betaFeedback

WorksMathematics

Residual-Driven Basis and Data Selection for Polynomial Approximation of High-Dimensional Hamilton-Jacobi-Bellman Equations

Abstract

We develop greedy strategies for selecting global polynomial basis functions and training data in the sparse approximation of high-dimensional partial differential equations, with emphasis on Hamilton-Jacobi-Bellman (HJB) equations arising in nonlinear optimal control. Inspired by the goal-oriented dual-weighted residual framework, we derive a closed-form, mesh-free indicator that predicts, to first order, the reduction of the PDE residual obtained by adding a single candidate basis function, without solving an adjoint problem. In the data-driven regime, we jointly select basis functions and gradient-augmented training samples by maximizing the predicted decrease of a supervised loss. Numerical experiments on benchmark PDEs and nonlinear optimal control problems in up to 35 dimensions show accuracy comparable to non-adaptive approaches with substantially fewer basis functions and training samples.

Introduction

The numerical solution of partial differential equations (PDEs) posed on high-dimensional state spaces is a central challenge in scientific computing, arising prominently in dynamic programming, where the value function of a dd-dimensional optimal control problem satisfies a Hamilton–Jacobi–Bellman (HJB) equation over a subset of Rd\mathbb{R}^{d}. Classical methods, such as finite differences [34], finite elements [61], and spectral methods [51], are effective in low dimensions but typically require meshes, grids, or tensor-product approximation spaces whose cardinality grows exponentially with dd, a phenomenon known as the curse of dimensionality [4]. Sparse grids [11], reduced basis methods [44], kernel-based approximation [58], and deep learning [27] mitigate this growth with ansatze whose complexity is not tied to a grid in Rd\mathbb{R}^{d}. For methods built on a dictionary of candidate features, their effectiveness relies on identifying which basis functions and training samples are most relevant to the solution at hand.

The prototypical problem motivating this work is the synthesis of optimal feedback laws for nonlinear dynamical systems. Consider a control-affine system with state y(t)∈Rd\mathbf{y}(t) \in\mathbb{R}^{d} and control signal u(t)∈Rm\mathbf{u}(t) \in\mathbb{R}^{m}, for which we seek a control minimizing an accumulated running cost over an infinite horizon:

min⁡u(⋅)∫0∞(ℓ(y(t))+β∥u(t)∥2) dt,β>0,subject toy˙(t)=f(y(t))+q(y(t))u(t),y(0)=x.(1)\min_{\mathbf{u}(\cdot)} \int_{0}^{\infty} \left(\ell(\mathbf{y}(t))+\beta\lVert\mathbf{u}(t)\rVert^{2}\right)\,\mathrm{d}t,\qquad\beta>0, \qquad\text{subject to}\quad\dot{\mathbf{y}}(t)=\mathbf{f}(\mathbf{y}(t))+\mathbf{q}(\mathbf{y}(t))\mathbf{u}(t),\qquad\mathbf{y}(0)=\mathbf{x}. \tag*{(1)}

2020 Mathematics Subject Classification. 49L12, 49L20, 49N35, 68T05, 93C20.

Key words and phrases. Optimal feedback control, Hamilton–Jacobi–Bellman equations, Adaptive basis selection, Greedy algorithms, Galerkin methods, Polynomial approximation.

The optimal cost of (1), understood as a function of the initial state x\mathbf{x}, defines the value function V(x)V(\mathbf{x}) of the control problem. Dynamic programming establishes that, wherever differentiable, the value function satisfies the stationary, first-order Hamilton–Jacobi–Bellman equation

0=min⁡u∈Rm{ℓ(x)+β∥u∥22+∇V(x)T(f(x)+q(x)u)}.(2)0 = \min_{\mathbf{u}\in\mathbb{R}^{m}} \left\{ \ell(\mathbf{x}) + \beta\lVert\mathbf{u}\rVert_2^2 + \nabla V(\mathbf{x})^{\mathsf{T}}\left(\mathbf{f}(\mathbf{x})+\mathbf{q}(\mathbf{x})\mathbf{u}\right) \right\}. \tag*{(2)}

