Skip to main content
Log in

Traveling Waves Near Couette Flow for the 2D Euler Equation

  • Published:
Communications in Mathematical Physics Aims and scope Submit manuscript

Abstract

In this paper we reveal the existence of a large family of new, nontrivial and smooth traveling waves for the 2D Euler equation at an arbitrarily small distance from the Couette flow in \(H^s\), with \(s<3/2\), at the level of the vorticity. The speed of these waves is of order 1 with respect to this distance. This result strongly contrasts with the setting of very high regularity in Gevrey spaces [see Bedrossian and Masmoudi (Publ Math Inst Hautes Etudes Sci 122:195–300, 2015)], where the problem exhibits an inviscid damping mechanism that leads to relaxation of perturbations back to nearby shear flows. It also complements the fact that there not exist nontrivial traveling waves in the \(H^{\frac{3}{2}+}\) neighborhoods of Couette flow [see Lin and Zeng (Arch Ration Mech Anal 200:1075–1097, 2011)].

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

Access this article

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

Instant access to the full article PDF.

Fig. 1

Similar content being viewed by others

References

  1. Antonelli, P., Dolce, M., Marcati, P.: Linear stability analysis of the homogeneous Couette flow in a 2D isentropic compressible fluid. Ann. PDE 7, 24 (2021)

    MathSciNet  MATH  Google Scholar 

  2. Ao, W., Davila, J., del Pino, M., Musso, M., Wei, J.: Travelling and rotating solutions to the generalized inviscid surface quasi-geostrophic equation. Trans. Am. Math. Soc. 374, 6665–6689 (2021)

    MathSciNet  MATH  Google Scholar 

  3. Bedrossian, J., Germain, P., Masmoudi, N.: On the stability threshold for the 3D Couette flow in Sobolev regularity. Ann. Math. 185, 541–608 (2017)

    MathSciNet  MATH  Google Scholar 

  4. Bedrossian, J., Masmoudi, N., Vicol, V.: Enhanced dissipation and inviscid damping in the inviscid limit of the Navier–Stokes equations near the 2D Couette flow. Arch. Rat. Mech. Anal. 219, 1087–1159 (2016)

    MATH  Google Scholar 

  5. Bedrossian, J., Zelati, M.C.: Enhanced dissipation, hypoellipticity, and anomalous small noise inviscid limits in shear flows. Arch. Ration. Mech. Anal. 224, 1161–1204 (2017)

    MathSciNet  MATH  Google Scholar 

  6. Bedrossian, J., Coti Zelati, M., Vicol, V.: Vortex Axisymmetrization, Inviscid Damping, and Vorticity Depletion in the Linearized 2D Euler Equations. Ann. PDE 5, 4 (2019)

    MathSciNet  MATH  Google Scholar 

  7. Bedrossian, J., Masmoudi, N.: Inviscid damping and the asymptotic stability of planar shear flows in the 2D Euler equations. Publ. Math. Inst. Hautes Études Sci. 122, 195–300 (2015)

    MathSciNet  MATH  Google Scholar 

  8. Bedrossian, J., Germain, P., Masmoudi, N.: Stability of the Couette flow at high Reynolds numers in 2D and 3D. Bull. Am. Math. Soc. 56(3), 373–414 (2019)

    MATH  Google Scholar 

  9. Bedrossian, J., Bianchini, R., Coti-Zelati, M., Dolce, M.: Nonlinear inviscid damping and shear-buoyancy instability in the two-dimensional Boussinesq equations. arXiv:2103.13713

  10. Bianchini, R., Coti-Zelati, M., Dolce, M.: Linear inviscid damping for shear flows near Couette in the 2D stably stratified regime. To appear in Indiana University Mathematics Journal, arXiv:2005.09058

  11. Bouchet, F., Morita, H.: Large time behavior and asymptotic stability of the 2D Euler and linearized Euler equations. Phys. D 239(12), 948–966 (2010)

    MathSciNet  MATH  Google Scholar 

  12. Burbea, J.: Motions of vortex patches. Lett. Math. Phys. 6, 1–16 (1982)

    ADS  MathSciNet  MATH  Google Scholar 

  13. Castro, A., Córdoba, D., Gómez-Serrano, J.: Uniformly rotating analytic global patch solutions for active scalars. Ann. PDE 2(1), 1–34 (2016)

    MathSciNet  MATH  Google Scholar 

  14. Castro, A., Córdoba, D., Gómez-Serrano, J.: Existence and regularity of rotating global solutions for the generalized surface quasi-geostrophic equations. Duke Math. J. 165(5), 935–984 (2016)

    MathSciNet  MATH  Google Scholar 

  15. Castro, A., Córdoba, D., Gómez-Serrano, J.: Global smooth solutions for the inviscid SQG equation. Mem. AMS 266(1292), 89 (2020)

    MathSciNet  MATH  Google Scholar 

  16. Castro, A., Córdoba, D., Gómez-Serrano, J.: Uniformly rotating smooth solutions for the incompressible 2D Euler equations. Arch. Ration. Mech. Anal. 231(2), 719–785 (2019)

    MathSciNet  MATH  Google Scholar 

  17. Córdoba, D., Granero-Belinchón, R., Orive, R.: The confined Muskat problem: differences with the deep water regime. Commun. Math. Sci. 12, 423–455 (2014)

