Pith. sign in

REVIEW 3 major objections 5 minor 24 references

Universal Approximation Theorem for Deep Q-Learning via FBSDE System

T0 review · 3 major / 5 minor · reviewed 2026-08-15 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read For any target accuracy, a residual Q-network whose layers emulate Bellman updates can match the optimal Q-function to that accuracy, with depth set by the horizon and the discount.

desk verdict A new and appealing architecture-to-Bellman-iteration correspondence, but the main theorem leans on a circular regularity proof and an operator existence assumption that is close to the result itself. read the letter →

arxiv 2505.06023 v1 pith:YHRMEVGQ submitted 2025-05-09 cs.LG cs.AImath.OC

Qian Qi This is my paper · ORCID
classification cs.LGcs.AImath.OC
keywords universalapproximationtheoremdeepQ-networksBellmanoperatorvalueiterationneuraloperatorsbackwardstochasticdifferentialequationsLipschitzregularityresidualnetworks
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

Standard universal approximation theorems justify deep Q-networks by treating the optimal Q-function as an arbitrary continuous function. This paper argues that $Q^*$ is not arbitrary: it is the unique fixed point of a Bellman operator with a contraction structure, and a network built from operator layers that approximate the Bellman residual can exploit that structure. The central claim is a universal approximation theorem for such networks: for any $\varepsilon > 0$, enough layers and a fine enough discretization grid give a network output within $\varepsilon$ of $Q^*$ in the sup norm. Depth plays a concrete role, corresponding to Bellman iterations, and the discount factor controls how operator approximation errors accumulate. A sympathetic reader should care because this gives a problem-aware, structural explanation of DQN approximation power rather than a generic function-approximation guarantee.

What carries the argument

The load-bearing object is the Bellman operator $B$ on the space $C(K_Q)$ of continuous Q-functions, defined for a $\delta$-discretized continuous-time MDP in which the chosen action is held constant for a short duration $\delta$. The operator $B$ is a contraction with factor $e^{-\lambda\delta}$, and its residual form $J(Q) = BQ - Q$ is what each network block approximates. A single application of $B$ can be represented as the solution of a short-horizon backward stochastic differential equation coupled to the forward state process, which is the FBSDE system of the title; this representation is what transfers regularity from the input $Q$ to the output $BQ$. The network is a residual stack of neural-operator blocks of the form $Q \mapsto Q + D_M(N_\theta(E_M(Q)))$, where $E_M$ samples $Q$ on a grid, $N_\theta$ is a finite-dimensional network, and $D_M$ reconstructs a function. The proof's key step, Lemma 3.1(d), establishes that all Bellman iterates are uniformly Lipschitz and uniformly bounded, placing them in a compact subset of $C(K_Q)$ so that the neural-operator approximation lemma applies uniformly across layers.

What would settle it

Take a simple MDP satisfying Assumption 2.1 (for instance, the one-dimensional deterministic action problem in Appendix E) and compute, as $\varepsilon \to 0$, the minimal output Lipschitz constant $L^*_F$ among all neural-operator blocks that approximate $J$ on the compact set of iterates to sup-norm accuracy $\varepsilon$. If $L^*_F$ cannot be bounded independently of $\varepsilon$, Assumption 4.1(b) fails and the theorem's conclusion is not guaranteed.

Watch

Extended reading notes

Core claim

The paper's central claim is Theorem 4.4: for any $\varepsilon > 0$, there exist a number $L$ of operator layers, a discretization scheme $(E_M, D_M)$, and network parameters $\theta$ such that the network's final output satisfies $\|\hat{Q}_{NN}^{(L)} - Q^*\|_\infty < \varepsilon$, where $L$ is tied to $T/\delta$ and the contraction factor $e^{-\lambda\delta}$. The proof shows that, under the standard Lipschitz assumptions on the MDP data, every Bellman iterate $Q^{(k)}$ and the limit $Q^*$ is uniformly Lipschitz and uniformly bounded on the compact domain $K_Q$. These functions therefore lie in a compact set of continuous functions, which allows a neural-operator universal approximation theorem to approximate the residual operator $J(Q) = BQ - Q$ uniformly on that set. Each residual network block then acts as one approximate Bellman step, and the contraction of $B$ keeps the accumulated approximation error from exploding. The paper presents this as an iterative-refinement proof technique in which network depth literally corresponds to value-iteration count.