Once VV is known, the optimal control is recovered in feedback form, u∗(x)=−12βq(x)T∇V(x)\mathbf{u}^{*}(\mathbf{x})=-\frac{1}{2\beta}\mathbf{q}(\mathbf{x})^{\mathsf{T}}\nabla V(\mathbf{x}), valid for every state x\mathbf{x}. Inserting u∗\mathbf{u}^{*} into (2) gives the fully nonlinear PDE LHJB(V)=0\mathcal{L}_{\mathrm{HJB}}(V)=0, with

LHJB(V):=−14β∇VTqqT∇V+∇VTf+ℓ,\mathcal{L}_{\mathrm{HJB}}(V):=-\frac{1}{4\beta}\nabla V^{\mathsf{T}}\mathbf{q}\mathbf{q}^{\mathsf{T}}\nabla V+\nabla V^{\mathsf{T}}\mathbf{f}+\ell,

cast over the state space of the dynamics, so that grid-based discretizations become unfeasible already for moderate values of dd. The construction of numerical schemes for (2) which remain tractable in high dimensions is the main purpose of this paper. The methodology applies to a general class of first-order PDEs L(V)=0\mathcal{L}(V)=0 (see § 3), with the HJB equation as our leading example.

Sparse greedy approximation. To approximate the value function VV, we introduce a large pool of globally defined candidate basis functions ΦΘ(x)={ϕj(x):j∈Θ}\Phi_{\Theta}(\mathbf{x})=\{\phi_j(\mathbf{x}):j\in\Theta\}, with ∣Θ∣=P≫1|\Theta|=P\gg1, such as multivariate polynomials, and consider approximations of the form

V^(x;c)=∑j∈Γcjϕj(x)=cTΦΓ(x).(3)\widehat{V}(\mathbf{x};\mathbf{c})=\sum_{j\in\Gamma}c_j\phi_j(\mathbf{x})=\mathbf{c}^{\mathsf{T}}\Phi_{\Gamma}(\mathbf{x}). \tag*{(3)}

where Γ⊂Θ\Gamma\subset\Theta is the index set of active basis functions, ΦΓ(x)=[ϕj(x)]j∈ΓT\Phi_{\Gamma}(\mathbf{x})=[\phi_j(\mathbf{x})]_{j\in\Gamma}^{\mathsf{T}} is the vector of active basis functions, and c=[cj]j∈ΓT\mathbf{c}=[c_j]_{j\in\Gamma}^{\mathsf{T}} is the coefficient vector. In practice only a relatively small subset of ΦΘ\Phi_{\Theta} contributes significantly to the approximation, while the cost of computing c\mathbf{c} grows rapidly with n:=∣Γ∣n:=|\Gamma|. Our aim is therefore to identify an accurate active set with n≪Pn\ll P. We construct it greedily: starting from a small set Γ(1)\Gamma^{(1)}, at iteration kk we compute the current approximation V^(k)=c(k)TΦΓ(k)\widehat{V}^{(k)}=\mathbf{c}^{(k)\mathsf{T}}\Phi_{\Gamma^{(k)}}, assign to each remaining candidate ϕj\phi_j, j∈Λ(k):=Θ∖Γ(k)j\in\Lambda^{(k)}:=\Theta\setminus\Gamma^{(k)}, an indicator ej(k)≥0e_j^{(k)}\ge0 that predicts the improvement obtained by adding ϕj\phi_j, and set

