Log in

Approximation Schemes for McKean–Vlasov and Boltzmann-Type Equations (Error Analysis in Total Variation Distance)

  • Published:
Journal of Theoretical Probability Aims and scope Submit manuscript

Abstract

We deal with McKean–Vlasov and Boltzmann-type jump equations. This means that the coefficients of the stochastic equation depend on the law of the solution, and the equation is driven by a Poisson point measure with intensity measure which depends on the law of the solution as well. Alfonsi and Bally (Construction of Boltzmann and McKean Vlasov type flows (the sewing lemma approach), 2021, ar**v:2105.12677) have proved that under some suitable conditions, the solution \(X_t\) of such equation exists and is unique. One also proves that \(X_t\) is the probabilistic interpretation of an analytical weak equation. Moreover, the Euler scheme \(X_t^{{\mathcal {P}}}\) of this equation converges to \(X_t\) in Wasserstein distance. In this paper, under more restrictive assumptions, we show that the Euler scheme \(X_t^{{\mathcal {P}}}\) converges to \(X_t\) in total variation distance and \(X_t\) has a smooth density (which is a function solution of the analytical weak equation). On the other hand, in view of simulation, we use a truncated Euler scheme \(X^{{\mathcal {P}},M}_t\) which has a finite numbers of jumps in any compact interval. We prove that \(X^{{\mathcal {P}},M}_{t}\) also converges to \(X_t\) in total variation distance. Finally, we give an algorithm based on a particle system associated with \(X^{{\mathcal {P}},M}_t\) in order to approximate the density of the law of \(X_t\). Complete estimates of the error are obtained.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Subscribe and save

Springer+ Basic
EUR 32.99 /Month
  • Get 10 units per month
  • Download Article/Chapter or Ebook
  • 1 Unit = 1 Article or 1 Chapter
  • Cancel anytime
Subscribe now

Buy Now

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Similar content being viewed by others

Data Availability

Data sharing was not applicable to this article as no datasets were generated or analyzed during the current study.

References

  1. Albeverio, S., Rüdiger, B., Sundar, P.: The Enskog process. J. Stat. Phys. 167(1), 90–122 (2017)

    MathSciNet  Google Scholar 

  2. Alexandre, R.: A review of Boltzmann equation with singular kernels. Kinet. Relat. Models 2(4), 551–646 (2009)

    MathSciNet  Google Scholar 

  3. Alfonsi, A., Bally, V.: Construction of Boltzmann and McKean Vlasov type flows (the sewing lemma approach). ar**v:2105.12677 [math.PR] (2021)

  4. Alfonsi, A., Jourdain, B., Kohatsu-Higa, A.: Optimal transport bounds between the time-marginals of a multidimensional diffusion and its Euler scheme. Electron. J. Probab. 20(6), 1–31 (2015)

    MathSciNet  Google Scholar 

  5. Antonelli, F., Kohatsu-Higa, A.: Rate of convergence of a particle method to the solution of the McKean-Vlasov equation. Ann. Appl0. Probab. 12(2), 423–476 (2002)

    MathSciNet  Google Scholar 

  6. Applebaum, D.: Lévy Processes and Stochastic Calculus. Cambridge Studies in Advanced Mathematics, 2nd edn. Cambridge University Press, Cambridge (2009). https://doi.org/10.1017/CBO9780511809781

    Book  Google Scholar 

  7. Bally, V., Caramellino, L., Poly, G.: Regularization lemmas and convergence in total variation. Electron. J. Probab. 25, 1–20 (2020)

    MathSciNet  Google Scholar 

  8. Bally, V., Clément, E.: Integration by parts formula and applications to equations with jumps. Probab. Theory Relat. Fields 151, 613–657 (2011)

    MathSciNet  Google Scholar 

  9. Bally, V., Fournier, N.: Regularization properties of the 2D homogeneous Boltzmann equation without cutoff. Probab. Theory Relat. Fields 151(3–4), 659–704 (2011)

    MathSciNet  Google Scholar 

  10. Bally, V., Qin, Y.: Total variation distance between a jump-equation and its Gaussian approximation. Stoch. Partial Differ. Equ. Anal. Comput. 10(3), 1211–1260 (2022)

    MathSciNet  Google Scholar 

  11. Bally, V., Talay, D.: The Law of the Euler scheme for stochastic differential equations: I. convergence rate of the distribution function. [Research Report] RR-2244, INRIA. (1994)

  12. Bao, J., Reisinger, C., Ren, P., et al.: First-order convergence of Milstein schemes for McKean-Vlasov equations and interacting particle systems. Proc. R. Soc. A 477(2245), 20200258 (2021)

    MathSciNet  Google Scholar 

  13. Bichteler, K., Gravereaux, J.B., Jacod, J.: Malliavin Calculus for Processes with Jumps. Gordon and Breach, Philadelphia (1987)

    Google Scholar 

  14. Carmona, R., Delarue, F.: Probability Theory of Mean Field Games with Applications. Probability Theory and Stochastic Modelling, Springer, Berlin (2018)

    Google Scholar 

  15. Cercignani, C.: The Boltzmann Equation and Its Applications. Applied Mathematical Sciences, Springer, New York (1988)

    Google Scholar 

  16. Cont, R., Tankov, P.: Finacial Modelling with Jump Processes. Chapman & Hall/CRC, Boca Raton (2004)

    Google Scholar 

  17. Desvillettes, L., Graham, C., Méléard, S.: Probabilistic interpretation and numerical approximation of a Kac equation without cutoff. Stoch. Process. Appl. 84(1), 115–135 (1999)

    MathSciNet  Google Scholar 

  18. Fournier, N., Guillin, A.: From a Kac-like particle system to the Landau equation for hard potentials and Maxwell molecules. Ann. Sci. Éc. Norm. Supér.(4) 50(1), 157–199 (2017)

    MathSciNet  Google Scholar 

  19. Fournier, N., Mischler, S.: Rate of convergence of the Nanbu particle system for hard potentials and Maxwell molecules. Ann. Probab. 44(1), 589–627 (2016)

    MathSciNet  Google Scholar 

  20. Friesen, M., Rüdiger, B., Sundar, P.: The Enskog process for hard and soft potentials. NoDEA Nonlinear Differ. Equ. Appl. 26(3), 42 (2019)

    MathSciNet  Google Scholar 

  21. Friesen, M., Rüdiger, B. , Sundar, P.: On uniqueness and stability for the Enskog equation (2020)

  22. Gobet, E., Pagès, G., Pham, H., Printems, J.: Discretization and simulation for a class of SPDEs with applications to Zakai and McKean-Vlasov equations (2005)

  23. Graham, C.: Mckean-Vlasov Ito-Skorohod equations, and nonlinear diffusions with discrete jump sets. Stoch. Process. Appl. 40, 69–82 (1992)

    MathSciNet  Google Scholar 

  24. Haji-Ali, A.L., Tempone, R.: Multilevel and multi-index monte Carlo methods for the McKean-Vlasov equation. Stat. Comput. 28(4), 923–935 (2018)

    MathSciNet  Google Scholar 

  25. Ikeda, N., Watanabe, S.: Stochastic Differential Equations and Diffusion Processes, 2nd edn. Elsevier, Amsterdam (1989)

    Google Scholar 

  26. Ishikawa, Y.: Stochastic Calculus of Variations for Jump Processes, Berlin. De Gruyter, Boston (2013)

    Google Scholar 

  27. Jourdain, B., Kohatsu-Higa, A.: A review of recent results on approximation of solutions of stochastic differential equations. In: Proceedings of the Workshop on Stochastic Analysis with Financial Applications: Hong Kong (2009). Birkhauser (2011)

  28. Kohatsu-Higa, A.: The Euler approximation for stochastic differential equations with boundary conditions. In: Proceedings of the Workshop on Turbulent Diffusion and Related Problems in Stochastic Numerics. The Institute of Statistical Mathematics, Tokyo (1996)

  29. Kohatsu-Higa, A., Ogawa, S.: Monte Carlo methods weak rate of convergence for an Euler scheme of nonlinear SDE’s. Monte Carlo Methods Appl. 3, 327–345 (1997)

    MathSciNet  Google Scholar 

  30. Kohatsu-Higa, A., Protter, P.: The Euler scheme for SDE’s driven by semimartingales. In: Kunita, H. and Kuo, H. (eds.) Stochastic Analysis on Infinite Dimensional Spaces, vol. 310, pp. 141–151 (1994)

  31. Kohatsu-Higa, A., Tankov, P.: Jump-adapted discretization schemes for Lévy-driven SDEs. Stoch. Process. Appl. 120, 2258–2285 (2010)

    Google Scholar 

  32. Kulik, A.M.: Malliavin calculus for Lévy processes with arbitrary Lévy measures. Theor. Probab. AD Math. Stat. No. 72, 75–92 (2006)

