Hostname: page-component-848d4c4894-p2v8j Total loading time: 0 Render date: 2024-06-11T14:45:41.692Z Has data issue: false hasContentIssue false
  • English
  • Français

Hybrid PDE solver for data-driven problems and modern branching†

Published online by Cambridge University Press:  22 May 2017

FRANCISCO BERNAL ,
GONÇALO DOS REIS  and
GREIG SMITH
FRANCISCO BERNAL
Affiliation:
CMAP - Centre de Mathématiques Appliquées, Ecole Polytechnique, Route de Saclay, 91128 Palaiseau Cedex, France email: Francisco.Bernal@polytechnique.edu
GONÇALO DOS REIS
Affiliation:
School of Mathematics, University of Edinburgh, Edinburgh, EH9 3FD, UK email: G.dosReis@ed.ac.uk Centro de Matemática e Aplicações (CMA), FCT, UNL, 2829-516 Caparica, Portugal
GREIG SMITH
Affiliation:
School of Mathematics, University of Edinburgh, Edinburgh, EH9 3FD, UK email: G.dosReis@ed.ac.uk Maxwell Institute Graduate School in Analysis and its Applications, University of Edinburgh, Edinburgh, EH9 3FD, UK email: G.Smith-13@sms.ed.ac.uk
Rights & Permissions [Opens in a new window]

Abstract

Core share and HTML view are not available for this content. However, as you have access to this content, a full PDF is available via the ‘Save PDF’ action button.

The numerical solution of large-scale PDEs, such as those occurring in data-driven applications, unavoidably require powerful parallel computers and tailored parallel algorithms to make the best possible use of them. In fact, considerations about the parallelization and scalability of realistic problems are often critical enough to warrant acknowledgement in the modelling phase. The purpose of this paper is to spread awareness of the Probabilistic Domain Decomposition (PDD) method, a fresh approach to the parallelization of PDEs with excellent scalability properties. The idea exploits the stochastic representation of the PDE and its approximation via Monte Carlo in combination with deterministic high-performance PDE solvers. We describe the ingredients of PDD and its applicability in the scope of data science. In particular, we highlight recent advances in stochastic representations for non-linear PDEs using branching diffusions, which have significantly broadened the scope of PDD. We envision this work as a dictionary giving large-scale PDE practitioners references on the very latest algorithms and techniques of a non-standard, yet highly parallelizable, methodology at the interface of deterministic and probabilistic numerical methods. We close this work with an invitation to the fully non-linear case and open research questions.

Keywords

Probabilistic Domain Decompositionhigh-performance parallel computingMarked branching diffusionshybrid non-linear PDE solversMonte Carlo methodsPrimary: 65C0565C30Secondary: 65N5560H3591-XX35CXX
Type
Papers
Information
European Journal of Applied Mathematics , Volume 28 , Special Issue 6: Big Data and Partial Differential Equations , December 2017 , pp. 949 - 972
Copyright
Copyright © Cambridge University Press 2017 

Footnotes

†

F. Bernal acknowledges funding from Centre de Mathématiques Appliquées (CMAP), École Polytechnique. G. dos Reis gratefully thanks the partial support by the Fundação para a Ciência e a Tecnologia (Portuguese Foundation for Science and Technology) through the project UID/MAT/00297/2013 (Centro de Matemática e Aplicações). G. Smith was supported by The Maxwell Institute Graduate School in Analysis and its Applications, a Centre for Doctoral Training funded by the UK Engineering and Physical Sciences Research Council (grant [EP/L016508/01]), the Scottish Funding Council, Heriot-Watt University and the University of Edinburgh.