Γ(k+1)=Γ(k)∪{j∗},j∗∈arg max⁡j∈Λ(k)ej(k),\Gamma^{(k+1)}=\Gamma^{(k)}\cup\{j^{*}\},\qquad j^{*}\in\operatorname*{arg\,max}_{j\in\Lambda^{(k)}}e_j^{(k)},

after which all coefficients are recomputed. The method is thus specified by how c(k)\mathbf{c}^{(k)} is computed and by the choice of ej(k)e_j^{(k)}. We consider two complementary settings.

i) Model-based framework. When data are unavailable or expensive to obtain, we compute the coefficients by a Galerkin projection of the PDE and select basis functions with the residual-decay indicator

ej,PDE(k)=∣⟨L(V^(k)),Lkϕj⟩L2(Ω)∣∥Lkϕj∥L2(Ω),Lk:=DL(V^(k)).e_{j,\mathrm{PDE}}^{(k)}=\frac{\left|\left\langle\mathcal{L}(\widehat{V}^{(k)}),\mathcal{L}_k\phi_j\right\rangle_{L^2(\Omega)}\right|}{\lVert\mathcal{L}_k\phi_j\rVert_{L^2(\Omega)}},\qquad\mathcal{L}_k:=D\mathcal{L}(\widehat{V}^{(k)}).

Its square, divided by two, is the largest decrease of the linearized residual functional α↦12∥L(V^(k))+αLkϕj∥L2(Ω)2\alpha\mapsto\frac{1}{2}\lVert\mathcal{L}(\widehat{V}^{(k)})+\alpha\mathcal{L}_k\phi_j\rVert_{L^2(\Omega)}^2 along ϕj\phi_j. We derive it from the dual-weighted residual (DWR) framework by projecting the adjoint onto span⁡{Lkϕj}\operatorname{span}\{\mathcal{L}_k\phi_j\}, which makes it closed-form and adjoint-free: it requires only the current residual and the linearized operator, and no mesh.

ii) Data-driven framework. Following data-driven approaches to PDE approximation [47], assume we are given a dataset D:={(xi,vi,gi)}i=1N\mathcal{D}:=\{(\mathbf{x}_i,v_i,\mathbf{g}_i)\}_{i=1}^{N} of function values vi=V(xi)v_i=V(\mathbf{x}_i) and gradients gi=∇V(xi)\mathbf{g}_i=\nabla V(\mathbf{x}_i). For HJB equations, such data can be generated in a causality-free fashion by solving open-loop optimal control problems via Pontryagin’s Maximum Principle (PMP), see § 2. We fit (3) by a gradient-augmented Lasso regression on an active subset of DD. Given prescribed weights ω1,ω2>0\omega_1,\omega_2>0 for the value and gradient errors, at each iteration we add both the sample with the largest misfit

ω1∣vi−V^(k)(xi)∣2+ω2∥gi−∇V^(k)(xi)∥22\omega_1\lvert v_i-\widehat{V}^{(k)}(\mathbf{x}_i)\rvert^2+\omega_2\lVert\mathbf{g}_i-\nabla\widehat{V}^{(k)}(\mathbf{x}_i)\rVert_2^2

and the basis function with the largest supervised-loss indicator ej,SL(k)e_{\mathrm{j,SL}}^{(k)}, which measures the exact one-step decrease of the weighted gradient-augmented least-squares loss along ϕj\phi_j. In this way, only a small subset of the available samples is used.

Related literature. Two choices central to this paper are the selection of the approximation space and the states used to construct the approximation. Existing approaches to high-dimensional HJB equations address these choices through different numerical formulations. A first family determines the approximation from the PDE itself. Polynomial Galerkin schemes combined with policy iteration [30], sparse-grid discretizations [8, 25], low-rank tensor formats [17, 18], and neural networks trained by minimizing the HJB residual, either pointwise as in physics-informed and Deep Galerkin methods [53] or over monotone finite-difference schemes [22], fall into this class, as do backward SDE formulations [27]. In these methods the approximation space, that is, the polynomial degree, the grid level, the tensor format or the network architecture, is fixed before the solve, or is refined by local error indicators tied to a grid, as in adaptive sparse grids.