    Google Scholar 

  33. Kulik, A. M.: Stochastic calculus of variations for general Lévy processes and its applications to jump-type SDE’s with non-degenerated drift. ar**v:math/0606427 (2007)

  34. Kunita, H.: Stochastic differential equations based on Lévy processes and stochastic flows of diffeomorphisms. In: Rao, M.M. (ed.) Real and Stochastic Analysis, pp. 305–373. Birkhaäuser, Boston (2004)

    Google Scholar 

  35. Kunita, H.: Stochastic Flows and Jump-diffusions. Springer, New York (2019)

    Google Scholar 

  36. Lapeyre, B., Pardoux, È., Sentis, R.: Mèthodes de Monte-Carlo Pour les Èquations de Transport et de Diffusion. Springer-Verlag, Berlin (1998)

    Google Scholar 

  37. Mariucci, E., Reiß, M.: Wasserstein and total variation distance between marginals of Lévy processes. Electron. J. Stat. 12, 2482–2514 (2018)

    MathSciNet  Google Scholar 

  38. Méléard, S.: Asymptotic behaviour of some interacting particle systems; McKean-Vlasov and Boltzmann models. In: Probabilistic Models for Nonlinear Partial Differential Equations, vol. 1627, pp. 42–95. Springer, Berlin (1996)

    Google Scholar 

  39. Mezerdi, M.A.: On the convergence of carathéodory numerical scheme for Mckean-Vlasov equations. Stoch. Anal. Appl. 39(5), 804–818 (2021)

    MathSciNet  Google Scholar 

  40. Naito, R., Yamada, T.: A higher order weak approximation of McKean-Vlasov type SDEs. BIT Comput. Sci. Numer. Math. (2022)

  41. Nualart, D.: The Malliavin Calculus and Related Topics. Springer, New York (2006)

    Google Scholar 

  42. Protter, P., Talay, D.: The Euler scheme for Lévy driven stochastic differential equations. Ann. Probab. 25(1), 393–423 (1997)

    MathSciNet  Google Scholar 

  43. Sato, K.: Lévy Processes and Infinitely Divisible Distributions. Cambridge University Press, Cambridge (1999)

    Google Scholar 

  44. Sato, K.: Basic results on Lévy processes. In: Barndorff-Nielsen, O.E., Mikosch, T., Resnick, S.I. (eds.) Lévy Processes, pp. 3–37. Basel, Theory and applications. Birkhäuser (2001)

    Google Scholar 

  45. Song, Y., Zhang,X.: Regularity of density for SDEs driven by degenerate Lévy noises. ar**v:1401.4624 (2014)

  46. Sznitman, A.-S.: Équations de type de Boltzmann, spatialement homogènes. Z. Wahrsch. Verw. Gebiete 66(4), 559–592 (1984)

    MathSciNet  Google Scholar 

  47. Szpruch, L., Tan, S., Tse, A.: Iterative multilevel particle approximation for Mckean-Vlasov SDEs. Ann. Appl. Probab. 29(4), 2230–2265 (2019)

    MathSciNet  Google Scholar 

  48. Tanaka, H.: Probabilistic treatment of the Boltzmann equation of Maxwellian molecules. Z. Wahrsch. Verw. Gebiete. 46(1), 67–105 (1978)

    MathSciNet  Google Scholar 

  49. Tanaka, H.: Stochastic differential equation corresponding to the spatially homogeneous Boltzmann equation of Maxwellian and noncutoff type. J. Fac. Sci. Univ. Tokyo Sect. IA Math 34(2), 351–369 (1987)

    MathSciNet  Google Scholar 

  50. Villani, C.: On a new class of weak solutions to the spatially homogeneous Boltzmann and Landau equations. Arch. Rational Mech. Anal. 143(3), 273–307 (1998)

    MathSciNet  Google Scholar 

  51. Villani, C.: Optimal Transport. Springer, New York (2009)

    Google Scholar 

  52. Zhang, X.: Densities for SDEs driven by degenerate \(\alpha \)-stable processes. Ann. Probab. 42(5), 1885–1910 (2014)

    MathSciNet  Google Scholar 

Download references

Acknowledgements

The author thanks Prof. Vlad Bally for his kind and patient guidance and many useful suggestions.

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to Yifeng Qin.

Ethics declarations

Conflict of interest statement.

The author certifies that he has NO affiliations with or involvement in any organization or entity with any financial interest or non-financial interest in the subject matter or materials discussed in this manuscript.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Appendix

Appendix

In Appendix, we give the proof of Lemma 3.9.

Proof

Proof of (i) We notice by the definitions (93) and Eqs. (52), (51) that for any partition \({\mathcal {P}}=\{0=r_0<r_1<\cdots<r_{n-1}<r_n=T\}\), \(M\in {\mathbb {N}}\), any \(k_0,{i_0}\in {\mathbb {N}},{j}\in \{1,\ldots ,d\}\),

$$\begin{aligned} D^Z_{ \big (k_0,{i_0},{j} \big )}{x}^{{\mathcal {P}},M}_{t}= & {} \int _{0}^{t}\nabla _xb \left( \tau (r),{x}^{{\mathcal {P}},M}_{\tau (r)},\rho ^{{\mathcal {P}},M}_{\tau (r)} \right) D^Z_{ \big (k_0,{i_0},{j} \big )}{x}^{{\mathcal {P}},M}_{\tau (r)}\textrm{d}r \nonumber \\{} & {} + \, \mathbb {1}_{\{0<T_{{i_0}}^{k_0}\le t\}}\mathbb {1}_{\{1\le k_0\le M\}}\xi _{{i_0}}^{k_0}\partial _{z_{{j}}}{c}\nonumber \\ {}{} & {} \left( \tau \left( {T}_{{i_0}}^{k_0} \right) ,\eta ^1_{T_{i_0}^{k_0}} \big (W_{i_0}^{k_0} \big ),{Z}_{{i_0}}^{k_0},{x}^{{{\mathcal {P}}},M}_{\tau ^{{{\mathcal {P}}}} \big ({T}_{{i_0}}^{k_0} \big )-},\rho ^{{\mathcal {P}},M}_{\tau ^{{{\mathcal {P}}}} \big ({T}_{{i_0}}^{k_0} \big )-} \right) \nonumber \\{} & {} + \, \sum _{k=1}^M\sum _{i=1}^{J^k_t}\nabla _x{c} \left( \tau \big ({T}_{i}^{k} \big ),\eta ^1_{T_{i}^{k}} \big (W_{i}^{k} \big ),{Z}_{i}^{k},{x}^{{{\mathcal {P}}},M}_{\tau ^{{{\mathcal {P}}}} \big ({T}_{i}^{k} \big )-},\rho ^{{\mathcal {P}},M}_{\tau ^{{{\mathcal {P}}}} \big ({T}_{i}^{k} \big )-} \right) \nonumber \\ {}{} & {} D^Z_{ \big (k_0,{i_0},{j} \big )}{x}^{{\mathcal {P}},M}_{\tau \big ({T}_{i}^{k} \big )-}, \end{aligned}$$
(140)
$$\begin{aligned} D^\Delta _j{x}_{t}^{{\mathcal {P}},M}= & {} a^M_{T}{\varvec{e_j}}+\int _{0}^{t}\nabla _xb \left( \tau (r),{x}^{{\mathcal {P}},M}_{\tau (r)},\rho ^{{\mathcal {P}},M}_{\tau (r)} \right) D^\Delta _{j}{x}^{{\mathcal {P}},M}_{\tau (r)}\textrm{d}r \nonumber \\{} & {} + \, \sum _{k=1}^M\sum _{i=1}^{J^k_t}\nabla _x{c} \left( \tau \big ({T}_{i}^{k} \big ),\eta ^1_{T_{i}^{k}} \big (W_{i}^k \big ),{Z}_{i}^{k},{x}^{{{\mathcal {P}}},M}_{\tau ^{{{\mathcal {P}}}} \big ({T}_{i}^{k} \big )-},\rho ^{{\mathcal {P}},M}_{\tau ^{{{\mathcal {P}}}} \big ({T}_{i}^{k} \big )-} \right) \nonumber \\ {}{} & {} D^\Delta _j{x}^{{\mathcal {P}},M}_{\tau \big ({T}_{i}^{k} \big )-}, \quad \quad \end{aligned}$$
(141)

where \({\varvec{e_j}}=(0,\ldots ,0,1,0,\ldots ,0)\) with value 1 at the \(j-\)th component. And