References

[1] Acebrón, J. A., Busico, M. P., Lanucara, P. & Spigler, R. (2005a) Domain decomposition solution of elliptic boundary-value problems via Monte Carlo & quasi-Monte Carlo methods. SIAM J. Sci. Comput. 27 (2), 440–457.CrossRefGoogle Scholar
[2] Acebrón, J. A., Busico, M. P., Lanucara, P. & Spigler, R. (2005b) Probabilistically induced domain decomposition methods for elliptic boundary-value problems. J. Comput. Phys. 210 (2), 421–438.CrossRefGoogle Scholar
[3] Acebrón, J. A. & Ribeiro, M. A. (2016) A Monte Carlo method for solving the one-dimensional Telegraph equations with boundary conditions. J. Comput. Phys. 305, 29–43.Google Scholar
[4] Acebrón, J. A. & Rodríguez-Rozas, Á. (2011) A new parallel solver suited for arbitrary semilinear parabolic partial differential equations based on generalized random trees. J. Comput. Phys. 230 (21), 7891–7909.Google Scholar
[5] Acebrón, J. A. & Rodríguez-Rozas, Á. (2013) Highly efficient numerical algorithm based on random trees for accelerating parallel Vlasov–Poisson simulations. J. Comput. Phys. 250, 224–245.Google Scholar
[6] Acebrón, J. A., Rodríguez-Rozas, Á. & Spigler, R. (2009) Domain decomposition solution of nonlinear two-dimensional parabolic problems by random trees. J. Comput. Phys. 228 (15), 5574–5591.Google Scholar
[7] Acebrón, J. A., Rodríguez-Rozas, Á. & Spigler, R. (2010a) Efficient parallel solution of nonlinear parabolic partial differential equations by a probabilistic domain decomposition. J. Sci. Comput. 43 (2), 135–157.Google Scholar
[8] Acebrón, J. A., Rodríguez-Rozas, Á. & Spigler, R. (2010b) A fully scalable algorithm suited for petascale computing and beyond. Comput. Sci. Res. Dev. 25 (1–2), 115–121.CrossRefGoogle Scholar
[9] Agarwal, A. & Claisse, J. (2017) Branching diffusion representation of quasi-linear elliptic pdes and estimation using monte carlo method. preprint, arXiv:1704.00328.Google Scholar
[10] Benth, F. E., Karlsen, K. H. & Reikvam, K. (2003) A semilinear Black and Scholes partial differential equation for valuing American options. Finance Stoch. 7 (3), 277–298.Google Scholar
[11] Bernal, F. & Acebrón, J. A. (2016a) A comparison of higher-order weak numerical schemes for stopped stochastic differential equations. Comm. Comput. Phys. 20 (3), 703–732.CrossRefGoogle Scholar
[12] Bernal, F. & Acebrón, J. A. (2016b) A multigrid-like algorithm for probabilistic domain decomposition. Comput. Math.Appl. 72 (7), 1790–1810.Google Scholar
[13] Bernal, F., Acebrón, J. A. & Anjam, I. (2014) A stochastic algorithm based on fast marching for automatic capacitance extraction in non-Manhattan geometries. SIAM J. Imag. Sci. 7 (4), 2657–2674.CrossRefGoogle Scholar
[14] Bihlo, A. & Haynes, R. D. (2014) Parallel stochastic methods for PDE based grid generation. Comput. Math. Appl. 68 (8), 804–820.Google Scholar
[15] Bihlo, A. & Haynes, R. D. (2016) A stochastic domain decomposition method for time dependent mesh generation. In: Domain Decomposition Methods in Science and Engineering XXII. Dickopf, T., Gander, M. J., Halpern, L., Krause, R., & Pavarino, L. F. (editors), Springer, Vol. 104, pp. 107–115.Google Scholar
[16] Bihlo, A., Haynes, R. D. & Walsh, E. (2015) Stochastic domain decomposition for time dependent adaptive mesh generation. J. Math. Stud. 48 (2), 106–124.Google Scholar
[17] Bossy, M., Champagnat, N., Leman, H., Maire, S., Violeau, L. & Yvinec, M. (2015) Monte Carlo methods for linear and non-linear Poisson-Boltzmann equation. ESAIM: Proc. Surv. 48, 420–446.Google Scholar
[18] Bouchard, B., Elie, R. & Touzi, N. (2009) Discrete-time approximation of BSDEs and probabilistic schemes for fully nonlinear PDEs. In: Advanced Financial Modelling, Albrecher, H., Runggaldier, W. J. & Schachermayer, W. (editors), Radon Ser. Comput. Appl. Math., Vol. 8, Walter de Gruyter, Berlin, pp. 91–124.Google Scholar