Load-bearing premise

The theorem rests on Assumption 4.1: that there exists a family of neural-operator blocks which can approximate the Bellman residual $J(Q)$ to any desired accuracy while keeping the Lipschitz constant of their output functions uniformly bounded; the paper sketches why bounded basis functions might achieve this but does not prove that such a family exists.

Editorial extensions

If this is right

  • Depth is assigned a quantitative meaning: a network with $L \approx (1/(\lambda\delta)) \log(M_Q/\varepsilon)$ layers performs roughly $L$ approximate Bellman steps, and adding layers is equivalent to running more value iterations.
  • The contraction factor $e^{-\lambda\delta}$ guarantees that per-layer operator approximation errors accumulate at most as $1/(1-e^{-\lambda\delta})$; the total error is the value-iteration truncation error plus this accumulated operator error.
  • Under the theorem's assumptions, the optimal Q-function and every Bellman iterate lie in a common compact, uniformly Lipschitz set, which is exactly the setting where neural-operator UATs apply; this justifies the use of operator-based residual architectures for Q-learning.
  • Quantitative rates require stronger smoothness of the Bellman iterates: with $s$-order smoothness, the discretization error scales like $M^{-s/d_Q}$, so improving regularity estimates on $Q^*$ directly improves the grid size and network size needed for a given $\varepsilon$.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • A natural next step, not taken in the paper, is to instantiate Assumption 4.1 for concrete architectures such as bounded-basis decoders, which would turn the conditional theorem into an unconditional one.
  • The FBSDE perspective suggests that structure-aware universal approximation theorems could be built for other control problems whose value functions have known regularity classes; this paper's construction is one instance of that pattern.
  • A direct empirical test of the key assumption would be to train the proposed operator-residual network on a small continuous-time MDP at several decreasing target errors and measure the Lipschitz constant of the learned operator blocks; boundedness of that constant would support Assumption 4.1, while divergence would indicate the assumption needs modification.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. The paper proposes a DQN architecture composed of neural-operator residual blocks, where each block approximates the residual Bellman operator J(Q)=BQ-Q, and claims a universal approximation theorem for the resulting network. The proof strategy is to show that the Bellman iterates Q(k) are uniformly bounded and uniformly Lipschitz, to place them in a compact subset of C(KQ), to apply a neural-operator universal approximation theorem for the operator J on that compact set, and to control the accumulation of per-layer errors using the contraction of the Bellman operator. The main theorem, Theorem 4.4, states that for any epsilon>0 there are L layers, a discretization scheme, and parameters such that the final network output approximates Q* to within epsilon in the sup norm, with L related to T/delta and the contraction factor e^{-lambda delta}.

Significance. If the claims were correct, the paper would offer a structural, problem-aware UAT for DQNs and a transparent link between network depth and Bellman iterations, with a clean error-propagation analysis. The operator view of residual layers and the explicit accumulation bound e_L <= epsilon_1/(1-e^{-lambda delta}) in Appendix C are useful and clearly presented. However, the result does not currently hold as proved: the uniform Lipschitz lemma is circular, and the central Assumption 4.1 is both unproved and used in a circle through the definition of K_target. These are not presentation issues but load-bearing gaps in the proof of Theorem 4.4.