$$\begin{aligned} D^Z_{ \big (k_0,{i_0},{j} \big )}{x}^{M}_{t}= & {} \int _{0}^{t}\nabla _xb \big (r,{x}^{M}_{r},\rho _{r} \big )D^Z_{ \big (k_0,{i_0},{j} \big )}{x}^{M}_{r}\textrm{d}r \nonumber \\{} & {} + \, \mathbb {1}_{ \big \{0<T_{{i_0}}^{k_0}\le t \big \}}\mathbb {1}_{ \big \{1\le k_0\le M \big \}}\xi _{{i_0}}^{k_0}\partial _{z_{{j}}}{c}\nonumber \\ {}{} & {} \left( {T}_{{i_0}}^{k_0},\eta ^2_{T_{i_0}^{k_0}} \big (W_{i_0}^{k_0} \big ),{Z}_{{i_0}}^{k_0},{x}^{M}_{{T}_{{i_0}}^{k_0}-},\rho _{{T}_{{i_0}}^{k_0}-} \right) \nonumber \\{} & {} + \, \sum _{k=1}^M\sum _{i=1}^{J^k_t}\nabla _x{c} \left( {T}_{i}^{k},\eta ^2_{T_{i}^{k}} \big (W_{i}^k \big ),{Z}_{i}^{k},{x}^{M}_{{T}_{i}^{k}-},\rho _{{T}_{i}^{k}-} \right) D^Z_{ \big (k_0,{i_0},{j} \big )}{x}^{M}_{{T}_{i}^{k}-}, \nonumber \\ \end{aligned}$$
(142)
$$\begin{aligned} D^\Delta _j{x}_{t}^{M}= & {} a^M_{T}{\varvec{e_j}}+\int _{0}^{t}\nabla _xb \big (r,{x}^{M}_{r},\rho _{r} \big )D^\Delta _{j}{x}^{M}_{r}\textrm{d}r\nonumber \\ {}{} & {} +\sum _{k=1}^M\sum _{i=1}^{J_t^k}\nabla _x{c} \left( {T}_{i}^{k},\eta ^2_{T_{i}^{k}} \big (W_{i}^k \big ),{Z}_{i}^{k},{x}^{M}_{{T}_{i}^{k}-},\rho _{{T}_{i}^{k}-} \right) D^\Delta _j{x}^{M}_{{T}_{i}^{k}-}. \end{aligned}$$
(143)

For \(u\in l_2\), we will use the notation \(\vert u\vert _{l_2}^2=\vert u_{({\bullet },{\circ },{\diamond })}\vert _{l_2}^2=\sum \nolimits _{k=1}^\infty \sum \nolimits _{i=1}^\infty \sum \nolimits _{j=1}^d\vert u_{(k,i,j)}\vert ^2\). We write \({\mathbb {E}}\vert D^Z_{({\bullet },{\circ },{\diamond })}{x}^{{\mathcal {P}},M}_{t}-D^Z_{({\bullet },{\circ },{\diamond })}{x}^{M}_{t}\vert _{l_2}^2\le C[H_1+H_2+H_3]\), with

$$\begin{aligned} H_1= & {} {\mathbb {E}} \Bigg \vert \int _{0}^{t}\nabla _xb \left( \tau (r),{x}^{{\mathcal {P}},M}_{\tau (r)},\rho ^{{\mathcal {P}},M}_{\tau (r)} \right) D^Z_{({\bullet },{\circ },{\diamond })}{x}^{{\mathcal {P}},M}_{\tau (r)}\textrm{d}r\\ {}{} & {} -\int _{0}^{t}\nabla _xb \left( r,{x}^{M}_{r},\rho _{r} \right) D^Z_{({\bullet },{\circ },{\diamond })}{x}^{M}_{r}\textrm{d}r \Bigg \vert _{l_2}^2, \\ H_2= & {} {\mathbb {E}} \left| \mathbb {1}_{\{0<T_{{\circ }}^{\bullet }\le t\}}\mathbb {1}_{\{1\le {\bullet }\le M\}} \left( \partial _{z_{\diamond }}{c} \big (\tau \big ({T}_{\circ }^{\bullet } \big ),\eta ^1_{T_{\circ }^{\bullet }} \big (W_{\circ }^{\bullet } \big ),{Z}_{\circ }^{\bullet },{x}^{{{\mathcal {P}}},M}_{\tau ^{{{\mathcal {P}}}} \big ({T}_{\circ }^{\bullet } \big )-},\rho ^{{\mathcal {P}},M}_{\tau ^{{{\mathcal {P}}}} \big ({T}_{\circ }^{\bullet } \big )-} \right) \right. \\{} & {} - \, \left. \partial _{z_{\diamond }}{c} \left( {T}_{\circ }^{\bullet },\eta ^2_{T_{\circ }^{\bullet }} \big (W_{\circ }^{\bullet } \big ),{Z}_{\circ }^{\bullet },{x}^{M}_{{T}_{\circ }^{\bullet }-},\rho _{{T}_{\circ }^{\bullet }-} \big ) \right) \right| _{l_2}^2, \\ H_3= & {} {\mathbb {E}} \left| \sum _{k=1}^M \sum _{i=1}^{J^k_t}\nabla _x{c} \left( \tau ({T}_{i}^{k} \big ),\eta ^1_{T_{i}^{k}}\big (W_{i}^k \big ),{Z}_{i}^{k},{x}^{{{\mathcal {P}}},M}_{\tau ^{{{\mathcal {P}}}} \big ({T}_{i}^{k} \big )-},\rho ^{{\mathcal {P}},M}_{\tau ^{{{\mathcal {P}}}} \big ({T}_{i}^{k} \big )-} \right) D^Z_{ \big ({\bullet },{\circ },{\diamond } \big )}{x}^{{\mathcal {P}},M}_{\tau \big ({T}_{i}^{k} \big )-} \right. \\{} & {} -\sum _{k=1}^M\sum _{i=1}^{J^k_t}\nabla _x{c} \left( {T}_{i}^{k},\eta ^2_{T_{i}^{k}} \big (W_{i}^k \big ),{Z}_{i}^{k},{x}^{M}_{{T}_{i}^{k}-},\rho _{{T}_{i}^{k}-} \right) D^Z_{\big ({\bullet },{\circ },{\diamond } \big )}{x}^{M}_{{T}_{i}^{k}-}\vert _{l_2}^2. \end{aligned}$$

We take a small \(\varepsilon _*>0\). We recall \(\varepsilon _M\) in (27). Firstly, using Hypothesis 2.1, we get

$$\begin{aligned} H_1\le & {} C \left[ {\mathbb {E}} \int _{0}^{t} \left| \nabla _xb \left( \tau (r),{x}^{{\mathcal {P}},M}_{\tau (r)},\rho ^{{\mathcal {P}},M}_{\tau (r)})-\nabla _xb(r,{x}^{M}_{r},\rho _{r} \right) \right| ^2 \left| D^Z_{ \left( {\bullet },{\circ },{\diamond } \right) }{x}^{M}_{r} \right| _{l_2}^2\textrm{d}r \right. \nonumber \\{} & {} \left. + \, {\mathbb {E}} \int _{0}^{t}\vert \nabla _xb \left( \tau (r),{x}^{{\mathcal {P}},M}_{\tau (r)},\rho ^{{\mathcal {P}},M}_{\tau (r)} \right) \vert ^2\vert D^Z_{ \left( {\bullet },{\circ },{\diamond } \right) }{x}^{{\mathcal {P}},M}_{\tau (r)}-D^Z_{ \left( {\bullet },{\circ },{\diamond } \right) }{x}^{M}_{r}\vert _{l_2}^2\textrm{d}r \right] \nonumber \\\le & {} C \left[ {\mathbb {E}}\int _0^t \left[ \big \vert {\mathcal {P}} \big \vert ^2+ \big \vert {x}^{{\mathcal {P}},M}_{\tau (r)}-{x}^{M}_{r} \big \vert ^{2}+ \left( W_1 \left( {\rho }^{{\mathcal {P}},M}_{\tau (r)},{\rho }_{r} \right) \right) ^2 \right] \left| D^Z_{ \left( {\bullet },{\circ },{\diamond } \right) }{x}^{M}_{r} \right| _{l_2}^2\textrm{d}r \right. \nonumber \\{} & {} + \, \left. \int _{0}^{t}{\mathbb {E}} \left| D^Z_{({\bullet },{\circ },{\diamond })}{x}^{{\mathcal {P}},M}_{\tau (r)}-D^Z_{({\bullet },{\circ },{\diamond })}{x}^{M}_{r} \right| _{l_2}^2\textrm{d}r \right] . \end{aligned}$$