A second family avoids discretizing the state space by working at individual states or along optimal trajectories. Since optimal trajectories are characteristics of the HJB equation, solving open-loop optimal control problems via the PMP yields the value function and its gradient at individual states without a spatial grid [31, 32, 54, 59], and for certain Hamiltonians, Lax–Hopf formulas reduce such pointwise evaluations to convex optimization problems [16]. These samples are then used to fit a global approximation, such as a sparse polynomial [2] or a neural network trained with adaptive sampling [38]; since each sample requires the solution of a two-point boundary value problem, the number of samples largely determines the cost. Related approaches regress on data from state-dependent Riccati equations [1], train networks on the control objective with HJB penalties along trajectories [35, 41], apply symplectic networks to the PMP system [37], or learn the feedback law directly [33].

Adaptive strategies have been developed in several related contexts. In data-driven settings, [49] proposes a greedy point selection algorithm for kernel-based solutions of Dirichlet problems. In machine learning, [52] develops a greedy training algorithm for shallow neural networks that incrementally adds neurons (basis functions) to minimize empirical risk. In optimal control, [19, 20] combine greedy selection of initial states for open-loop solves with greedy kernel approximation, including Hermite interpolation based on value and gradient data. Their subsequent work [21] introduces online enrichment driven by the HJB residual. In these kernel constructions, the approximation space is generated by kernel functions associated with the selected interpolation centers.

In model-based settings, adaptivity is often guided by a posteriori error estimation. A well-established framework is the DWR method, originally developed for finite element methods, which drives adaptive mesh refinement to achieve target accuracy in a prescribed quantity of interest (QoI) [3, 10, 26, 28, 40]. DWR provides a first-order error approximation for the QoI by weighting the primal residual with the solution of a (linearized) adjoint problem, typically computed with the same numerical scheme as the primal one, such as finite elements, discontinuous Galerkin [15], finite volumes [13], or feedforward neural networks [46]. It has been applied to many low-dimensional problems [29, 45, 57] and extended to neural networks [23], reduced-order models [7], multiscale [14] and virtual element methods [50]. However, these formulations remain mesh-based, and their effectiveness depends critically on the accuracy of the adjoint approximation [39]. The approach in [7] projects the adjoint onto a reduced subspace, but still requires solving a reduced adjoint system. Adjoint-based adaptivity for HJB equations has so far been demonstrated only in one- or two-dimensional settings [9, 12, 24, 48].

The proposed greedy enrichment is closely related to Orthogonal Matching Pursuit (OMP) [42, 56] and, more broadly, to greedy dictionary approximation [6, 36]. OMP selects the dictionary element with the largest normalized correlation with the current residual and then refits by least squares over the span of the selected elements. The supervised-loss indicator is a weighted OMP selection rule on the gradient-augmented feature matrix; our data-driven algorithm differs from OMP through the greedy selection of training data and the sparsity-promoting refit. The residual-decay indicator is an OMP-type rule in function space, with dictionary {Lkϕj}\{\mathcal{L}_k\phi_j\}, the PDE residual in place of the data residual, and a Galerkin refit.

Contributions. The main contributions of this paper are as follows.

  1. We derive the closed-form residual-decay indicator ej,PDE(k)e_{j,\mathrm{PDE}}^{(k)} for model-based greedy basis enrichment, show that it predicts the first-order reduction of the PDE residual obtained by adding a single candidate basis function, and specialize it to the HJB operator. It is inspired by the goal-oriented DWR framework, but applies to high-dimensional, mesh-free settings while bypassing any adjoint solve. It also admits an empirical version computed from samples.

  1. We develop a data-driven greedy algorithm for joint basis and training sample selection, which couples a weighted OMP step for the basis with a greedy selection of gradient-augmented training samples.