[19] Bouchard, B., Tan, X. & Zou, Y. (2016) Numerical approximation of BSDEs using local polynomial drivers and branching processes. arXiv:1612.06790.Google Scholar
[20] Cheridito, P., Soner, H. M., Touzi, N. & Victoir, N. (2007) Second-order backward stochastic differential equations and fully nonlinear parabolic PDEs. Comm. Pure Appl. Math. 60 (7), 1081–1110.Google Scholar
[21] Costantini, C., Pacchiarotti, B. & Sartoretto, F. (1998) Numerical approximation for functionals of reflecting diffusion processes. SIAM J. Appl. Math. 58 (1), 73–102.Google Scholar
[22] Crisan, D. & Manolarakis, K. (2010) Probabilistic methods for semilinear partial differential equations. Applications to finance. Math. Modelling Numer. Anal. 44 (5), 1107.Google Scholar
[23] Cruzeiro, A. B. & Shamarova, E. (2009) Navier–Stokes equations and forward–backward SDEs on the group of diffeomorphisms of a torus. Stoch. Process. Appl. 119 (12), 4034–4060.Google Scholar
[24] Dancer, E. N. & Du, Y. H. (1994) Competing species equations with diffusion, large interactions, and jumping nonlinearities. J. Differ. Equ. 114 (2), 434–475.Google Scholar
[25] Doumbia, M., Oudjane, N., & Warin, X. (2017) Unbiased Monte Carlo estimate of stochastic differential equations expectations. ESAIM: Probability and Statistics 21, 56–87.CrossRefGoogle Scholar
[26] Dynkin, E. B. (2004) Superdiffusions and Positive Solutions of Nonlinear Partial Differential Equations, University Lecture Series, Vol. 34, American Mathematical Society, Providence, RI. Appendix A by J.-F. Le Gall and Appendix B by I. E. Verbitsky.Google Scholar
[27] El Karoui, N., Kapoudjian, C., Pardoux, E., Peng, S. & Quenez, M. C. (1997) Reflected solutions of backward SDEs, and related obstacle problems for PDEs. Ann. Probab. 25 (2), 702–737.CrossRefGoogle Scholar
[28] El Karoui, N., Peng, S. & Quenez, M. C. (1997) Backward stochastic differential equations in finance. Math. Finance 7 (1), 1–71.Google Scholar
[29] Escher, J. & Matioc, A.-V. (2010) Radially symmetric growth of nonnecrotic tumors. Nonlinear Differ. Equ. Appl. NoDEA 17 (1), 1–20.CrossRefGoogle Scholar
[30] Fahim, A., Touzi, N. & Warin, X. (2011) A probabilistic numerical method for fully nonlinear parabolic PDEs. Ann. Appl. Probab. 21 (4), 1322–1364.Google Scholar
[31] Fisher, R. A. (1937) The wave of advance of advantageous genes. Ann. eugenics 7 (4), 355–369.Google Scholar
[32] Frei, C. & dos Reis, G. (2013) Quadratic FBSDE with generalized Burgers' type nonlinearities, perturbations and large deviations. Stoch. Dynam. 13 (2), 1250015.Google Scholar
[33] Freidlin, M. (1985) Functional integration and Partial Differential Equations, Annals of Mathematics Studies, Vol. 109, Princeton University Press, Princeton, NJ.Google Scholar
[34] Giles, M. B. & Bernal, F. (2017) Multilevel estimation of expected exit times and other functionals of stopped diffusions. Submitted.CrossRefGoogle Scholar
[35] Gobet, E. (2001) Euler schemes and half-space approximation for the simulation of diffusion in a domain. ESAIM Probab. Statist. 5, 261–297.CrossRefGoogle Scholar
[36] Gobet, E. (2016) Monte–Carlo Methods and Stochastic Processes: From Linear to Non-Linear, CRC Press.CrossRefGoogle Scholar
[37] Gobet, E., Liu, G. & Zubelli, J. (2016) A non-intrusive stratified resampler for regression Monte Carlo: Application to solving non-linear equations. hal-01291056.Google Scholar
[38] Gobet, E. & Maire, S. (2005) Sequential Monte Carlo domain decomposition for the Poisson equation. In: Proceedings of the 17th IMACS World Congress, Scientific Computation, Applied Mathematics and Simulation (11–15 July 2005, Paris).Google Scholar
[39] Gobet, E. & Menozzi, S. (2010) Stopped diffusion processes: Boundary corrections and overshoot. Stoch. Process. Appl. 120 (2), 130–162.Google Scholar
[40] Guyon, J. & Henry-Labordère, P. (2013) Nonlinear Option Pricing, CRC Press.Google Scholar