Then by Lemma 3.7, using Hölder inequality with conjugates \(1+\frac{\varepsilon _*}{2}\) and \(\frac{2+\varepsilon _*}{\varepsilon _*}\), by Lemma 2.6 and (50), we have

$$\begin{aligned} H_1\le & {} C \left[ \vert {\mathcal {P}}\vert ^2+\int _0^t \left( {\mathbb {E}}\vert {x}^{{\mathcal {P}},M}_{\tau (r)}-{x}^{M}_{r}\vert ^{2+\varepsilon _*} \right) ^{\frac{2}{2+\varepsilon _*}}\textrm{d}r+\int _0^t \left( W_{2+\varepsilon _*} \left( {\rho }^{{\mathcal {P}},M}_{\tau (r)},{\rho }_{r} \right) \right) ^2\textrm{d}r \right. \nonumber \\{} & {} + \left. \int _{0}^{t}{\mathbb {E}} \left| D^Z_{ ({\bullet },{\circ },{\diamond })}{x}^{{\mathcal {P}},M}_{\tau (r)}-D^Z_{({\bullet },{\circ },{\diamond })}{x}^{M}_{r} \right| _{l_2}^2\textrm{d}r \right] \nonumber \\\le & {} C \left[ \left( \vert {\mathcal {P}}\vert +\varepsilon _M \right) ^{\frac{2}{2+\varepsilon _*}}+\int _{0}^{t}{\mathbb {E}}\vert D^Z_{ ({\bullet },{\circ },{\diamond })}{x}^{{\mathcal {P}},M}_{\tau (r)}-D^Z_{({\bullet },{\circ },{\diamond })}{x}^{M}_{r}\vert _{l_2}^2\textrm{d}r \right] . \end{aligned}$$
(144)

Secondly, using Hypothesis 2.1 and the isometry of the Poisson point measure \({\mathcal {N}}\), we get

$$\begin{aligned} H_2= & {} {\mathbb {E}}\sum _{k=1}^M\sum _{i=1}^{J^k_t}\sum _{j=1}^d \left| \partial _{z_{j}}{c} \left( \tau \big ({T}_{i}^{k} \big ),\eta ^1_{T_{i}^{k}} \big (W_{i}^k \big ),{Z}_{i}^{k},{x}^{{{\mathcal {P}}},M}_{\tau ^{{{\mathcal {P}}}} \big ({T}_{i}^{k} \big )-},\rho ^{{\mathcal {P}},M}_{\tau ^{{{\mathcal {P}}}} \big ({T}_{i}^{k} \big )-} \right) \right. \\{} & {} - \, \left. \partial _{z_{j}}{c} \left( {T}_{i}^{k},\eta ^2_{T_{i}^{k}} \big (W_{i}^k \big ),{Z}_{i}^{k},{x}^{M}_{{T}_{i}^{k}-},\rho _{{T}_{i}^{k}-} \right) \right| ^2\nonumber \\\le & {} C{\mathbb {E}}\sum _{k=1}^M\sum _{i=1}^{J^k_t}\vert {\bar{c}}(Z^k_i)\vert ^2 \left[ \left| \tau ^{{{\mathcal {P}}}} ({T}_{i}^{k} \big )-T^k_i \right| + \left| \eta ^1_{T_{i}^{k}} \big (W_{i}^k \big )-\eta ^2_{T_{i}^{k}} \big (W_{i}^k \big ) \right| \right. \nonumber \\{} & {} + \, \left. \left| {x}^{{{\mathcal {P}}},M}_{\tau ^{{{\mathcal {P}}}}\big ({T}_{i}^{k} \big )-}-{x}^{M}_{{T}_{i}^{k}-} \right| +W_1 \left( \rho ^{{\mathcal {P}},M}_{\tau ^{{{\mathcal {P}}}} \big ({T}_{i}^{k} \big )-},\rho _{{T}_{i}^{k}-} \right) \right] ^2\nonumber \\\le & {} C{\mathbb {E}}\int _0^t\int _{[0,1]\times {\mathbb {R}}^d}\vert {\bar{c}}(z)\vert ^2 \left[ \vert \tau ^{{{\mathcal {P}}}}(r)-r\vert ^2+\vert \eta ^1_{r}(w)-\eta ^2_{r}(w)\vert ^2 \right. \nonumber \\{} & {} + \left. \vert {x}^{{{\mathcal {P}}},M}_{\tau ^{{{\mathcal {P}}}}(r)-}-{x}^{M}_{r-}\vert ^2+ \left( W_1 \left( \rho ^{{\mathcal {P}},M}_{\tau ^{{{\mathcal {P}}}}(r)-},\rho _{r-} \right) \right) ^2 \right] {\mathcal {N}} (\textrm{d}w,\textrm{d}z,\textrm{d}r)\nonumber \\= & {} C{\mathbb {E}}\int _0^t\int _{[0,1]\times {\mathbb {R}}^d}\vert {\bar{c}}(z)\vert ^2 \left[ \vert \tau ^{{{\mathcal {P}}}}(r)-r\vert ^2+\vert \eta ^1_{r}(w)-\eta ^2_{r}(w)\vert ^2 \right. \nonumber \\{} & {} + \left. \vert {x}^{{{\mathcal {P}}},M}_{\tau ^{{{\mathcal {P}}}}(r)-}-{x}^{M}_{r-}\vert ^2+ \left( W_1 \left( \rho ^{{\mathcal {P}},M}_{\tau ^{{{\mathcal {P}}}}(r)-},\rho _{r-} \right) \right) ^2 \right] \textrm{d}w\mu (\textrm{d}z)\textrm{d}r. \end{aligned}$$

Then by (39), (43), Lemma 2.6, (50), and Hölder inequality with conjugates \(1+\frac{\varepsilon _*}{2}\) and \(\frac{2+\varepsilon _*}{\varepsilon _*}\), we have

$$\begin{aligned} H_2\le & {} C \left[ \vert {\mathcal {P}}\vert ^2+\int _0^t \left( {\mathbb {E}} \left| {x}^{{\mathcal {P}},M}_{\tau (r)}-{x}_{r} \right| ^{2+\varepsilon _*} \right) ^{\frac{2}{2+\varepsilon _*}}\textrm{d}r \right. \nonumber \\{} & {} \left. +\int _0^t \left( {\mathbb {E}} \left| {x}^{{\mathcal {P}},M}_{\tau (r)}-{x}^{M}_{r} \right| ^{2+\varepsilon _*} \right) ^{\frac{2}{2+\varepsilon _*}}\textrm{d}r+\int _0^t \left( W_{2+\varepsilon _*} \left( \rho ^{{\mathcal {P}},M}_{\tau ^{{{\mathcal {P}}}}(r)},\rho _{r} \right) \right) ^2\textrm{d}r \right] \nonumber \\\le & {} C(\vert {\mathcal {P}}\vert +\varepsilon _M)^{\frac{2}{2+\varepsilon _*}}. \end{aligned}$$
(145)

Thirdly, we write \(H_3\le C[H_{3,1}+H_{3,2}]\), where

$$\begin{aligned} H_{3,1}= & {} {\mathbb {E}} \left( \sum _{k=1}^M\sum _{i=1}^{J_t^k}\vert \nabla _x{c} \left( \tau \big ({T}_{i}^{k} \big ),\eta ^1_{T_{i}^{k}} \big (W_{i}^k \big ),{Z}_{i}^{k}, {x}^{{{\mathcal {P}}},M}_{\tau ^{{{\mathcal {P}}}} \big ({T}_{i}^{k} \big )-},\rho ^{{\mathcal {P}},M}_{\tau ^{{{\mathcal {P}}}} \big ({T}_{i}^{k} \big )-} \right) \right. \\{} & {} - \left. \left. \nabla _x{c} \left( {T}_{i}^{k},\eta ^2_{T_{i}^{k}} \big (W_{i}^k \big ),{Z}_{i}^{k},{x}^{M}_{{T}_{i}^{k}-}, \rho _{{T}_{i}^{k}-} \right) \right| \left| D^Z_{ \left( {\bullet },{\circ },{\diamond } \right) }{x}^{M}_{{T}_{i}^{k}-} \right| _{l_2} \right) ^2, \\ H_{3,2}= & {} {\mathbb {E}} \left( \sum _{k=1}^M\sum _{i=1}^{J^k_t} \left| \nabla _x{c} \left( \tau \big ({T}_{i}^{k} \big ),\eta ^1_{T_{i}^{k}} \big (W_{i}^k \big ),{Z}_{i}^{k},{x}^{{{\mathcal {P}}},M}_{\tau ^{{{\mathcal {P}}}} \big ({T}_{i}^{k} \big )-},\rho ^{{\mathcal {P}},M}_{\tau ^{{{\mathcal {P}}}} \big ({T}_{i}^{k} \big )-} \right) \right| \right. \\{} & {} \left. \left| D^Z_{ \big ({\bullet },{\circ },{\diamond } \big )}{x}^{{\mathcal {P}},M}_{\tau \big ({T}_{i}^{k} \big )-}-D^Z_{({\bullet },{\circ },{\diamond })}{x}^{M}_{{T}_{i}^{k}-} \right| _{l_2} \right) ^2. \end{aligned}$$