    MathSciNet  MATH  Google Scholar 

  18. Coti-Zelati, M., Elgindi, T.M., Widmayer, K.: Enhanced Dissipation in the Navier-Stokes Equations Near the Poiseuille Flow. Commun. Math. Phys. 378, 987–1010 (2020)

    ADS  MathSciNet  MATH  Google Scholar 

  19. Coti Zelati, M., Elgindi, T.M., Widmayer, K.: Stationary Structures near the Kolmogorov and Poiseuille Flows in the 2D Euler Equations. arXiv:2007.11547

  20. Coti Zelati, M., Zillinger, C.: On degenerate circular and shear flows: the point vortex and power law circular flows. Commun. Part. Differ. Equ. 44(2), 110–155 (2019)

    MathSciNet  MATH  Google Scholar 

  21. Crandall, M.G., Rabinowitz, P.H.: Bifurcation from simple eigenvalues. J. Funct. Anal. 8, 321–340 (1971)

    MathSciNet  MATH  Google Scholar 

  22. Deng, Y., Masmoudi, N.: Long time instability of the Couette flow in low Gevrey spaces. To appear in Comm. Pure Appl. Math. arXiv:1803.01246

  23. Deng, Y., Zillinger, C.: On the smallness condition in linear inviscid damping: monotonicity and resonance chains. Nonlinearity 33, 6176 (2020)

    ADS  MathSciNet  MATH  Google Scholar 

  24. Deng, Y., Zillinger, C.: Echo chains as a linear mechanism: norm inflation, modified exponents and asymptotics. Arch. Ration. Mech. Anal. 242(1), 643–700 (2021)

    MathSciNet  MATH  Google Scholar 

  25. Deng, W., Wu, J., Zhang, P.: Stability of couette flow for 2D Boussinesq system with vertical dissipation. J. Funct. Anal. 281(12), 109255 (2021)

    MathSciNet  MATH  Google Scholar 

  26. De la Hoz, F., Hassainia, Z., Hmidi, T.: Doubly Connected V-States for the Generalized Surface Quasi-Geostrophic Equations. Arch. Ration. Mech. Anal. 220, 1209–1281 (2016)

    MathSciNet  MATH  Google Scholar 

  27. De la Hoz, F., Hmidi, T., Mateu, J., Verdera, J.: Doubly connected V-states for the planar Euler equations. SIAM J. Math. Anal. 48, 1892–1928 (2016)

    MathSciNet  MATH  Google Scholar 

  28. Dieudonné, J.: Foundations of Modern Analysis. Academic Press, New York (1960)

    MATH  Google Scholar 

  29. Dritschel, D.G., Hmidi, T., Renault, C.: Imperfect bifurcation for the quasi-geostrophic shallow-water equations. Arch. Ration. Mech. Anal. 231, 1853–1915 (2019)

    MathSciNet  MATH  Google Scholar 

  30. Gallay, T.: Stability of vortices in ideal fluids: the legacy of Kelvin and Rayleigh. Hyperb. Prob. Theory Numer. Appl. Proc. HYP2018 10, 42–59 (2018)

    MathSciNet  MATH  Google Scholar 

  31. Gallay, T., Wayne, C.E.: Global stability of vortex solutions of the two-dimensional Navier–Stokes equation. Commun. Math. Phys. 255, 97–129 (2005)

    ADS  MathSciNet  MATH  Google Scholar 

  32. García, C., Hmidi, T., Soler, J.: Non uniform rotating vortices and periodic orbits for the two-dimensional Euler equations. Arch. Ration. Mech. Anal. 238, 929–1086 (2020)

    MathSciNet  MATH  Google Scholar 

  33. Gómez-Serrano, J.: On the existence of stationary patches. Adv. Math. 343, 110–140 (2019)

    MathSciNet  MATH  Google Scholar 

  34. Gómez-Serrano, J., Park, J., Shi, J., Yao, Y.: Symmetry in stationary and uniformly rotating solutions of active scalar equations. Duke Math. J. 170(13), 2957–3038 (2021)

    MathSciNet  MATH  Google Scholar 

  35. Gómez-Serrano, J., Park, J., Shi, J., Yao, Y.: Remarks on stationary and uniformly-rotating vortex sheets: rigidity results. Commun. Math. Phys. 386, 1845–1879 (2021)

    ADS  MathSciNet  MATH  Google Scholar 

  36. Gómez-Serrano, J., Park, J., Shi, J., Yao, Y.: Remarks on stationary and uniformly-rotating vortex sheets: flexibility results. To appear in Phil. Trans. R. Soc. A. arXiv:2012.08709

  37. Gravejat, P., Smets, D.: Smooth traveling-wave solutions to the inviscid surface quasi-geostrophic equation. Int. Math. Res. Not. 2019(6), 1744–1757 (2019)

    MATH  Google Scholar 

  38. Grenier, E., Nguyen, T., Rousset, F., Soffer, A.: Linear inviscid damping and enhanced viscous dissipation of shear flows by using the conjugate operator method. J. Funct. Anal. 278(3), 108339 (2020)