major comments (3)
  1. [Appendix A, proof of Lemma 3.1(d)] The argument proving uniform Lipschitz continuity of all iterates is circular. After deriving the recurrence L(k+1) <= K_A + K_B L(k), the proof considers the case K_B >= 1 and asserts that because Q(k) -> Q* uniformly and Q* is Lipschitz, the set {Q(k)} union {Q*} is compact by Lemma 3.2, which in turn is proved from Lemma 3.1(d). The additional appeal to 'the number of relevant recursions is bounded by N_max = ceil(T/delta)' is not justified: the value iteration sequence is indexed by k >= 0 and the recurrence has no built-in truncation at N_max; if K_B >= 1, the displayed bound involving N_max applies only to the first N_max iterates and says nothing about later iterates unless one already knows uniform Lipschitz boundedness. Thus the uniform regularity of all iterates, which is the load-bearing premise for Lemma 3.2 and for the compactness needed by the neural-operator UAT, is not established.
  2. [Assumption 4.1 and Appendix C, Step 3] The definition of the compact set K_target is circular. In Step 3 of Appendix C, L*_unif-Lip is defined through the quantities A' = 2K_A + L*_F and B' = 2K_B + 1, so L*_unif-Lip depends on L*_F. Lemma 4.3 then invokes Assumption 4.1(b) to obtain an L*_F for an architecture approximating J on exactly this set K_target = {Q : ||Q||_infty <= M*_unif-bound, Lip(Q) <= L*_unif-Lip}. Assumption 4.1(b) explicitly allows L*_F to depend on the Lipschitz data L_K of the task, and since J(Q) = BQ - Q has Lipschitz constant at least K_A + (1 - e^{-lambda delta}) L for Q in K_target, any realization has L*_F >= K_A + (1 - e^{-lambda delta}) L*_unif-Lip; substituting this into the recurrence gives no positive solution for L*_unif-Lip once L is moderately large, so the induction that all approximate iterates lie in K_target cannot be closed. Remark 4.2 only asserts that L*_F is a 'given constant' for the eventual K_target; that is precisely the fixed-point step that is missing.
  3. [Assumption 4.1(b)] The main theorem is conditional on an unproved existence statement. Assumption 4.1(b) postulates that, for every compact set of uniformly Lipschitz functions, a neural operator can uniformly approximate a given continuous operator while all output functions share a uniform Lipschitz constant L*_F. Standard neural-operator UATs cited in the paper provide sup-norm approximation on a prescribed compact set, but they do not control the Lipschitz constant of the output functions. Remark 4.2's basis-function estimate L*_F <= M B_y L_phi is only a sketch: it assumes bounded network outputs B_y and does not show that those bounds are compatible with the required approximation accuracy epsilon_op for J on K_target, nor does it constitute a proof of existence for the architecture used in Lemma 4.3. Since Theorem 4.4 is stated only under this assumption, the paper's central approximation claim is conditional on a property whose realization is not demonstrated.
minor comments (5)
  1. [Appendix C, Step 1] The sentence 'we set L = 1 if epsilon >= 2M_Q' is inconsistent with the preceding formula L = ceil(ln(2M_Q/epsilon)/(lambda delta)) + 1, since the formula already handles all epsilon > 0.
  2. [Appendix A, proof of Lemma 3.1(a)] The continuity argument relies on a convergence statement for SDE solutions on a common time interval but does not spell out how the integrals with different lower limits t and t' are compared; this is a presentation gap in a non-central part of the proof.
  3. [Figure 1] The axis tick labels in Figure 1 are rendered as raw Unicode codepoints, making the figure unreadable; the figure should be regenerated.
  4. [Lemma 4.3 and Appendix C] Lemma 4.3 is stated before the set K_target is defined and refers to 'as defined in the proof of Theorem 4.4 in Appendix C and shown therein', which makes the lemma difficult to read independently; a self-contained definition before the lemma would improve clarity.
  5. [Section 5.1.1] The citation 'Krylov and Krylov [1987]' appears in the text but the reference list contains a single Krylov entry with the same author name repeated; this should be corrected.

Circularity Check

2 steps flagged · score 7.0 of 10

The central regularity/approximation argument is self-referential: Ktarget is defined through L*_unif-Lip, which is defined through L*_F, which Assumption 4.1 only supplies for that same Ktarget; Lemma 3.1(d) also cites Lemma 3.2, whose proof presupposes Lemma 3.1(d).

  1. self definitional [Assumption 4.1(b), Remark 4.2, Lemma 4.3, and Appendix C, Step 3 (definition of Ktarget)]
    "For the subsequent analysis of L∗unif-Lip in Appendix C, this L∗F (associated with the neural operator chosen to approximate J on Ktarget) is treated as a given constant. This avoids circularity where L∗F would depend iteratively on L∗unif-Lip."

    Appendix C first computes L∗unif-Lip from A′ = 2K_A + L∗F via L∗unif-Lip = A′((B′)L−1−1)/(B′−1), and only then defines Ktarget = {Q : ‖Q‖∞ ≤ M∗unif-bound and Q is L∗unif-Lip-Lipschitz}. Lemma 4.3 then invokes Assumption 4.1(b) to supply L∗F for exactly this Ktarget. But Assumption 4.1(b) permits L∗F to depend on LK = L∗unif-Lip and on LG(K) = K_A + (K_B+1)L∗unif-Lip, so L∗F is a function of L∗unif-Lip, not a given constant. The remark's assertion that this avoids circularity is the circular step: no fixed point is proved.