Using Hypothesis 2.1 and (45) with

$$\begin{aligned}{\bar{\Phi }}(r,w,z,\omega ,\rho )= & {} \left| \nabla _x{c} \left( \tau (r),\eta ^1_{r}(w),z,{x}^{{{\mathcal {P}}},M}_{\tau ^{{{\mathcal {P}}}}(r)-},\rho ^{{\mathcal {P}},M}_{\tau ^{{{\mathcal {P}}}}(r)-} \right) \right. \\{} & {} - \, \left. \nabla _x{c} \left( r,\eta ^2_{r}(w),z,{x}^{M}_{r-},\rho _{r-} \right) \right| \left| D^Z_{({\bullet },{\circ },{\diamond })}{x}^{M}_{r-} \right| _{l_2},\end{aligned}$$

we get

$$\begin{aligned} H_{3,1}\le & {} {\mathbb {E}} \left( \int _0^t\int _{[0,1]\times {\mathbb {R}}^d} \left| \nabla _x{c} \left( \tau (r),\eta ^1_{r}(w),z,{x}^{{{\mathcal {P}}},M}_{\tau ^{{{\mathcal {P}}}}(r)-},\rho ^{{\mathcal {P}},M}_{\tau ^{{{\mathcal {P}}}}(r)-} \right) \right. \right. \\{} & {} - \, \nabla _x{c} \big (r,\eta ^2_{r}(w),z,{x}^{M}_{r-},\rho _{r-} \big ) \big \vert \big \vert D^Z_{({\bullet },{\circ },{\diamond })}{x}^{M}_{r-}\vert _{l_2}{\mathcal {N}}(\textrm{d}w,\textrm{d}z,\textrm{d}r) \big )^2\\\le & {} C{\mathbb {E}}\int _0^t\int _{[0,1]} \big [ \big \vert \tau ^{{{\mathcal {P}}}}(r)-r \big \vert ^2+\big \vert \eta ^1_{r}(w)-\eta ^2_{r}(w) \big \vert ^2\\{} & {} +\, \big \vert {x}^{{{\mathcal {P}}},M}_{\tau ^{{{\mathcal {P}}}}(r)-}-{x}^{M}_{r-} \big \vert ^2+ \big (W_1 \big (\rho ^{{\mathcal {P}},M}_{\tau ^{{{\mathcal {P}}}}(r)-},\rho _{r-} \big ) \big )^2 \big ] \big \vert D^Z_{({\bullet },{\circ },{\diamond })}{x}^{M}_{r-}\vert _{l_2}^2\textrm{d}w\textrm{d}r. \end{aligned}$$

Then using (39), (43), Lemma 3.7, and Hölder inequality with conjugates \(1+\frac{\varepsilon _*}{2}\) and \(\frac{2+\varepsilon _*}{\varepsilon _*}\), we have

$$\begin{aligned} H_{3,1}\le & {} C\Big [\vert {\mathcal {P}}\vert ^2+\int _0^t \left( {\mathbb {E}}\vert {x}^{{\mathcal {P}},M}_{\tau (r)-}-{x}_{r-}\vert ^{2+\varepsilon _*} \right) ^{\frac{2}{2+\varepsilon _*}}\textrm{d}r\\{} & {} \left. +\int _0^t \left( {\mathbb {E}}\vert {x}^{{\mathcal {P}},M}_{\tau (r)-}-{x}^{M}_{r-}\vert ^{2+\varepsilon _*} \right) ^{\frac{2}{2+\varepsilon _*}}\textrm{d}r+\int _0^t \left( W_{2+\varepsilon _*} \left( \rho ^{{\mathcal {P}},M}_{\tau ^{{{\mathcal {P}}}}(r)-},\rho _{r-} \right) \right) ^2\textrm{d}r \right] \\\le & {} C \left( \vert {\mathcal {P}}\vert +\varepsilon _M \right) ^{\frac{2}{2+\varepsilon _*}}, \end{aligned}$$

where the last inequality is a consequence of Lemma 2.6 and (50).

Using Hypothesis 2.1, (45) with

$$\begin{aligned}{} & {} {\bar{\Phi }}(r,w,z,\omega ,\rho )= \left| \nabla _x{c} \left( \tau (r),\eta ^1_{r}(w),z, {x}^{{{\mathcal {P}}},M}_{\tau ^{{{\mathcal {P}}}}(r)-},\rho ^{{\mathcal {P}},M}_{\tau ^{{{\mathcal {P}}}}(r)-} \right) \right| \\ {}{} & {} \left| D^Z_{({\bullet },{\circ },{\diamond })}{x}^{{\mathcal {P}},M}_{\tau (r)-}-D^Z_{({\bullet },{\circ },{\diamond })}{x}^{M}_{r-} \right| _{l_2}, \end{aligned}$$

we have

$$\begin{aligned} H_{3,2}\le & {} {\mathbb {E}} \left( \int _0^t\int _{[0,1]\times {\mathbb {R}}^d} \left| \nabla _x{c} \left( \tau (r),\eta ^1_{r}(w),z,{x}^{{{\mathcal {P}}},M}_{\tau ^{{{\mathcal {P}}}}(r)-}, \rho ^{{\mathcal {P}},M}_{\tau ^{{{\mathcal {P}}}}(r)-} \right) \right| \left| D^Z_{({\bullet },{\circ },{\diamond })}{x}^{{\mathcal {P}},M}_{\tau (r)-} \right. \right. \\{} & {} \left. -D^Z_{({\bullet },{\circ },{\diamond })}{x}^{M}_{r-} \right| _{l_2}{\mathcal {N}}(\textrm{d}w,\textrm{d}z,\textrm{d}r))^2\\\le & {} C\int _{0}^{t}{\mathbb {E}} \left| D^Z_{({\bullet },{\circ },{\diamond })}{x}^{{\mathcal {P}},M}_{\tau (r)-}-D^Z_{({\bullet },{\circ },{\diamond })}{x}^{M}_{r-} \right| _{l_2}^2\textrm{d}r. \end{aligned}$$

Therefore,

$$\begin{aligned} H_3\le & {} C[H_{3,1}+H_{3,2}]\nonumber \\ {}{} & {} \le C \left[ (\vert {\mathcal {P}}\vert +\varepsilon _M)^{\frac{2}{2+\varepsilon _*}}+\int _{0}^{t}{\mathbb {E}}\vert D^Z_{({\bullet },{\circ },{\diamond })}{x}^{{\mathcal {P}},M}_{\tau (r)}-D^Z_{({\bullet },{\circ },{\diamond })} {x}^{M}_{r}\vert _{l_2}^2\textrm{d}r \right] .\nonumber \\ \end{aligned}$$
(146)

Combining (144), (145) and (146),

$$\begin{aligned}{} & {} {\mathbb {E}} \left| D^Z_{({\bullet },{\circ },{\diamond })}{x}^{{\mathcal {P}},M}_{t}-D^Z_{({\bullet },{\circ },{\diamond })}{x}^{M}_{t} \right| _{l_2}^2\le C \left[ (\vert {\mathcal {P}}\vert +\varepsilon _M)^{\frac{2}{2+\varepsilon _*}} \right. \\{} & {} \quad \left. +\int _{0}^{t}{\mathbb {E}} \left| D^Z_{({\bullet },{\circ },{\diamond })}{x}^{{\mathcal {P}},M}_{\tau (r)}-D^Z_{({\bullet },{\circ },{\diamond })}{x}^{M}_{r} \right| _{l_2}^2\textrm{d}r \right] . \end{aligned}$$

In a similar way, we notice, by (116), the isometry of the Poisson point measure N, and (45) with

$$\begin{aligned}{\bar{\Phi }}(r,w,z,\omega ,\rho )=\nabla _x{c} \left( \tau (r),\eta ^1_{r}(w),z,{x}^{{{\mathcal {P}}},M}_{\tau ^{{{\mathcal {P}}}}(r)-},\rho ^{{\mathcal {P}},M}_{\tau ^{{{\mathcal {P}}}}(r)-} \right) D^Z_{(k_0,{i_0},{j})}{x}^{{\mathcal {P}},M}_{\tau (r)-}\end{aligned}$$

that

$$\begin{aligned} {\mathbb {E}} \left| D^Z_{({\bullet },{\circ },{\diamond })}{x}^{{\mathcal {P}},M}_{\tau (t)}-D^Z_{({\bullet },{\circ },{\diamond })}{x}^{{\mathcal {P}},M}_{t} \right| _{l_2}^2\le C\vert {\mathcal {P}}\vert ,\end{aligned}$$
(147)

so

$$\begin{aligned}{} & {} {\mathbb {E}} \left| D^Z_{({\bullet },{\circ },{\diamond })}{x}^{{\mathcal {P}},M}_{t}-D^Z_{({\bullet },{\circ },{\diamond })}{x}^{M}_{t} \right| _{l_2}^2\le C \left[ (\vert {\mathcal {P}}\vert +\varepsilon _M)^{\frac{2}{2+\varepsilon _*}} \right. \\{} & {} \quad \left. +\int _{0}^{t}{\mathbb {E}} \left| D^Z_{({\bullet },{\circ },{\diamond })}{x}^{{\mathcal {P}},M}_{r}-D^Z_{({\bullet },{\circ },{\diamond })}{x}^{M}_{r} \right| _{l_2}^2\textrm{d}r \right] . \end{aligned}$$