[41] Henry-Labordere, P. (2012) Counterparty risk valuation: A marked branching diffusion approach. Available at SSRN: https://ssrn.com/abstract=1995503.Google Scholar
[42] Henry-Labordere, P., Oudjane, N., Tan, X., Touzi, N. & Warin, X. (2016) Branching diffusion representation of semilinear PDEs and Monte Carlo approximation. arXiv:1603.01727.Google Scholar
[43] Henry-Labordere, P., Tan, X. & Touzi, N. (2014) A numerical algorithm for a class of BSDEs via the branching process. Stoch. Process. Appl. 124 (2), 1112–1140.Google Scholar
[44] Higham, D. J., Mao, X., Roj, M., Song, Q. & Yin, G. (2013) Mean exit times and the multilevel Monte Carlo method. SIAM/ASA J. Uncertain. Quantification 1 (1), 2–18.Google Scholar
[45] Karatzas, I. & Shreve, S. (1991) Brownian Motion and Stochastic Calculus, Vol. 113, Springer-Verlag, New York.Google Scholar
[46] Kloeden, P. E. & Platen, E. (1992) Numerical Solution of Stochastic Differential Equations, Applications of Mathematics (New York), Vol. 23, Springer-Verlag, Berlin.Google Scholar
[47] Kolmogorov, A. N., Petrovsky, I. G. & Piskunov, N. S. (1937) Étude de l'équation de la diffusion avec croissance de la quantité de matiere et son application a un probleme biologique. Moscow Univ. Math. Bull 1 (1–25), 129.Google Scholar
[48] Lionnet, A., dos Reis, G. & Szpruch, L. (2015) Time discretization of FBSDE with polynomial growth drivers and reaction–diffusion PDEs. Ann. Appl. Probab. 25 (5), 2563–2625.Google Scholar
[49] Lionnet, A., dos Reis, G. & Szpruch, L. (2016) Convergence and properties of modified explicit schemes for BSDEs with polynomial growth. arXiv:1607.06733.Google Scholar
[50] Ma, J. & Yong, J. (1999) Forward-Backward Stochastic Differential Equations and Their Applications, Lecture Notes in Mathematics, Vol. 1702, Springer-Verlag, Berlin.Google Scholar
[51] McKean, H. P. (1975) Application of Brownian motion to the equation of Kolmogorov-Petrovskii-Piskunov. Commun. Pure Appl. Math. 28 (3), 323–331.Google Scholar
[52] Mendes, R. V. (2010) Poisson–Vlasov in a strong magnetic field: A stochastic solution approach. J. Math. Phys. 51 (4), 043101.CrossRefGoogle Scholar
[53] Milstein, G. N. & Tretyakov, M. V. (2004) Stochastic Numerics for Mathematical Physics, Springer-Verlag, Berlin.Google Scholar
[54] Pardoux, É. & Peng, S. (1992) Backward stochastic differential equations and quasilinear parabolic partial differential equations. In: Stochastic partial differential equations and their Applications (Charlotte, NC, 1991), Lec. Notes in Control and Inform. Sci., Vol. 176, Springer, Berlin, pp. 200–217.Google Scholar
[55] Peng, S. (1991) Probabilistic interpretation for systems of quasilinear parabolic partial differential equations. Stoch. Stoch. Rep. 37 (1–2), 61–74.Google Scholar
[56] Rasulov, A., Raimova, G. & Mascagni, M. (2010) Monte Carlo solution of Cauchy problem for a nonlinear parabolic equation. Math. Comput. Simul. 80 (6), 1118–1123.Google Scholar
[57] Skorokhod, A. V. (1964) Branching diffusion processes. Teor. Verojatnost. i Primenen. 9 (3), 492–497.Google Scholar
[58] Smith, B., Bjorstad, P. & Gropp, W. (2004) Domain Decomposition: Parallel Multilevel Methods for Elliptic Partial Differential Equations, Cambridge, Cambridge University Press.Google Scholar
[59] Struwe, M. (1996) Geometric evolution problems. In: Nonlinear Partial Differential Equations in Differential Geometry (Park City, UT, 1992). IAS/Park City Math. Ser., Vol. 2, Amer. Math. Soc., Providence, RI, pp. 257–339.Google Scholar
[60] Warin, X. (2017) Variations on branching methods for nonlinear PDEs. arXiv:1701.07660.Google Scholar
[61] Watanabe, S. (1965) On the branching process for Brownian particles with an absorbing boundary. J. Math. Kyoto Univ. 4 (2), 385–398.Google Scholar
[62] Xu, Y. (2015) A complex Feynman-Kac formula via linear backward stochastic differential equations. arXiv:1505.03590.Google Scholar