  1. We demonstrate the proposed framework on benchmark PDEs and nonlinear optimal control problems in up to 35 dimensions. The numerical results show that the greedy strategies achieve comparable accuracy to nonadaptive baselines while using substantially fewer basis functions and, in the data-driven setting, fewer training samples.

The rest of the paper is organized as follows. Section 2 presents the finite-horizon optimal control problem, the associated HJB equation, its link with the PMP, and the large-horizon recovery of the infinite-horizon value function. Section 3 introduces the polynomial dictionaries and the Galerkin and supervised-learning schemes for the coefficients. Sections 4 and 5 develop the model-based and data-driven greedy algorithms, and Section 6 presents the numerical experiments.

Optimal control and Hamilton–Jacobi–Bellman PDEs

A primary motivation of this work is the approximation of nonlinear Hamilton-Jacobi-Bellman PDEs arising in optimal control, where the dimensionality make grid-based discretizations intractable. For an initial time t∈[0,T]t \in[0,T] and an initial state x∈Rd\mathbf{x} \in\mathbb{R}^d, we consider a deterministic nonlinear optimal control problem of the form

min⁡u(⋅)∈L2(T;Rm)J(u;t,x):=∫tTℓ(y(s))+β∥u(s)∥22 ds+ℓT(y(T)),β>0,(4)\min_{\mathbf{u}(\cdot)\in L^2(T;\mathbb{R}^m)} J(\mathbf{u};t,\mathbf{x}) := \int_t^T \ell(\mathbf{y}(s))+\beta\|\mathbf{u}(s)\|_2^2\,\mathrm{d}s+\ell_T(\mathbf{y}(T)), \qquad\beta>0, \tag*{(4)}

subject to y˙(s)=f(y(s))+q(y(s))u(s),y(t)=x\dot{\mathbf{y}}(s)=\mathbf{f}(\mathbf{y}(s))+\mathbf{q}(\mathbf{y}(s))\mathbf{u}(s), \quad\mathbf{y}(t)=\mathbf{x}.