    MathSciNet  MATH  Google Scholar 

  39. Hassainia, Z., Hmidi, T.: Existence of corotating asymmetric vortex pairs for Euler equations. arXiv:2004.02306

  40. Hassainia, Z., Hmidi, T.: On the V-states for the generalized quasi-geostrophic equations. Comm. Math. Phys. 337(1), 321–377 (2015)

    ADS  MathSciNet  MATH  Google Scholar 

  41. Hmidi, T., Renault, C.: Existence of small loops in a bifurcation diagram near degenerate eigenvalues. Nonlinearity 30(10), 3821 (2017)

    ADS  MathSciNet  MATH  Google Scholar 

  42. Hmidi, T., Mateu, J.: Existence of corotating and counter-rotating vortex pairs for active scalar equations. Comm. Math. Phys. 350(2), 699–747 (2017)

    ADS  MathSciNet  MATH  Google Scholar 

  43. Hmidi, T., Mateu, J., Verdera, J.: Boundary regularity of rotating vortex patches. Arch. Ration. Mech. Anal. 209, 171–208 (2013)

    MathSciNet  MATH  Google Scholar 

  44. Ionescu, A.D., Jia, H.: Inviscid damping near the couette flow in a channel. Commun. Math. Phys. 374(3), 2015–2096 (2019)

    ADS  MathSciNet  MATH  Google Scholar 

  45. Ionescu, A. D., Jia, H.: Nonlinear inviscid damping near monotonic shear flows. To appear in Acta Mathematica arXiv:2001.03087

  46. Ionescu, A.D., Jia, H.: Axi-symmetrization near Point Vortex Solutions for the 2D Euler Equation. Comm. Pure Appl. Math. (2021). https://doi.org/10.1002/cpa.21974

    Article  MATH  Google Scholar 

  47. Ionescu, A. D., Jia, H.: Linear vortex symmetrization: the spectral density function. arXiv:2109.12815

  48. Jia, H.: Linear Inviscid damping near monotone shear flows. SIAM J. Math Anal. 52(1), 623–652 (2020)

    MathSciNet  MATH  Google Scholar 

  49. Jia, H.: Linear inviscid damping in Gevrey spaces. Arch. Ration. Mech. Anal. 235, 1327–1355 (2020)

    MathSciNet  MATH  Google Scholar 

  50. Li, Y.C., Lin, Z.: A resolution of the Sommerfeld paradox. SIAM J. Math. Anal. 43, 1923–1954 (2011)

    MathSciNet  MATH  Google Scholar 

  51. Lin, Z., Zeng, C.: Inviscid dynamical structures near Couette flow. Arch. Ration. Mech. Anal. 200, 1075–1097 (2011)

    MathSciNet  MATH  Google Scholar 

  52. Lin, Z., Yang, J., Zhu, H.: Barotropic instability of shear flows. Stud. Appl. Math. 144, 289–326 (2020)

    MathSciNet  MATH  Google Scholar 

  53. Lin, Z.: Nonlinear instability of ideal plane flows. Int. Math. Res. Not. 2004, 2147–2178 (2004)

    MathSciNet  MATH  Google Scholar 

  54. Lin, Z., Wei, D., Zhang, Z., Zhu, H.: The number of traveling wave families in a running water with Coriolis force. arXiv:2009.05733

  55. Masmoudi, N., Said-Houari, B., Zhao, W.: Stability of Couette flow for 2D Boussinesq system without thermal diffusivity. arXiv:2010.01612

  56. Masmoudi, N., Zhao, W.: Nonlinear inviscid damping for a class of monotone shear flows in finite channel. arXiv:2001.08564

  57. Mouhot, C., Villani, C.: On Landau damping. Acta Math. 207(1), 29–201 (2011)

    MathSciNet  MATH  Google Scholar 

  58. Renault, C.: Relative equilibria with holes for the surface quasi-geostrophic equations. J. Differ. Equ. Vol. 263(1), 567–614 (2017)

    ADS  MathSciNet  MATH  Google Scholar 

  59. Trefethen, L.N.: Pseudospectra of linear operators. SIAM Rev. 39(3), 383–406 (1997)

    ADS  MathSciNet  MATH  Google Scholar 

  60. Trefethen, L.N., Trefethen, A.E., Reddy, S.C., Driscoll, T.A.: Hydrodynamic stability without eigenvalues. Science 261(5121), 578–584 (1993)

    ADS  MathSciNet  MATH  Google Scholar 

  61. Wei, D., Zhang, Z., Zhao, W.: Linear inviscid damping and vorticity depletion for shear flows. Ann. PDE 5, 3 (2019)

    MathSciNet  MATH  Google Scholar 

  62. Wei, D., Zhang, Z., Zhao, W.: Linear inviscid damping for a class of monotone shear flow in Sobolev spaces. Comm. Pure Appl. Math. 71, 617–687 (2018)

    MathSciNet  MATH  Google Scholar 

  63. Wei, D., Zhang, Z., Zhao, W.: Linear inviscid damping and enhanced dissipation for the Kolmogorov flow. Adv. Math. 362, 106963 (2020)

    MathSciNet  MATH  Google Scholar 

  64. Zillinger, C.: Linear inviscid damping for monotone shear flows in a finite periodic channel, boundary effects, blow-up and critical sobolev regularity. Arch. Ration. Mech. Anal. 221, 1449–1509 (2016)