  2. other [Appendix A, proof of Lemma 3.1(d)]
    "Even if KB ≥ 1, since Q(k) → Q∗ uniformly, and Q∗ is L∗Q∗-Lipschitz, the set {Q(k)}k≥0 ∪ {Q∗} is compact in C(KQ) by Lemma 3.2 (if Q(0) is Lipschitz). This implies the L(k) must be uniformly bounded."

    Lemma 3.2 is proved from exactly the property being established: its proof says 'This follows from Lemma 3.1 (d) (uniform boundedness and uniform Lipschitz continuity of all Q(k) on the compact domain KQ).' Thus Lemma 3.1(d) uses Lemma 3.2 to conclude uniform boundedness of the Lipschitz constants, while Lemma 3.2 presupposes uniform Lipschitz continuity. The subsequent backward-induction sentence is only a sketch and does not remove the quoted self-dependency in the written proof.

full rationale

Most of the paper is a conditional theorem: given Assumption 4.1 (a neural-operator class with a UAT plus output-Lipschitz control), the proof constructs a DQN whose layers emulate Bellman updates. An explicitly assumed UAT is not by itself circular, and the motivational self-citation [Qi 2025] is not load-bearing. However, the proof of Theorem 4.4 does not merely assume an external UAT; it constructs the target class Ktarget after L∗unif-Lip has been computed from L∗F and then asks Assumption 4.1 to return that same L∗F for this Ktarget. Since Assumption 4.1(b) explicitly allows L∗F to depend on LK and LG(K), and since for J = B − I these quantities grow with L∗unif-Lip, the bound L∗unif-Lip = A′((B′)L−1−1)/(B′−1) is a self-referential equation rather than a closed form. No fixed-point argument or monotone coupling is supplied; Remark 4.2 simply asserts the problem away. This is a genuine circularity in the central induction. The compactness proof for the exact Bellman iterates contains a second, smaller circularity: Lemma 3.1(d) invokes Lemma 3.2, which is proved from Lemma 3.1(d). The paper is not circular by self-citation or by renaming a known result; rather its load-bearing regularity step is self-definitional. Score 7 reflects that the main theorem's approximation claim is not actually closed as written.

Assumptions & free parameters 0 free parameters · 5 assumptions · 0 invented entities

The central claim rests on the standard Lipschitz model assumptions (Assumption 2.1), on unproved existence of a regularizing neural operator class (Assumption 4.1, which is nearly as strong as the theorem), and on a circular argument for uniform Lipschitz continuity of the Bellman iterates. There are no fitted numerical parameters: the grid size M, layer count L, and network capacity are existence parameters set by the error tolerance. No new physical entities are introduced.

assumptions (5)
  • domain assumption Assumption 2.1: drift h, diffusion sigma, reward r, and terminal reward g are uniformly Lipschitz and bounded on compact domains.
    This is the standard controllability and regularity assumption used to prove continuity, contraction, and Lipschitz propagation of value iteration. It is restrictive for practical RL, excluding unbounded state spaces and nonsmooth rewards, but it is standard in stochastic control (Fleming and Soner [2006]).
  • ad hoc to paper Assumption 4.1: existence of a neural operator class D_M(N_theta(E_M(·))) with universal approximation of continuous operators on compact sets of uniformly Lipschitz functions and a uniform output Lipschitz constant L*_F.
    This is the load-bearing existence assumption for Theorem 4.4. No explicit construction or proof is provided; Remark 4.2 only gives a heuristic bound for bounded basis functions. It is strong enough to be nearly equivalent to the theorem being claimed.
  • standard math Standard BSDE and SDE stability estimates: uniqueness and Lipschitz dependence of solutions on initial data and parameters.
    Used in Section 3 and Appendix A to justify Lipschitz preservation of a single Bellman step. These are textbook results (Peng 1992, El Karoui et al. 1997, Yong and Zhou 1999).
  • standard math Arzela-Ascoli theorem and Banach fixed-point theorem.
    Used to assert precompactness of value iterates and convergence of value iteration in C(K_Q).
  • ad hoc to paper Finite-horizon dynamic programming implies uniform Lipschitz continuity of all Bellman iterates Q(k) and Q*.
    Appealed to in Lemma 3.1(d) after the fixed-point inequality L <= K_A + K_B L fails for K_B >= 1. The paper asserts the number of relevant recursions is bounded by N_max = ceil(T/delta) and cites Fleming and Soner, but this part of the proof is not written out and the argument as presented is circular.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Universal Approximation Theorem for Deep Q-Learning via FBSDE System." pith.science (2026). https://pith.science/paper/YHRMEVGQ