References

  1. [1]G. Albi, S. Bicego, and D. Kalise, Gradient-augmented supervised learning of optimal feedback laws using state-dependent Riccati equations, IEEE Control Systems Letters 6 (2021), 836–841.arxiv.org/abs/2103.04091
  2. [2]B. Azmi, D. Kalise, and K. Kunisch, Optimal feedback law recovery by gradient-augmented sparse polynomial regression, Journal of Machine Learning Research 22 (2021), no. 48, 1–32.arxiv.org/abs/2007.09753
  3. [3]R. Becker and R. Rannacher, An optimal control approach to a posteriori error estimation in finite element methods, Acta Numerica 10 (2001), 1–102.
  4. [4]R. E. Bellman, Adaptive processes: a guided tour, Princeton university press, 1961.
  5. [5]D. Bertsekas, Dynamic programming and optimal control: Volume i, Vol. 4, Athena scientific, 2012.
  6. [6]P. Binev, A. Cohen, W. Dahmen, R. DeVore, G. Petrova, and P. Wojtaszczyk, Convergence rates for greedy algorithms in reduced basis methods, SIAM Journal on Mathematical Analysis 43 (2011), no. 3, 1457–1472.
  7. [7]P. J. Blonigan and E. J. Parish, Evaluation of dual-weighted residual and machine learning error estimation for projection-based reduced-order models of steady partial differential equations, Computer Methods in Applied Mechanics and Engineering 409 (2023), 115988.DOI
  8. [8]O. Bokanowski, J. Garcke, M. Griebel, and I. Klompmaker, An adaptive sparse grid semi-Lagrangian scheme for first order Hamilton–Jacobi Bellman equations, Journal of Scientific Computing 55 (2013), no. 3, 575–605.DOI
  9. [9]S. C. Brenner and E. L. Kawecki, Adaptive C⁰ interior penalty methods for Hamilton–Jacobi–Bellman equations with Cordes coefficients, Journal of Computational and Applied Mathematics 388 (2021), 113241.
  10. [10]M. P. Bruchhäuser, K. Schwegler, and M. Bause, Dual weighted residual based error control for nonstationary convection-dominated equations: potential or ballast?, Boundary and interior layers, computational and asymptotic methods bail 2018, 2020, pp. 1–17.DOI
  11. [11]H.-J. Bungartz and M. Griebel, Sparse grids, Acta Numerica 13 (2004), 147–269.
  12. [12]F. Cagnetti, D Gomes, and H. V. Tran, Convergence of a semi-discretization scheme for the Hamilton–Jacobi equation: a new approach with the adjoint method, Applied Numerical Mathematics 73 (2013), 2–15.DOI
  13. [13]Q. Chen and M. Gunzburger, Goal-oriented a posteriori error estimation for finite volume methods, Journal of computational and applied mathematics 265 (2014), 69–82.arxiv.org/abs/1011.0368
  14. [14]E. T Chung, W. T. Leung, and S. Pollock, Goal-oriented adaptivity for GMsFEM, Journal of Computational and Applied Mathematics 296 (2016), 625–637.arxiv.org/abs/1509.05643
  15. [15]B. Cockburn and S. Xia, An adjoint-based adaptive error approximation of functionals by the hybridizable discontinuous Galerkin method for second-order elliptic equations, Journal of Computational Physics 457 (2022), 111078.DOI
  16. [16]J. Darbon and S. Osher, Algorithms for overcoming the curse of dimensionality for certain Hamilton–Jacobi equations arising in control theory and elsewhere, Research in the Mathematical Sciences 3 (2016), 19.arxiv.org/abs/1605.01799
  17. [17]S. Dolgov, D. Kalise, and L. Saluzzi, Data-driven tensor train approximation for Hamilton–Jacobi–Bellman equations, SIAM Journal on Scientific Computing 45 (2023), no. 5, A2153–A2184.
  18. [18]S. Dolgov, D. Kalise, and K. K Kunisch, Tensor decomposition methods for high-dimensional Hamilton–Jacobi–Bellman equations, SIAM Journal on Scientific Computing 43 (2021), no. 3, A1625–A1650.arxiv.org/abs/1908.01533
  19. [19]T. Ehring and B. Haasdonk, Greedy sampling and approximation for realizing feedback control for high dimensional nonlinear systems, IFAC-PapersOnLine 55 (2022), no. 20, 325–330.DOI
  20. [20]———, Hermite kernel surrogates for the value function of high-dimensional nonlinear optimal control problems, Advances in Computational Mathematics 50 (2024), no. 3, 36.
