution of This Paper: First systematic study of MBSDE with Poisson jumps and time-dependent moving boundaries

Application Fields

• Mathematical Finance: El Karoui et al. 7, Becherer 2, Morlais 20
• PDE Theory: Pardoux 26,27,28,30, Barles et al. 1
• Stochastic Control: Hamadène-Lepeltier 12,13
• Homogenization Theory: Essaky-Ouknine 9

Advantages of This Paper

1. Theoretical Extension: Extends MBSDE theory from continuous to jump cases
2. Generality: Handles time-dependent moving boundaries, more general than fixed boundaries
3. Technical Innovation: Introduces local integrability conditions, relaxing global integrability requirements
4. Completeness: Complete theoretical framework from negative-valued to real-valued cases

Conclusions and Discussion

Main Conclusions

1. Existence and Uniqueness: Under assumptions (A) and (B), when the operator graph lies in $\mathbb{R} \times \mathbb{R}_-$, MBSDE has a unique solution (Theorem 4.1)
2. Extension to General Case: Under additional assumption (C), extends to operator graphs in $\mathbb{R} \times \mathbb{R}$ (Theorem 5.1)
3. Comparison Principle: Under monotonicity conditions, establishes comparison theorem (Theorem 3.2), overcoming difficulties from potential failure of comparison principles in jump cases
4. Methodology: Penalty function method combined with Yosida approximation and backward Skorokhod problem provides systematic framework for handling multivalued operators

Limitations

1. Dimension Restriction: Only handles one-dimensional case; higher-dimensional maximal monotone operator theory is more complex
2. Integrability Assumptions:
   • Assumption (B) requires local time integrability, potentially too strong for some applications
   • Assumption (C) requires function $\ell$ to satisfy specific growth conditions
3. Monotonicity Conditions: Assumption (A.2)-(iv) requires specific monotonicity of driver in $\psi$, limiting applicability
4. Boundary Continuity: Requires $t \mapsto a_t$ continuous, cannot handle jump boundaries
5. Theoretical Nature: No numerical algorithms or practical application cases provided

Future Directions

1. High-dimensional Generalization: Extend to multidimensional MBSDE, requiring handling of maximal monotone operators on $\mathbb{R}^d$
2. Weakening Assumptions:
   • Relax Lipschitz conditions to locally Lipschitz or monotone continuous
   • Study more general integrability conditions
3. Numerical Methods: Develop effective numerical algorithms for solving MBSDE
4. Application Research:
   • Optimal stopping problems in finance
   • Stochastic control with constraints
   • Barrier option pricing
5. Double Reflection Case: Study MBSDE with both upper and lower moving boundaries
6. Related PDEs: Establish connections with integro-partial differential equations

In-Depth Evaluation

Strengths

1. Theoretical Rigor:
   • Complete and detailed proofs with sufficient justification for each step
   • Sophisticated technical handling, particularly convergence analysis in Step 4
   • Effective use of counterexamples and proof by contradiction strengthens argumentation
2. Method Innovation:
   • Clever application of penalty function method through Lipschitz approximation for multivalued operators
   • Local integrability assumption (B) is key innovation, balancing theoretical generality and technical feasibility
   • Patching technique (Section 5) demonstrates systematic approach from local to global
3. Theoretical Contribution:
   • First systematic study of MBSDE with jumps and moving boundaries
   • Establishment of comparison theorem fills theoretical gap
   • Extension from negative-valued to real-valued demonstrates theoretical completeness
4. Writing Clarity:
   • Logical structure: from uniqueness to existence, from special to general
   • Clear and explicit assumption conditions facilitating reader understanding
   • Remark sections provide valuable supplementary explanations
5. Mathematical Tool Usage:
   • Skillful application of Itô's formula, BDG inequality, various convergence theorems
   • Appropriate use of measure theory and functional analysis tools

Weaknesses

1. Practical Limitations:
   • Lacks numerical experiments verifying theoretical results
   • No concrete application case studies provided
   • Verification methods for assumption conditions unclear
2. Technical Complexity:
   • Highly technical proofs unfriendly to non-specialists
   • Some steps (e.g., inequalities 21-23 in Step 4) lack intuitive explanation
   • Constant dependencies ($C_{C,T}$) not explicitly characterized
3. Theoretical Completeness:
   • No discussion of solution regularity (e.g., Hölder continuity)
   • Lacks stability analysis (solution behavior under parameter perturbations)
   • Limited exploration of connections with PDEs
4. Necessity of Assumptions:
   • Insufficient justification for necessity of assumption (C)
   • Unclear whether local integrability assumption (B) can be further weakened
   • Optimality of monotonicity condition (A.2)-(iv) not discussed
5. Literature Review:
   • Insufficient detailed comparison with related work
   • No discussion of relationship between this paper's methods and other approaches (e.g., viscosity solution methods)

Impact

1. Theoretical Value:
   • Provides important extension to MBSDE theory
   • Methodology inspiring for related problems (e.g., double-reflected BSDE)
   • Comparison theorem establishment has independent value
2. Application Potential:
   • Mathematical Finance: Constrained optimal portfolio problems
   • Stochastic Control: State-constrained optimization problems
   • PDE Theory: Probabilistic representation of obstacle problems
3. Reproducibility:
   • Detailed theoretical proofs verifiable by experts
   • Lacks code implementation, numerical aspects difficult to reproduce
   • Constructive proofs in principle convertible to algorithms
4. Follow-up Research:
   • Lays foundation for high-dimensional generalizations
   • Stimulates numerical method research
   • Promotes exploration in application fields

Applicable Scenarios

1. Theoretical Research:
   • Further development of stochastic analysis theory
   • Intersection of maximal monotone operator theory and stochastic processes
   • Probabilistic methods for nonlinear PDEs
2. Financial Engineering:
   • Option pricing with transaction costs
   • Portfolio optimization with constraints
   • Credit risk models (jump defaults)
3. Stochastic Control:
   • State-constrained control problems
   • Optimal control of reflected diffusion processes
   • Obstacle problems in differential games
4. Engineering Applications:
   • Filtering problems with boundary constraints
   • Queueing network optimization
   • Inventory management (jump demand)

References

This paper cites 37 important references, key ones including:

1. 3 Bismut (1973): Pioneering work on BSDE
2. 29 Pardoux-Peng (1990): Foundational theory of nonlinear BSDE
3. 4 Brézis (1973): Theory of maximal monotone operators
4. 23,24 N'zi-Ouknine (1997): MBSDE in continuous case
5. 19 Marois (1990): Multivalued SDE with moving boundaries
6. 34 Royer (2006): BSDE with jumps and comparison theorems
7. 5 Delong (2014): Monograph on BSDE with jumps
8. 32 Protter (2004): Stochastic integration and differential equations

Overall Assessment: This is a high-quality theoretical mathematics paper achieving important progress in the theory of multivalued backward stochastic differential equations with jumps. Proofs are rigorous, methods innovative, laying solid foundation for subsequent research in this field. Main shortcomings are lack of numerical verification and practical application cases, and insufficient justification for necessity of certain technical assumptions. For researchers in stochastic analysis and related application fields, this is a paper worthy of in-depth study.