We conclude by Gronwall lemma that \({\mathbb {E}}\vert D^Z_{({\bullet },{\circ },{\diamond })}{x}^{{\mathcal {P}},M}_{t}-D^Z_{({\bullet },{\circ },{\diamond })}{x}^{M}_{t}\vert _{l_2}^2\le C(\vert {\mathcal {P}}\vert +\varepsilon _M)^{\frac{2}{2+\varepsilon _*}}\). Finally, by a similar argument, \({\mathbb {E}}\vert D^{\Delta }_{({\diamond })}{x}^{{\mathcal {P}},M}_{t}-D^{\Delta }_{({\diamond })}{x}^{M}_{t}\vert _{{\mathbb {R}}^d}^2\le C(\vert {\mathcal {P}}\vert +\varepsilon _M)^{\frac{2}{2+\varepsilon _*}}\), and we obtain what we need.

Proof of (ii) We only need to prove that for any \(M_1,M_2\in {\mathbb {N}}\) with \(\varepsilon _{M_1\wedge M_2}\le 1\) and \(\vert {\bar{c}}(z)\vert ^2\mathbb {1}_{\{\vert z\vert >M_1\wedge M_2\}}\le 1\), we have

$$\begin{aligned} \left\| Dx^{M_1}_t-Dx^{M_2}_t \right\| _{L^2(\Omega ;l^2\times {\mathbb {R}}^d)}\le (\varepsilon _{M_1}+\varepsilon _{M_2})^{\frac{1}{2+\varepsilon _*}}. \end{aligned}$$
(148)

In fact, if \((Dx^M_t)_{M\in {\mathbb {N}}}\) is a Cauchy sequence in \(L^2(\Omega ;l^2\times {\mathbb {R}}^d)\), then it has a limit Y in \(L^2(\Omega ;l^2\times {\mathbb {R}}^d)\). But when we apply Lemma 3.3(A) with \(F_M=X^M_t\) and \(F=X_t\), we know that there exists a convex combination \(\sum \limits _{M^\prime =M}^{m_{M}}\gamma _{M^\prime }^{M}\times F^{M^\prime }_t,\) with \(\gamma _{M^\prime }^{M}\ge 0,M^\prime =M, \ldots ,m_{M}\) and \( \sum \limits _{M^\prime =M}^{m_{M}}\gamma _{M^\prime }^{M}=1\), such that

$$\begin{aligned} \left\| \sum _{M^\prime =M}^{m_{M}}\gamma _{M^\prime }^{M}\times Dx^{M^\prime }_t-Dx_t\right\| _{L^2(\Omega ;l^2\times {\mathbb {R}}^d)}\rightarrow 0, \end{aligned}$$

as \(M\rightarrow \infty \). Meanwhile, we have

$$\begin{aligned}\left\| \sum _{M^\prime =M}^{m_{M}}\gamma _{M^\prime }^{M}\times DF^{M^\prime }_t-Y\right\| _{L^2(\Omega ;l^2\times {\mathbb {R}}^d)}\le \sum _{M^\prime =M}^{m_{M}}\gamma _{M^\prime }^{M}\left\| Dx^{M^\prime }_t-Y\right\| _{L^2(\Omega ;l^2\times {\mathbb {R}}^d)}\rightarrow 0.\end{aligned}$$

So \(Y=Dx_t\) and we conclude by passing to the limit \(M_2\rightarrow \infty \) in (148).

Now we prove (148). We recall the equation (142) and we write \({\mathbb {E}}\vert D^Z_{({\bullet },{\circ },{\diamond })}{x}^{M_1}_{t}-D^Z_{({\bullet },{\circ },{\diamond })}{x}^{M_2}_{t}\vert _{l_2}^2\le C[O_1+O_2+O_3]\), with

$$\begin{aligned} O_1= & {} {\mathbb {E}}\vert \int _{0}^{t}\nabla _xb \left( r,{x}^{M_1}_{r},\rho _{r} \right) D^Z_{({\bullet },{\circ },{\diamond })}{x}^{M_1}_{r}\textrm{d}r-\int _{0}^{t}\nabla _xb(r,{x}^{M_2}_{r},\rho _{r})D^Z_{({\bullet },{\circ },{\diamond })}{x}^{M_2}_{r}\textrm{d}r \vert _{l_2}^2, \\ O_2= & {} {\mathbb {E}} \big \vert \mathbb {1}_{\{0<T_{{\circ }}^{\bullet }\le t\}} \big (\mathbb {1}_{\{1\le {\bullet }\le M_1\}}\partial _{z_{\diamond }}{c} \big ({T}_{\circ }^{\bullet },\eta ^2_{T_{\circ }^{\bullet }} \big (W_{\circ }^{\bullet } \big ),{Z}_{\circ }^{\bullet },{x}^{M_1}_{{T}_{\circ }^{\bullet }-},\rho _{{T}_{\circ }^{\bullet }-} \big )\\{} & {} - \, \mathbb {1}_{\{1\le {\bullet }\le M_2\}}\partial _{z_{\diamond }}{c} \big ({T}_{\circ }^{\bullet },\eta ^2_{T_{\circ }^{\bullet }} \big (W_{\circ }^{\bullet } \big ),{Z}_{\circ }^{\bullet },{x}^{M_2}_{{T}_{\circ }^{\bullet }-},\rho _{{T}_{\circ }^{\bullet }-} \big ) \big )\vert _{l_2}^2, \\ O_3= & {} {\mathbb {E}}\vert \sum _{k=1}^{M_1}\sum _{i=1}^{J^k_t}\nabla _x{c} \big ({T}_{i}^{k},\eta ^2_{T_{i}^{k}} \big (W_{i}^k \big ),{Z}_{i}^{k},{x}^{M_1}_{{T}_{i}^{k}-},\rho _{{T}_{i}^{k}-})D^Z_{({\bullet },{\circ },{\diamond })}{x}^{M_1}_{{T}_{i}^{k}-}\\{} & {} -\sum _{k=1}^{M_2}\sum _{i=1}^{J^k_t}\nabla _x{c} \big ({T}_{i}^{k},\eta ^2_{T_{i}^{k}} \big (W_{i}^k \big ),{Z}_{i}^{k},{x}^{M_2}_{{T}_{i}^{k}-},\rho _{{T}_{i}^{k}-} \big )D^Z_{({\bullet },{\circ },{\diamond })}{x}^{M_2}_{{T}_{i}^{k}-} \big \vert _{l_2}^2. \end{aligned}$$

Firstly, using Hypothesis 2.1, we have

$$\begin{aligned} O_1\le & {} C \Big [{\mathbb {E}} \int _{0}^{t}\vert \nabla _xb \big (r,{x}^{M_1}_{r},\rho _{r} \big )-\nabla _xb \big (r,{x}^{M_2}_{r},\rho _{r} \big ) \big \vert ^2 \big \vert D^Z_{({\bullet },{\circ },{\diamond })}{x}^{M_2}_{r} \big \vert _{l_2}^2\textrm{d}r\nonumber \\{} & {} + \, {\mathbb {E}} \int _{0}^{t} \big \vert \nabla _xb \big (r,{x}^{M_1}_{r},\rho _{r} \big ) \big \vert ^2 \big \vert D^Z_{({\bullet },{\circ },{\diamond })}{x}^{M_1}_{r}-D^Z_{({\bullet },{\circ },{\diamond })}{x}^{M_2}_{r}\vert _{l_2}^2\textrm{d}r \Big ]\nonumber \\\le & {} C \Big [{\mathbb {E}}\int _0^t \big \vert {x}^{M_1}_{r}-{x}^{M_2}_{r} \big \vert ^{2} \big \vert D^Z_{({\bullet },{\circ },{\diamond })}{x}^{M_2}_{r} \big \vert _{l_2}^2\textrm{d}r+ \int _{0}^{t}{\mathbb {E}} \big \vert D^Z_{({\bullet },{\circ },{\diamond })}{x}^{M_1}_{r}-D^Z_{({\bullet },{\circ },{\diamond })}{x}^{M_2}_{r} \big \vert _{l_2}^2\textrm{d}r \Big ]. \end{aligned}$$