  21. [21]———, Online adaptive surrogates for the value function of high-dimensional nonlinear optimal control problems, IFAC-PapersOnLine 59 (2025), no. 1, 331–336.DOI
  22. [22]C. Esteve-Yagüe, R. Tsai, and A. Massouco, Finite-difference least square methods for solving Hamilton–Jacobi equations using neural networks, arXiv preprint arXiv:2406.10758 (2024).
  23. [23]K. Fidkowski, Adjoint-based adaptive training of deep neural networks, AIAA aviation 2021 forum, 2021, pp. 2904.DOI
  24. [24]D. Gallistl and E. Suli, Mixed finite element approximation of the Hamilton–Jacobi–Bellman equation with Cordes coefficients, SIAM Journal on Numerical Analysis 57 (2019), no. 2, 592–614.DOI
  25. [25]J. Garcke and A. Kröner, Suboptimal feedback control of PDEs by solving HJB equations on adaptive sparse grids, Journal of Scientific Computing 70 (2017), no. 1, 1–28.DOI
  26. [26]M. B Giles and E. Süli, Adjoint methods for PDEs: a posteriori error analysis and postprocessing by duality, Acta Numerica 11 (2002), 145–236.
  27. [27]J. Han, A. Jentzen, and W. E, Solving high-dimensional partial differential equations using deep learning, Proceedings of the National Academy of Sciences 115 (2018), no. 34, 8505–8510.DOI
  28. [28]V. Helml, M. Innerberger, and D. Praetorius, Plain convergence of goal-oriented adaptive FEM, Computers and Mathematics with Applications 147 (2023), 130–149.DOI
  29. [29]S. Jayasinghe, D. L Darmofal, N. K Burgess, M. C Galbraith, and S. R Allmaras, A space-time adaptive method for reservoir flows: formulation and one-dimensional application, Computational Geosciences 22 (2018), no. 1, 107–123.DOI
  30. [30]D. Kalise and K. Kunisch, Polynomial approximation of high-dimensional Hamilton–Jacobi–Bellman equations and applications to feedback control of semilinear parabolic PDEs, SIAM Journal on Scientific Computing 40 (2018jan), no. 2, A629–A652.DOI
  31. [31]W. Kang and L. Wilcox, A causality free computational method for HJB equations with application to rigid body satellites, Aiaa guidance, navigation, and control conference, 2015, pp. 2009.DOI
  32. [32]W. Kang and L. C Wilcox, Mitigating the curse of dimensionality: sparse grid characteristics method for optimal feedback control and HJB equations, Computational Optimization and Applications 68 (2017), no. 2, 289–315.arxiv.org/abs/1507.04769
  33. [33]K. Kunisch and D. Walter, Semiglobal optimal feedback stabilization of autonomous ODEs via deep neural network approximation, ESAIM: Control, Optimisation and Calculus of Variations 27 (2021), 16.arxiv.org/abs/2002.08625
  34. [34]R. J LeVeque, Finite difference methods for ordinary and partial differential equations: steady-state and time-dependent problems, SIAM, 2007.
  35. [35]X. Li, D. Verma, and L. Ruthotto, A neural network approach for stochastic optimal control, SIAM Journal on Scientific Computing 46 (2024), no. 5, C535–C556.arxiv.org/abs/2209.13104
  36. [36]S. G. Mallat and Z. Zhang, Matching pursuits with time-frequency dictionaries, IEEE Transactions on Signal Processing 41 (1993), no. 12, 3397–3415.DOI
  37. [37]T. Meng, Z. Zhang, J. Dorban, and G. Karniadakis, SympOCnet: Solving optimal control problems with applications to high-dimensional multiagent path planning problems, SIAM Journal on Scientific Computing 44 (2022), no. 6, B1341–B1368.DOI
  38. [38]K. Nakamura-Zimmerer, Q. Gong, and W. Kang, Adaptive deep learning for high-dimensional Hamilton–Jacobi–Bellman equations, SIAM Journal on Scientific Computing 43 (2021), no. 2, A1221–A1247.
  39. [39]R. H Nochetto, A. Veeser, and M. Verani, A safeguarded dual weighted residual method, IMA Journal of Numerical Analysis 29 (2009), no. 1, 126–140.DOI
  40. [40]J. T. Oden and S. Prudhomme, Goal-oriented error estimation and adaptivity for the finite element method, Computers & Mathematics with Applications 41 (2001), no. 5-6, 735–756.DOI