    MathSciNet  MATH  Google Scholar 

  65. Zillinger, C.: Linear inviscid damping for monotone shear flows. Trans. Am. Math. Soc. 369, 8799–8855 (2017)

    MathSciNet  MATH  Google Scholar 

  66. Zillinger, C.: On enhanced dissipation for the Boussinesq equations. J. Differ. Equ. 282, 407–445 (2021)

    ADS  MathSciNet  MATH  Google Scholar 

  67. Zillinger, C.: On echo chains in the linearized Boussinesq equations around traveling waves, arXiv:2103.15441

Download references

Acknowledgements

AC is supported by grants CEX2019-000904-S and RED2018-102650-T funded by: MCIN/AEI/ 10.13039/501100011033 and by grant Europa Excelencia program ERC2018-092824. AC and DL are partially supported by grants MTM2017-89976-P and PID2020-114703GB-I00 funded by: MCIN/AEI/ 10.13039/501100011033.

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to Ángel Castro.

Ethics declarations

Conflict of interest

The authors declare that they have no conflict of interest.

Additional information

Communicated by A. Ionescu.

Publisher's Note

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

Appendix

Appendix

In order to facilitate the presentation of the manuscript, we collect in this section all the technical lemmas used previously. We start recalling the definition of the kernel given by

$$\begin{aligned} K[g](\textbf{x},{\bar{\textbf{x}}}):=\log \left[ \cosh ({\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[g](\textbf{x}))-\cos ({\bar{x}})\right] , \end{aligned}$$
(125)

where

$$\begin{aligned} \Delta _{{\bar{\textbf{x}}}}[g](\textbf{x}):=g(\textbf{x})-g(\textbf{x}-{\bar{\textbf{x}}}). \end{aligned}$$

In order to take derivatives into the kernel (125) we have introduced the following two functions

$$\begin{aligned} \Psi _1[g](\textbf{x},{\bar{\textbf{x}}})&:=\frac{\sinh ({\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[g](\textbf{x}))}{\cosh ({\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[g](\textbf{x}))-\cos ({\bar{x}})},\\ \Psi _2[g](\textbf{x},{\bar{\textbf{x}}})&:=\frac{\cosh ({\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[g](\textbf{x}))}{\cosh ({\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[g](\textbf{x}))-\cos ({\bar{x}})}, \end{aligned}$$

with derivatives given by the expressions

$$\begin{aligned} \partial _{\textbf{x}}\Psi _1[g](\textbf{x},{\bar{\textbf{x}}})&=[\Psi _2[g]-\Psi _1^2[g]](\textbf{x},{\bar{\textbf{x}}}) \Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}g](\textbf{x}), \end{aligned}$$
(126)
$$\begin{aligned} \partial _{\textbf{x}}\Psi _2[g](\textbf{x},{\bar{\textbf{x}}})&=[\Psi _1[g]\left( 1-\Psi _2[g]\right) ](\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}g](\textbf{x}). \end{aligned}$$
(127)

Combining (125) with (126) and (127) it is just a matter of algebra to obtain

$$\begin{aligned} \partial _{\textbf{x}}K[g](\textbf{x},{\bar{\textbf{x}}})&=\Psi _1[g](\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}g](\textbf{x}),\\ \partial _{\textbf{x}}^2 K[g](\textbf{x},{\bar{\textbf{x}}})&=[\Psi _2[g]-\Psi _1^2[g]](\textbf{x},{\bar{\textbf{x}}})\left( \Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}g](\textbf{x})\right) ^2+\Psi _1[g](\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}^2 g](\textbf{x}),\\ \partial _{\textbf{x}}^3 K[g](\textbf{x},{\bar{\textbf{x}}})&=\Psi _1[g](\textbf{x},{\bar{\textbf{x}}})[ 1-3\Psi _2[g]+ 2 \Psi _1^2[g]](\textbf{x},{\bar{\textbf{x}}}) \left( \Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}g](\textbf{x})\right) ^3\\&\quad +3[\Psi _2[g]- \Psi _1^2[g]](\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}g](\textbf{x})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}^2 g](\textbf{x})\\&\quad + \Psi _1[g](\textbf{x},{\bar{\textbf{x}}}) \Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}^3 g](\textbf{x}). \end{aligned}$$

Now, before starting with the proof of the lemmas, we remember the Taylor expansion of the trigonometric functions involved in the definition of \(\Psi _{i}(\textbf{x},{\bar{\textbf{x}}})\) for \(i=1,2.\)

$$\begin{aligned} \sinh (z):=\sum _{n=0}^{+\infty }\frac{z^{2n+1}}{(2n+1)!},\qquad \cosh (z):=\sum _{n=0}^{+\infty }\frac{z^{2n}}{(2n)!},\qquad \text {and} \qquad \cos (z):=\sum _{n=0}^{+\infty }\frac{(-1)^n z^{2n}}{(2n)!}. \end{aligned}$$

Using the above expressions, it is simple to check that there exists some constant \(C(\Vert g\Vert _{H^{3}(D_{\varepsilon })})\) such that for any \({\bar{\textbf{x}}}\in {\mathbb {T}}\times {\mathbb {R}}\) with \(|{\bar{\textbf{x}}}|\ll 1\), we get

$$\begin{aligned} |\cosh ({\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[g](\textbf{x}))|&\le C(\Vert g\Vert _{H^{3}(D_{\varepsilon })}), \nonumber \\ |\sinh ({\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[g](\textbf{x}))|&\le C(\Vert g\Vert _{H^{3}(D_{\varepsilon })}) |{\bar{\textbf{x}}}|, \nonumber \\ |\cosh ({\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[g](\textbf{x}))-\cos ({\bar{x}})|&\ge \left( \tfrac{1}{2}-\Vert g\Vert _{H^{3}(D_{\varepsilon })}\right) |{\bar{\textbf{x}}}|^2-C(\Vert g\Vert _{H^{3}(D_{\varepsilon })}) |{\bar{\textbf{x}}}|^4, \end{aligned}$$
(128)

and as an immediate consequence we obtain the following result.