@misc{pith2026250506023,
  author       = {Pith},
  title        = {Pith review of: Universal Approximation Theorem for Deep Q-Learning via FBSDE System},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/YHRMEVGQ}},
  note         = {Machine review of arXiv:2505.06023}
}
read the original abstract

The approximation capabilities of Deep Q-Networks (DQNs) are commonly justified by general Universal Approximation Theorems (UATs) that do not leverage the intrinsic structural properties of the optimal Q-function, the solution to a Bellman equation. This paper establishes a UAT for a class of DQNs whose architecture is designed to emulate the iterative refinement process inherent in Bellman updates. A central element of our analysis is the propagation of regularity: while the transformation induced by a single Bellman operator application exhibits regularity, for which Backward Stochastic Differential Equations (BSDEs) theory provides analytical tools, the uniform regularity of the entire sequence of value iteration iterates--specifically, their uniform Lipschitz continuity on compact domains under standard Lipschitz assumptions on the problem data--is derived from finite-horizon dynamic programming principles. We demonstrate that layers of a deep residual network, conceived as neural operators acting on function spaces, can approximate the action of the Bellman operator. The resulting approximation theorem is thus intrinsically linked to the control problem's structure, offering a proof technique wherein network depth directly corresponds to iterations of value function refinement, accompanied by controlled error propagation. This perspective reveals a dynamic systems view of the network's operation on a space of value functions.

Figures

Figures reproduced from arXiv: 2505.06023 by the authors.

Figure 1
Figure 1. Illustration of Bellman iterates Q(k) (s, a) for k = 0, 1, 2. The state space is s ∈ [0, 1]. Actions aL (move left) and aR (move right) are shown. Q(0) is zero. Q(1) is identical for both actions. Q(2) shows distinct values for aL and aR, reflecting the one-step lookahead with V (1). All iterates are bounded and visually Lipschitz continuous. 33 [PITH_FULL_IMAGE:figures/full_fig_p033_1.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

24 extracted references · 16 canonical work pages

  1. [1]

    Universal approximation bounds for superpositions of a sigmoidal function

    Andrew R Barron. Universal approximation bounds for superpositions of a sigmoidal function. IEEE Transactions on Information theory, 39 0 (3): 0 930--945, 1993

    1993

  2. [2]

    Sparse grids

    Hans-Joachim Bungartz and Michael Griebel. Sparse grids. Acta numerica, 13: 0 147--269, 2004

    work page 2004

  3. [3]

    Universal approximation to nonlinear operators by neural networks with arbitrary activation functions and its application to dynamical systems

    Tianping Chen and Hong Chen. Universal approximation to nonlinear operators by neural networks with arbitrary activation functions and its application to dynamical systems. IEEE transactions on neural networks, 6 0 (4): 0 911--917, 1995

    work page 1995

  4. [4]

    Approximation by superpositions of a sigmoidal function

    George Cybenko. Approximation by superpositions of a sigmoidal function. Mathematics of control, signals and systems, 2 0 (4): 0 303--314, 1989

    1989

  5. [5]

    Backward stochastic differential equations in finance

    Nicole El Karoui, Shige Peng, and Marie Claire Quenez. Backward stochastic differential equations in finance. Mathematical finance, 7 0 (1): 0 1--71, 1997

    work page 1997

  6. [6]

    Controlled Markov processes and viscosity solutions, volume 25

    Wendell H Fleming and Halil M Soner. Controlled Markov processes and viscosity solutions, volume 25. Springer Science & Business Media, 2006

    work page 2006

  7. [7]

    Tensor spaces and numerical tensor calculus, volume 42

    Wolfgang Hackbusch. Tensor spaces and numerical tensor calculus, volume 42. Springer, 2012

    work page 2012

  8. [8]

    Deep residual learning for image recognition

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Deep residual learning for image recognition. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 770--778, 2016

    2016

Show all 24 references
  1. [9]