Then by Lemma 3.7, Hölder inequality with conjugates \(1+\frac{\varepsilon _*}{2}\) and \(\frac{2+\varepsilon _*}{\varepsilon _*}\), and Lemma 2.6, we obtain

$$\begin{aligned} O_1\le & {} C \left[ \int _0^t \big ({\mathbb {E}} \big \vert {x}^{M_1}_{r}-{x}^{M_2}_{r} \big \vert ^{2+\varepsilon _*} \big )^{\frac{2}{2+\varepsilon _*}}\textrm{d}r+ \int _{0}^{t}{\mathbb {E}} \big \vert D^Z_{({\bullet },{\circ },{\diamond })}{x}^{M_1}_{r}-D^Z_{({\bullet },{\circ },{\diamond })}{x}^{M_2}_{r} \big \vert _{l_2}^2\textrm{d}r \right] \nonumber \\\le & {} C \left[ \big (\varepsilon _{M_1}+\varepsilon _{M_2} \big )^{\frac{2}{2+\varepsilon _*}}+\int _{0}^{t}{\mathbb {E}} \big \vert D^Z_{({\bullet },{\circ },{\diamond })}{x}^{M_1}_{r}-D^Z_{({\bullet },{\circ },{\diamond })}{x}^{M_2}_{r} \big \vert _{l_2}^2\textrm{d}r \right] . \end{aligned}$$
(149)

Secondly, using Hypothesis 2.1, the isometry of the Poisson point measure \({\mathcal {N}}\), we have

$$\begin{aligned} O_2\le & {} C \left[ {\mathbb {E}}\sum _{k=1}^{M_1}\sum _{i=1}^{J^k_t}\sum _{j=1}^d\vert \partial _{z_{j}}{c} \left( {T}_{i}^{k},\eta ^2_{T_{i}^{k}} \big (W_{i}^k \big ),{Z}_{i}^{k},{x}^{M_1}_{{T}_{i}^{k}-},\rho _{{T}_{i}^{k}-} \right) \right. \\{} & {} - \, \partial _{z_{j}}{c} \left( {T}_{i}^{k},\eta ^2_{T_{i}^{k}} \big (W_{i}^k \big ),{Z}_{i}^{k},{x}^{M_2}_{{T}_{i}^{k}-},\rho _{{T}_{i}^{k}-} \right) \Big \vert ^2\\{} & {} + \left. \, {\mathbb {E}}\sum _{k=M_1\wedge M_2}^{M_1\vee M_2}\sum _{i=1}^{J^k_t}\sum _{j=1}^d \Big \vert \partial _{z_{j}}{c} \Big ({T}_{i}^{k},\eta ^2_{T_{i}^{k}} \big (W_{i}^k \big ),{Z}_{i}^{k},{x}^{M_2}_{{T}_{i}^{k}-},\rho _{{T}_{i}^{k}-} \Big ) \Big \vert ^2 \right] \nonumber \\\le & {} C \left[ {\mathbb {E}}\sum _{k=1}^M\sum _{i=1}^{J^k_t} \left| {\bar{c}}(Z^k_i) \right| ^2 \left| {x}^{M_1}_{{T}_{i}^{k}-}-{x}^{M_2}_{{T}_{i}^{k}-} \right| ^2 \right. \\{} & {} + \left. \, {\mathbb {E}}\sum _{k=M_1\wedge M_2}^{M_1\vee M_2}\sum _{i=1}^{J^k_t}\vert {\bar{c}} (Z^k_i)\vert ^2 \right] \\\le & {} C \left[ {\mathbb {E}}\int _0^t\int _{[0,1]\times {\mathbb {R}}^d}\vert {\bar{c}}(z)\vert ^2\vert {x}^{M_1}_{r-}-{x}^{M_2}_{r-}\vert ^2{\mathcal {N}}(\textrm{d}w,\textrm{d}z,\textrm{d}r) \right. \\{} & {} + \left. \, {\mathbb {E}}\int _0^t\int _{[0,1]}\int _{\{\vert z\vert>M_1\wedge M_2\}}\vert {\bar{c}}(z)\vert ^2{\mathcal {N}}(\textrm{d}w,\textrm{d}z,\textrm{d}r) \right] \\= & {} C \left[ {\mathbb {E}}\int _0^t\int _{[0,1]\times {\mathbb {R}}^d}\vert {\bar{c}}(z)\vert ^2\vert {x}^{M_1}_{r-}-{x}^{M_2}_{r-}\vert ^2\textrm{d}w\mu (\textrm{d}z)\textrm{d}r \right. \\{} & {} + \left. \, {\mathbb {E}}\int _0^t\int _{[0,1]}\int _{\{\vert z\vert >M_1\wedge M_2\}}\vert {\bar{c}}(z)\vert ^2\textrm{d}w\mu (\textrm{d}z)\textrm{d}r \right] . \end{aligned}$$

Then by Hölder inequality with conjugates \(1+\frac{\varepsilon _*}{2}\) and \(\frac{2+\varepsilon _*}{\varepsilon _*}\), Hypothesis 2.1 and Lemma 2.6,

$$\begin{aligned} O_2\le & {} C \left[ \int _0^t \left( {\mathbb {E}}\vert {x}^{M_1}_{r}-{x}^{M_2}_{r}\vert ^{2+\varepsilon _*} \right) ^{\frac{2}{2+\varepsilon _*}}\textrm{d}r+\varepsilon _{M_1\wedge M_2} \right] \nonumber \\\le & {} C(\varepsilon _{M_1}+\varepsilon _{M_2})^{\frac{2}{2+\varepsilon _*}}. \end{aligned}$$
(150)

Thirdly, we write \(O_3\le C[O_{3,1}+O_{3,2}+O_{3,3}]\), where

$$\begin{aligned} O_{3,1}= & {} {\mathbb {E}} \left( \sum _{k=M_1\wedge M_2}^{M_1\vee M_2}\sum _{i=1}^{J^k_t}\vert \nabla _x{c}\left( {T}_{i}^{k},\eta ^2_{T_{i}^{k}}(W_{i}^k),{Z}_{i}^{k},{x}^{M_1\vee M_2}_{{T}_{i}^{k}-},\rho _{{T}_{i}^{k}-} \right) \vert \vert D^Z_{({\bullet },{\circ },{\diamond })}{x}^{M_1\vee M_2}_{{T}_{i}^{k}-}\vert _{l_2} \right) ^2, \\ O_{3,2}= & {} {\mathbb {E}} \left( \sum _{k=1}^{M_1\wedge M_2}\sum _{i=1}^{J_t^k}\vert \nabla _x{c} \left( {T}_{i}^{k},\eta ^2_{T_{i}^{k}}(W_{i}^k),{Z}_{i}^{k},{x}^{M_1}_{{T}_{i}^{k}-},\rho _{{T}_{i}^{k}-} \right) \right. \\{} & {} -\nabla _x{c} \left( {T}_{i}^{k},\eta ^2_{T_{i}^{k}}(W_{i}^k),{Z}_{i}^{k},{x}^{M_2}_{{T}_{i}^{k}-},\rho _{{T}_{i}^{k}-} \right) \vert \vert D^Z_{({\bullet },{\circ },{\diamond })}{x}^{M_2}_{{T}_{i}^{k}-}\vert _{l_2})^2, \\ O_{3,3}= & {} {\mathbb {E}} \left( \sum _{k=1}^{M_1\wedge M_2}\sum _{i=1}^{J^k_t}\vert \nabla _x{c} \left( {T}_{i}^{k},\eta ^2_{T_{i}^{k}}(W_{i}^k),{Z}_{i}^{k},{x}^{M_1}_{{T}_{i}^{k}-},\rho _{{T}_{i}^{k}-} \right) \vert \vert D^Z_{({\bullet },{\circ },{\diamond })}{x}^{M_1}_{{T}_{i}^{k}-}-D^Z_{({\bullet },{\circ },{\diamond })}{x}^{M_2}_{{T}_{i}^{k}-}\vert _{l_2} \right) ^2. \end{aligned}$$