Corollary 7.1

Let \(g\in {\mathbb {B}}_{\delta }(H^3(D_{\varepsilon }))\) with \(0< \delta (\varepsilon )\ll 1\). For \({\bar{\textbf{x}}}\in {\mathbb {T}}\times {\mathbb {R}}\) such that \(|{\bar{\textbf{x}}}|\ll 1\) we have

$$\begin{aligned} |\Psi _i[g](\textbf{x},{\bar{\textbf{x}}})|\le \frac{C(\Vert g \Vert _{H^{3}(D_{\varepsilon })})}{ 1-C(\Vert g \Vert _{H^{3}(D_{\varepsilon })}) }\frac{1}{|{\bar{\textbf{x}}}|^i} \qquad \text {for} \quad i=1,2. \end{aligned}$$

Now, we focus our attention in the higher order derivatives \(\partial _{\textbf{x}}^2 K[g](\textbf{x},{\bar{\textbf{x}}})\) and \(\partial _{\textbf{x}}^3 K[g](\textbf{x},{\bar{\textbf{x}}})\). It will be convenient to introduce some notation in order to handle these terms. To be more specific, we will use the following expressions:

$$\begin{aligned} \partial _{\textbf{x}}^2 K[g](\textbf{x},{\bar{\textbf{x}}})&={\textsf{I}}_1[g](\textbf{x},{\bar{\textbf{x}}})+{\textsf{I}}_2[g](\textbf{x},{\bar{\textbf{x}}}), \end{aligned}$$
(129)
$$\begin{aligned} \partial _{\textbf{x}}^3 K[g](\textbf{x},{\bar{\textbf{x}}})&={\textsf{J}}_1[g](\textbf{x},{\bar{\textbf{x}}})+{\textsf{J}}_2[g](\textbf{x},{\bar{\textbf{x}}})+{\textsf{J}}_3[g](\textbf{x},{\bar{\textbf{x}}}) , \end{aligned}$$
(130)

where

$$\begin{aligned} \begin{aligned} {\textsf{I}}_1[g](\textbf{x},{\bar{\textbf{x}}})&:={\tilde{{\textsf{I}}}}_1[g](\textbf{x},{\bar{\textbf{x}}})\left( \Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}g](\textbf{x})\right) ^2,\\ {\textsf{I}}_2[g](\textbf{x},{\bar{\textbf{x}}})&:={\tilde{{\textsf{I}}}}_2[g](\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}^2 g](\textbf{x}),\\ {\textsf{J}}_1[g](\textbf{x},{\bar{\textbf{x}}})&:={\tilde{{\textsf{J}}}}_1[g](\textbf{x},{\bar{\textbf{x}}})\left( \Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}g](\textbf{x})\right) ^3,\\ {\textsf{J}}_2[g](\textbf{x},{\bar{\textbf{x}}})&:={\tilde{{\textsf{J}}}}_2[g](\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}g](\textbf{x})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}^2 g](\textbf{x}),\\ {\textsf{J}}_3[g](\textbf{x},{\bar{\textbf{x}}})&:={\tilde{{\textsf{J}}}}_3[g](\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}^3 g](\textbf{x}), \end{aligned} \quad \quad \begin{aligned} {\tilde{{\textsf{I}}}}_1[g]&:=\Psi _2[g]-\Psi _1^2[g],\\ {\tilde{{\textsf{I}}}}_2[g]&:=\Psi _1[g],\\ {\tilde{{\textsf{J}}}}_1[g]&:=\Psi _1[g]\left( 1- 3\Psi _2[g]+2 \Psi _1^2[g]\right) ,\\ {\tilde{{\textsf{J}}}}_2[g]&:=3\Psi _2[g]-\Psi _1^2[g],\\ {\tilde{{\textsf{J}}}}_3[g]&:=\Psi _1[g]. \end{aligned} \end{aligned}$$

As an immediate consequence of Corollary 7.1 we get the following bounds for the above expressions.