    Approximation capabilities of multilayer feedforward networks

    Kurt Hornik. Approximation capabilities of multilayer feedforward networks. Neural networks, 4 0 (2): 0 251--257, 1991

    1991

  2. [10]

    General topology

    John L Kelley. General topology. Courier Dover Publications, 2017

    2017

  3. [11]

    Neural operator: Learning maps between function spaces with applications to pdes

    Nikola Kovachki, Zongyi Li, Burigede Liu, Kamyar Azizzadenesheli, Kaushik Bhattacharya, Andrew Stuart, and Anima Anandkumar. Neural operator: Learning maps between function spaces with applications to pdes. Journal of Machine Learning Research, 24 0 (89): 0 1--97, 2023

    2023

  4. [12]

    Nonlinear elliptic and parabolic equations of the second order

    Nikolai Vladimirovich Krylov and NV Krylov. Nonlinear elliptic and parabolic equations of the second order. Springer, 1987

    1987

  5. [13]

    Deep learning via dynamical systems: An approximation perspective

    Qianxiao Li, Ting Lin, and Zuowei Shen. Deep learning via dynamical systems: An approximation perspective. Journal of the European Mathematical Society, 2022

    2022

  6. [14]

    Fourier neural operator for parametric partial differential equations

    Zongyi Li, Nikola Borislavov Kovachki, Kamyar Azizzadenesheli, Burigede liu, Kaushik Bhattacharya, Andrew Stuart, and Anima Anandkumar. Fourier neural operator for parametric partial differential equations. In International Conference on Learning Representations, 2021. URL htt...

    2021

  7. [15]

    Learning nonlinear operators via deeponet based on the universal approximation theorem of operators

    Lu Lu, Pengzhan Jin, Guofei Pang, Zhongqiang Zhang, and George Em Karniadakis. Learning nonlinear operators via deeponet based on the universal approximation theorem of operators. Nature machine intelligence, 3 0 (3): 0 218--229, 2021

    2021

  8. [16]

    Forward-backward stochastic differential equations and their applications

    Jin Ma and Jiongmin Yong. Forward-backward stochastic differential equations and their applications. Number 1702. Springer Science & Business Media, 1999

    1999

  9. [17]

    Human-level control through deep reinforcement learning

    Volodymyr Mnih, Koray Kavukcuoglu, David Silver, Andrei A Rusu, Joel Veness, Marc G Bellemare, Alex Graves, Martin Riedmiller, Andreas K Fidjeland, Georg Ostrovski, et al. Human-level control through deep reinforcement learning. In Nature, volume 518, pages 529--533. Nature Pu...

    2015

  10. [18]

    Stochastic differential equations: an introduction with applications

    Bernt Oksendal. Stochastic differential equations: an introduction with applications. Springer Science & Business Media, 2003

    2003

  11. [19]

    Bsdes, weak convergence and homogenization of semilinear pdes

    \'E tienne Pardoux. Bsdes, weak convergence and homogenization of semilinear pdes. In Nonlinear analysis, differential equations and control, pages 503--549. Springer, 1999

    1999

  12. [20]

    Stochastic hamilton--jacobi--bellman equations

    Shige Peng. Stochastic hamilton--jacobi--bellman equations. SIAM Journal on Control and Optimization, 30 0 (2): 0 284--304, 1992

    1992

  13. [21]

    Universal approximation theorem of deep q-networks

    Qian Qi. Universal approximation theorem of deep q-networks. In International Conference on Machine Learning. PMLR, 2025. URL https://arxiv.org/abs/2505.02288

    2025 arXiv

  14. [22]

    A mean-field optimal control formulation of deep learning

    E Weinan, Jiequn Han, and Qianxiao Li. A mean-field optimal control formulation of deep learning. Research in the Mathematical Sciences, 6 0 (1): 0 1--41, 2019

    2019

  15. [23]

    Error bounds for approximations with deep relu networks

    Dmitry Yarotsky. Error bounds for approximations with deep relu networks. Neural networks, 94: 0 103--114, 2017

    2017

  16. [24]

    Stochastic controls: Hamiltonian systems and HJB equations, volume 43 of Applications of Mathematics

    Jiongmin Yong and Xun Yu Zhou. Stochastic controls: Hamiltonian systems and HJB equations, volume 43 of Applications of Mathematics. Springer Science & Business Media, 1999

    1999

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.