Using Hypothesis 2.1, Lemma 3.7, (44) with

$$\begin{aligned}{\bar{\Phi }}(r,w,z,\omega ,\rho )= \left| \nabla _x{c} \left( r,\eta ^2_{r}(w),z,{x}^{M_1\vee M_2}_{r-},\rho _{r-} \right) \right| \left| D^Z_{({\bullet },{\circ },{\diamond })}{x}^{M_1\vee M_2}_{r-} \right| _{l_2},\end{aligned}$$

we get

$$\begin{aligned} O_{3,1}\le & {} {\mathbb {E}} \Big (\int _0^t\int _{[0,1]}\int _{\{ \vert z\vert>M_1\wedge M_2\}} \Big \vert \nabla _x{c}(r,\eta ^2_{r}(w),z,{x}^{M_1\vee M_2}_{r-},\rho _{r-} \Big ) \Big \vert \\{} & {} \left| D^Z_{({\bullet },{\circ },{\diamond })}{x}^{M_1\vee M_2}_{r-} \right| _{l_2}{\mathcal {N}}(\textrm{d}w,\textrm{d}z,\textrm{d}r))^2\\\le & {} C \left[ \left( \int _{\{\vert z\vert>M_1\wedge M_2\}}{\bar{c}}(z)\mu (\textrm{d}z))^2+\int _{\{\vert z\vert >M_1\wedge M_2\}}\vert {\bar{c}}(z)\vert ^2\mu (\textrm{d}z) \right. \right] \\= & {} C\varepsilon _{M_1\wedge M_2}. \end{aligned}$$

Using Hypothesis 2.1, Lemma 3.7, (45) with

$$\begin{aligned}{} & {} {\bar{\Phi }}(r,w,z,\omega ,\rho )=\vert \nabla _x{c} \left( r,\eta ^2_{r}(w),z,{x}^{M_1}_{r-},\rho _{r-} \right) \\{} & {} \quad -\nabla _x{c} \left( r,\eta ^2_{r}(w),z,{x}^{M_2}_{r-},\rho _{r-} \right) \big \vert \big \vert D^Z_{({\bullet },{\circ },{\diamond })}{x}^{M_2}_{r-} \big \vert _{l_2}, \end{aligned}$$

by Lemma 2.6, and Hölder inequality with conjugates \(1+\frac{\varepsilon _*}{2}\) and \(\frac{2+\varepsilon _*}{\varepsilon _*}\), we have

$$\begin{aligned} O_{3,2}\le & {} {\mathbb {E}} \left( \int _0^t\int _{[0,1]\times {\mathbb {R}}^d}\vert \nabla _x{c} \left( r,\eta ^2_{r}(w),z,{x}^{M_1}_{r-},\rho _{r-} \right) \right. \\{} & {} -\nabla _x{c} \big (r,\eta ^2_{r}(w),z,{x}^{M_2}_{r-},\rho _{r-} \big )\vert \vert D^Z_{({\bullet },{\circ },{\diamond })}{x}^{M_2}_{r-}\vert _{l_2}{\mathcal {N}} (\textrm{d}w,\textrm{d}z,\textrm{d}r) \big )^2\\\le & {} C{\mathbb {E}}\int _0^t\int _{[0,1]} \big \vert {x}^{M_1}_{r-}-{x}^{M_2}_{r-} \big \vert ^2 \big \vert D^Z_{({\bullet },{\circ },{\diamond })}{x}^{M_2}_{r-} \big \vert _{l_2}^2\textrm{d}w\textrm{d}r\\\le & {} C\int _0^t \big ({\mathbb {E}} \big \vert {x}^{M_1}_{r-}-{x}^{M_2}_{r-} \big \vert ^{2+\varepsilon _*} \big )^{\frac{2}{2+\varepsilon _*}}\textrm{d}r\\\le & {} C(\varepsilon _{M_1}+\varepsilon _{M_2})^{\frac{2}{2+\varepsilon _*}}. \end{aligned}$$

Using Hypothesis 2.1, (45) with

$$\begin{aligned}{\bar{\Phi }}(r,w,z,\omega ,\rho )= \big \vert \nabla _x{c} \big (r,\eta ^2_{r}(w),z,{x}^{M_1}_{r-},\rho _{r-} \big ) \big \vert \big \vert D^Z_{({\bullet },{\circ },{\diamond })}{x}^{M_1}_{r-}-D^Z_{({\bullet },{\circ },{\diamond })}{x}^{M_2}_{r-} \big \vert _{l_2},\end{aligned}$$

we have

$$\begin{aligned} O_{3,3}\le & {} {\mathbb {E}} \left( \int _0^t\int _{[0,1]\times {\mathbb {R}}^d} \big \vert \nabla _x{c} \big (r,\eta ^2_{r}(w),z,{x}^{M_1}_{r-},\rho _{r-} \big ) \big \vert \big \vert D^Z_{({\bullet },{\circ },{\diamond })}{x}^{M_1}_{r-} \right. \\{} & {} -D^Z_{({\bullet },{\circ },{\diamond })}{x}^{M_2}_{r-} \big \vert _{l_2}{\mathcal {N}}(\textrm{d}w,\textrm{d}z,\textrm{d}r) \big )^2\\\le & {} C\int _{0}^{t}{\mathbb {E}} \big \vert D^Z_{({\bullet },{\circ },{\diamond })}{x}^{M_1}_{r-}-D^Z_{({\bullet },{\circ },{\diamond })}{x}^{M_2}_{r-}\big \vert _{l_2}^2\textrm{d}r. \end{aligned}$$

Therefore,

$$\begin{aligned} O_3\le & {} C \left[ O_{3,1}+O_{3,2}+O_{3,3}]\le C[(\varepsilon _{M_1}+\varepsilon _{M_2})^{\frac{2}{2+\varepsilon _*}} \right. \nonumber \\{} & {} +\, \left. \int _{0}^{t}{\mathbb {E}} \left| D^Z_{({\bullet },{\circ },{\diamond })}{x}^{M_1}_{r}\textrm{d}r-D^Z_{({\bullet },{\circ },{\diamond })}{x}^{M_2}_{r} \right| _{l_2}^2\textrm{d}r \right] .\quad \end{aligned}$$
(151)

Combining (149), (150) and (151),

$$\begin{aligned}{} & {} {\mathbb {E}} \left| D^Z_{({\bullet },{\circ },{\diamond })}{x}^{M_1}_{t}-D^Z_{({\bullet },{\circ },{\diamond })}{x}^{M_2}_{t} \right| _{l_2}^2\\{} & {} \quad \le C \left[ (\varepsilon _{M_1}+\varepsilon _{M_2})^{\frac{2}{2+\varepsilon _*}}+\int _{0}^{t}{\mathbb {E}} \left| D^Z_{({\bullet },{\circ },{\diamond })}{x}^{M_1}_{r}-D^Z_{({\bullet },{\circ },{\diamond })}{x}^{M_2}_{r} \right| _{l_2}^2\textrm{d}r \right] . \end{aligned}$$

So we conclude by Gronwall lemma that \({\mathbb {E}}\vert D^Z_{({\bullet },{\circ },{\diamond })}{x}^{M_1}_{t}-D^Z_{({\bullet },{\circ },{\diamond })}{x}^{M_2}_{t}\vert _{l_2}^2\le C(\varepsilon _{M_1}+\varepsilon _{M_2})^{\frac{2}{2+\varepsilon _*}}\).

Finally, we recall by (24) that \(a^M_{T}=\sqrt{T\int _{\{\vert z\vert >M\}}{\underline{c}}(z) \mu (\textrm{d}z)}\) and by Hypothesis 2.3 that \({\underline{c}}(z)\le \vert {\bar{c}}(z)\vert ^2\). We notice that

$$\begin{aligned}{\mathbb {E}} \big \vert a^{M_1}_T{\varvec{e_{\diamond }}}-a^{M_2}_T{\varvec{e_{\diamond }}} \big \vert _{{\mathbb {R}}^d}^2\le C{\mathbb {E}} \big \vert a^{M_1}_T-a^{M_2}_T \big \vert ^2\le C\int _{\{\vert z\vert >{M_1\wedge M_2}\}}{\underline{c}}(z) \mu (\textrm{d}z)\le \varepsilon _{M_1\wedge M_2}.\end{aligned}$$

Then by a similar argument as above, \({\mathbb {E}}\vert D^{\Delta }_{({\diamond })}{x}^{M_1}_{t}-D^{\Delta }_{({\diamond })}{x}^{M_2}_{t}\vert _{{\mathbb {R}}^d}^2\le C(\varepsilon _{M_1}+\varepsilon _{M_2})^{\frac{2}{2+\varepsilon _*}}\), and we obtain (148).

Proof of (iii) (iii) is an immediate consequence of (i) and (ii).

\(\square \)

Rights and permissions

Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Qin, Y. Approximation Schemes for McKean–Vlasov and Boltzmann-Type Equations (Error Analysis in Total Variation Distance). J Theor Probab 37, 1523–1596 (2024). https://doi.org/10.1007/s10959-024-01324-6

Download citation

  • Received:

  • Revised:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s10959-024-01324-6

Keywords

Mathematics Subject Classification

Navigation