Corollary 7.2

Let \(g\in {\mathbb {B}}_{\delta }(H^3(D_{\varepsilon }))\) with \(0< \delta (\varepsilon )\ll 1\). For \({\bar{\textbf{x}}}\in {\mathbb {T}}\times {\mathbb {R}}\) such that \(|{\bar{\textbf{x}}}|\ll 1\) we have

$$\begin{aligned} |{\tilde{{\textsf{I}}}}_2[g](\textbf{x},{\bar{\textbf{x}}})|, |{\tilde{{\textsf{J}}}}_3[g](\textbf{x},{\bar{\textbf{x}}})|\le C(\Vert g\Vert _{H^{3}(D_{\varepsilon })}) |{\bar{\textbf{x}}}|^{-1},\\ |{\tilde{{\textsf{I}}}}_1[g](\textbf{x},{\bar{\textbf{x}}})|, |{\tilde{{\textsf{J}}}}_2[g](\textbf{x},{\bar{\textbf{x}}})|\le C(\Vert g\Vert _{H^{3}(D_{\varepsilon })}) |{\bar{\textbf{x}}}|^{-2},\\ |{\tilde{{\textsf{J}}}}_1[g](\textbf{x},{\bar{\textbf{x}}})|\le C(\Vert g \Vert _{H^{3}(D_{\varepsilon })}) |{\bar{\textbf{x}}}|^{-3}. \end{aligned}$$

Before starting with the proof of the main lemmas of the Appendix, we will collect a couple of auxiliary results. In first place, we note that for any \(g\in H^{4,3}(D_{\varepsilon }),\) an inequality that we will repeatedly apply in our arguments is the following:

$$\begin{aligned} |\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}^ig](\textbf{x})|\le |{\bar{\textbf{x}}}|^{2-i} \Vert g\Vert _{H^{4,3}(D_{\varepsilon })}, \qquad i=1,2. \end{aligned}$$
(131)

Note that the above reduces to the Sobolev embedding \( H^{3,2}(D_\varepsilon )\subset C^{1}(D_{\varepsilon })\) and \(H^{2,1}(D_\varepsilon )\subset C(D_{\varepsilon })\). Secondly, we obtain an uniform bound for the difference of the auxiliary functions \(\Psi _i[{\textsf{f}}']-\Psi _i[{\textsf{f}}'']\) and their derivatives in terms of the \(H^{4,3}(D_{\varepsilon })\)-norm of the difference \({\textsf{f}}'-{\textsf{f}}''.\)

Lemma 7.3

Let \({\textsf{f}}', {\textsf{f}}'' \in {\mathbb {B}}_{\delta }(H^{4,3}(D_{\varepsilon }))\) with \(0<\delta (\varepsilon )\ll 1\) small enough. The following bounds hold

  1. i)
    $$\begin{aligned} |(\Psi _i[{\textsf{f}}']-\Psi _i[{\textsf{f}}''])(\textbf{x},{\bar{\textbf{x}}})|^2 |{\bar{\textbf{x}}}|^{2i}\le C(\varepsilon ) \Vert {\textsf{f}}'-{\textsf{f}}''\Vert _{H^{3}(D_{\varepsilon })}^2, \qquad i=1,2. \end{aligned}$$
  2. ii)
    $$\begin{aligned} |(\partial _{\textbf{x}}\Psi _i[{\textsf{f}}']-\partial _{\textbf{x}}\Psi _i[{\textsf{f}}''])(\textbf{x},{\bar{\textbf{x}}})|^2 |{\bar{\textbf{x}}}|^{2i}\le C(\varepsilon ) \Vert {\textsf{f}}'-{\textsf{f}}''\Vert _{H^{4,3}(D_{\varepsilon })}^2, \qquad i=1,2. \end{aligned}$$
  3. iii)
    $$\begin{aligned} |(\partial _{\textbf{x}}^2\Psi _i[{\textsf{f}}']-\partial _{\textbf{x}}^2\Psi _i[{\textsf{f}}''])(\textbf{x},{\bar{\textbf{x}}})|^2 |{\bar{\textbf{x}}}|^{2(i+1)}\le C(\varepsilon ) \Vert {\textsf{f}}'-{\textsf{f}}''\Vert _{H^{4,3}(D_{\varepsilon })}^2, \qquad i=1,2. \end{aligned}$$

Proof of (i)

Using some trigonometric identities for hyperbolic functions \(\sinh (\cdot ), \cosh (\cdot )\) we get

$$\begin{aligned}&\left( \Psi _1[{\textsf{f}}']-\Psi _1[{\textsf{f}}'']\right) (\textbf{x},{\bar{\textbf{x}}})\\&\quad =\frac{\sinh (\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}'-{\textsf{f}}''](\textbf{x}))-\cos ({\bar{x}})\cosh ({\bar{y}})\left[ \sinh (\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}'](\textbf{x}))-\sinh (\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}''](\textbf{x}))\right] }{\left[ \cosh ({\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}'](\textbf{x}))-\cos ({\bar{x}})\right] \left[ \cosh ({\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}''](\textbf{x}))-\cos ({\bar{x}})\right] }\nonumber \\&\qquad -\frac{\cos ({\bar{x}})\sinh ({\bar{y}})\left[ \cosh (\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}'](\textbf{x}))-\cosh (\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}''](\textbf{x}))\right] }{\left[ \cosh ({\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}'](\textbf{x}))-\cos ({\bar{x}})\right] \left[ \cosh ({\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}''](\textbf{x}))-\cos ({\bar{x}})\right] }, \end{aligned}$$

and

$$\begin{aligned}&\left( \Psi _2[{\textsf{f}}']-\Psi _2[{\textsf{f}}'']\right) (\textbf{x},{\bar{\textbf{x}}})\\&\quad =\frac{\cos ({\bar{x}})\left[ \cosh ({\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}''](\textbf{x}))-\cosh ({\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}'](\textbf{x}))\right] }{\left[ \cosh ({\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}'](\textbf{x}))-\cos ({\bar{x}})\right] \left[ \cosh ({\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}''](\textbf{x}))-\cos ({\bar{x}})\right] }. \end{aligned}$$

Notice that both denominators vanish if and only if \({\bar{\textbf{x}}}=0\). Now, since \({\textsf{f}}',{\textsf{f}}''\in H^{4,3}(D_{\varepsilon })\subset C^2(D_{\varepsilon })\) are continuous functions on a bounded domain, the outer region is easily bounded by the required term. Then, we just focus our attention on the inner region \(|{\bar{\textbf{x}}}|\ll 1\).