  41. [41]D. Onken, L. Nurbekyan, X. Li, S. W. Fung, S. Osher, and L. Ruthotto, A neural network approach for high-dimensional optimal control applied to multiagent path finding, IEEE Transactions on Control Systems Technology 31 (2022), no. 1, 235–251.DOI
  42. [42]Y. C. Pati, R. Rezaiifar, and P. S. Krishnaprasad, Orthogonal matching pursuit: Recursive function approximation with applications to wavelet decomposition, Proceedings of 27th asilomar conference on signals, systems and computers, 1993, pp. 40–44.
  43. [43]L. S. Pontryagin, Mathematical theory of optimal processes, Routledge, 2018.DOI
  44. [44]A. Quarteroni, A. Manzoni, and F. Negri, Reduced basis methods for partial differential equations: an introduction, Springer, 2015.DOI
  45. [45]E. Rabizadeh, A. S. Bagherzadeh, C. Anitescu, N. Alajlan, and T. Rabczuk, Pointwise dual weighted residual based goal-oriented a posteriori error estimation and adaptive mesh refinement in 2D/3D thermo-mechanical multifield problems, Computer Methods in Applied Mechanics and Engineering 359 (2020), 112666.DOI
  46. [46]J. Roth, M. Schröder, and T. Wick, Neural network guided adjoint computations in dual weighted residual error estimation, SN Applied Sciences 4 (2022), no. 2, 62.arxiv.org/abs/2102.12450
  47. [47]S. Rudy, A. Alla, S. L Brunton, and J N. Kutz, Data-driven identification of parametric partial differential equations, SIAM Journal on Applied Dynamical Systems 18 (2019), no. 2, 643–660.arxiv.org/abs/1806.00732
  48. [48]M. Sagona and A. Seghini, An a posteriori error estimate for a semi-Lagrangian scheme for Hamilton–Jacobi equations, Numerical Algorithms 33 (2003), 453–460.DOI
  49. [49]R. Schaback, A greedy method for solving classes of PDE problems, arXiv preprint arXiv:1903.11536 (2019).arxiv.org/abs/1903.11536
  50. [50]C Sellmann and P Junker, Goal-oriented dual-weighted residual error estimation for the virtual elements method, Computer Methods in Applied Mechanics and Engineering 443 (2025), 118034.DOI
  51. [51]J. Shen, T. Tang, and L.-L. Wang, Spectral methods: algorithms, analysis and applications, Vol. 41, Springer Science & Business Media, 2011.DOI
  52. [52]J. W Siegel, Q. Hong, X. Jin, W. Hao, and J. Xu, Greedy training algorithms for neural networks and applications to PDEs, Journal of Computational Physics 484 (2023), 112084.arxiv.org/abs/2107.04466
  53. [53]J. Sirignano and K. Spiliopoulos, DGM: A deep learning algorithm for solving partial differential equations, Journal of Computational Physics 375 (2018), 1339–1364.DOI
  54. [54]N. Subbotina, The method of characteristics for Hamilton–Jacobi equations and applications to dynamical optimization, Journal of Mathematical Sciences 135 (2006), no. 3, 2955–3091.DOI
  55. [55]F.-T. Suttmeier, Numerical solution of variational inequalities by adaptive finite elements, Springer, 2008.DOI
  56. [56]J. A. Tropp, Greed is good: algorithmic results for sparse approximation, IEEE Transactions on Information Theory 50 (2004), no. 10, 2231–2242.DOI
  57. [57]D. A Venditti and D. L Darmofal, Anisotropic grid adaptation for functional outputs: application to two-dimensional viscous flows, Journal of Computational Physics 187 (2003), no. 1, 22–46.DOI
  58. [58]H. Wendland, Scattered data approximation, Vol. 17, Cambridge university press, 2004.DOI
  59. [59]I. Yegorov and P. M. Dower, Perspectives on characteristics-based curse-of-dimensionality-free numerical approaches for solving Hamilton–Jacobi equations, Applied Mathematics and Optimization 81 (2021), 1167–1218.
  60. [60]X. Zhou, Maximum principle, dynamic programming, and their connection in deterministic control, Journal of Optimization Theory and Applications 65 (1990), no. 2, 363–373.DOI
  61. [61]O. C. Zienkiewicz, R. L. Taylor, P. Nithiarasu, and J. Zhu, The finite element method, Vol. 3, Elsevier, 2000.

Paper details

vlnc:W_qfzsm29a
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Mathematics

Not recorded for this paper

View statement
Looking for related papers…
Contents