Applying Taylor expansion of each trigonometric function involved we obtain that each numerator of the previous expression can be bound as follows

$$\begin{aligned}{} & {} \left| \sinh (\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}'-{\textsf{f}}''](\textbf{x}))-\cos ({\bar{x}})\cosh ({\bar{y}})\left[ \sinh (\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}'](\textbf{x}))-\sinh (\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}''](\textbf{x}))\right] \right| \\{} & {} \quad \le C(\varepsilon ) |{\bar{\textbf{x}}}|^3 \frac{1}{1-\max \{\Vert {\textsf{f}}' - {\textsf{f}}''\Vert _{H^{3}},\Vert {\textsf{f}}'\Vert _{H^{3}},\Vert {\textsf{f}}''\Vert _{H^{3}}\}\text {diam}(D_{\varepsilon })} \Vert {\textsf{f}}'-{\textsf{f}}''\Vert _{H^{3}},\\{} & {} \left| \cos ({\bar{x}})\sinh ({\bar{y}})\left[ \cosh (\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}'](\textbf{x}))-\cosh (\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}''](\textbf{x}))\right] \right| \\{} & {} \quad \le C(\varepsilon ) |{\bar{\textbf{x}}}|^3 \frac{\max \{\Vert {\textsf{f}}'\Vert _{H^{3}},\Vert {\textsf{f}}''\Vert _{H^{3}}\}}{1-\max \{\Vert {\textsf{f}}'\Vert _{H^{3}},\Vert {\textsf{f}}''\Vert _{H^{3}}\}\text {diam}(D_{\varepsilon })} \Vert {\textsf{f}}'-{\textsf{f}}''\Vert _{H^{3}}, \end{aligned}$$

and

$$\begin{aligned}{} & {} \left| \cos ({\bar{x}})\left[ \cosh ({\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}'](\textbf{x}))-\cosh ({\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}''](\textbf{x}))\right] \right| \\{} & {} \quad \le C(\varepsilon ) |{\bar{\textbf{x}}}|^2 \frac{\max \{\Vert {\textsf{f}}'\Vert _{H^{3}},\Vert {\textsf{f}}''\Vert _{H^{3}}\}}{1-\max \{\Vert {\textsf{f}}'\Vert _{H^{3}},\Vert {\textsf{f}}''\Vert _{H^{3}}\}\text {diam}(D_{\varepsilon })} \Vert {\textsf{f}}'-{\textsf{f}}''\Vert _{H^{3}}. \end{aligned}$$

Combining all with the usual lower bound for the denominator (see (128)), we obtain

$$\begin{aligned} |(\Psi _i[{\textsf{f}}']-\Psi _i[{\textsf{f}}''])(\textbf{x},{\bar{\textbf{x}}})|^2 |{\bar{\textbf{x}}}|^{2j} \le \left( \frac{C(\varepsilon ,\Vert {\textsf{f}}'\Vert _{H^{3}(D_{\varepsilon })},\Vert {\textsf{f}}''\Vert _{H^{3}(D_{\varepsilon })})}{1-\max \{\Vert {\textsf{f}}'\Vert _{H^{3}},\Vert {\textsf{f}}''\Vert _{H^{3}}\}\text {diam}(D_{\varepsilon })}\right) \Vert {\textsf{f}}'-{\textsf{f}}''\Vert _{H^3(D_{\varepsilon })}^2. \end{aligned}$$

Finally, as \({\textsf{f}}', {\textsf{f}}'' \in {\mathbb {B}}_{\delta }(H^{4,3}(D_{\varepsilon })),\) taking \(0<\delta <\text {diam}(D_{\varepsilon })^{-1}\) we have proved our goal. \(\square \)

Proof of (ii). Due to the relations

$$\begin{aligned} \partial _{\textbf{x}}\Psi _1[g](\textbf{x},{\bar{\textbf{x}}})&=(\Psi _2[g]-\Psi _1^2[g])(\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}g](\textbf{x}),\\ \partial _{\textbf{x}}\Psi _2[g](\textbf{x},{\bar{\textbf{x}}})&=(\Psi _1[g]-\Psi _1[g]\Psi _2[g])(\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}g](\textbf{x}), \end{aligned}$$

it is clear that adding and subtracting some appropriate term we obtain the expressions

$$\begin{aligned} (\partial _{\textbf{x}}\Psi _1[{\textsf{f}}']-\partial _{\textbf{x}}\Psi _1[{\textsf{f}}''])(\textbf{x},{\bar{\textbf{x}}})&=(\Psi _2[{\textsf{f}}']-(\Psi _1[{\textsf{f}}'])^2)(\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}({\textsf{f}}'-{\textsf{f}}'')](\textbf{x}) \nonumber \\&\quad +(\Psi _2[{\textsf{f}}']-\Psi _2[{\textsf{f}}''])(\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}{\textsf{f}}''](\textbf{x})\nonumber \\&\quad - (\Psi _1[{\textsf{f}}']-\Psi _1[{\textsf{f}}''])(\textbf{x},{\bar{\textbf{x}}})(\Psi _1[{\textsf{f}}']+\Psi _1[{\textsf{f}}''])(\textbf{x},{\bar{\textbf{x}}}) \Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}{\textsf{f}}'](\textbf{x}), \end{aligned}$$
(132)

and

$$\begin{aligned} (\partial _{\textbf{x}}\Psi _2[{\textsf{f}}']-\partial _{\textbf{x}}\Psi _2[{\textsf{f}}''])(\textbf{x},{\bar{\textbf{x}}})&=(\Psi _1[{\textsf{f}}']-\Psi _1[{\textsf{f}}']\Psi _2[{\textsf{f}}'])(\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}({\textsf{f}}'-{\textsf{f}}'')](\textbf{x})\nonumber \\&\quad + (\Psi _1[{\textsf{f}}']-\Psi _1[{\textsf{f}}''] )(\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}{\textsf{f}}''](\textbf{x})\nonumber \\&\quad - \Psi _1[{\textsf{f}}'](\textbf{x},{\bar{\textbf{x}}})(\Psi _2[{\textsf{f}}']-\Psi _2[{\textsf{f}}''] )(\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}{\textsf{f}}''](\textbf{x})\nonumber \\&\quad - \Psi _2[{\textsf{f}}''](\textbf{x},{\bar{\textbf{x}}})(\Psi _1[{\textsf{f}}']-\Psi _1[{\textsf{f}}''] )(\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}{\textsf{f}}''](\textbf{x}). \end{aligned}$$
(133)

As before, we just focus on the inner region \(|{\bar{\textbf{x}}}|\ll 1,\) which is the most singular part. Now, remembering (131) and applying repeatedly Corollary 7.1, we get for \(1\le i \le 2 \) and \(|{\bar{\textbf{x}}}|\ll 1\) that

$$\begin{aligned}{} & {} |(\partial _{\textbf{x}}\Psi _i[{\textsf{f}}']-\partial _{\textbf{x}}\Psi _i[{\textsf{f}}''])(\textbf{x},{\bar{\textbf{x}}})||{\bar{\textbf{x}}}|^{i}\\{} & {} \quad \lesssim \Vert {\textsf{f}}'-{\textsf{f}}''\Vert _{H^{4,3}(D_{\varepsilon })}+ |{\bar{\textbf{x}}}||(\Psi _1[{\textsf{f}}']-\Psi _1[{\textsf{f}}''])(\textbf{x},{\bar{\textbf{x}}})| +|{\bar{\textbf{x}}}|^2|(\Psi _2[{\textsf{f}}']-\Psi _2[{\textsf{f}}''])(\textbf{x},{\bar{\textbf{x}}})|. \end{aligned}$$

Finally, applying i) in each of the last two terms of the above expressions we have proved ii). \(\square \)

Proof of (iii)

Taking one derivative on (132), (133) and adding and subtracting appropriate terms, we obtain the expressions

$$\begin{aligned}&(\partial _{\textbf{x}}^2\Psi _1[{\textsf{f}}']-\partial _{\textbf{x}}^2\Psi _1[{\textsf{f}}'])(\textbf{x},{\bar{\textbf{x}}})\\&\quad =(\Psi _2[{\textsf{f}}']-\Psi _1^2[{\textsf{f}}''])(\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}^2({\textsf{f}}'-{\textsf{f}}'')](\textbf{x})\\&\qquad + (\Psi _1[{\textsf{f}}']-3\Psi _1[{\textsf{f}}'] \Psi _2[{\textsf{f}}']+2\Psi _1^3[{\textsf{f}}'])(\textbf{x},{\bar{\textbf{x}}}) \Delta _{{\bar{\textbf{x}}}}[\partial _x{\textsf{f}}'](\textbf{x})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}({\textsf{f}}'-{\textsf{f}}'')](\textbf{x})\\&\qquad + (\Psi _2[{\textsf{f}}']-\Psi _2[{\textsf{f}}''])(\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}^2{\textsf{f}}''](\textbf{x})\\&\qquad + (\partial _\textbf{x}\Psi _2[{\textsf{f}}']-\partial _{\textbf{x}}\Psi _2[{\textsf{f}}''])(\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}{\textsf{f}}''](\textbf{x})\\&\qquad - (\Psi _1[{\textsf{f}}']-\Psi _1[{\textsf{f}}''])(\Psi _1[{\textsf{f}}']+\Psi _1[{\textsf{f}}''])(\textbf{x},{\bar{\textbf{x}}}) \Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}^2{\textsf{f}}'](\textbf{x}),\\&\qquad - (\partial _{\textbf{x}}\Psi _1[{\textsf{f}}']-\partial _{\textbf{x}}\Psi _1[{\textsf{f}}''])(\Psi _1[{\textsf{f}}']+\Psi _1[{\textsf{f}}''])(\textbf{x},{\bar{\textbf{x}}}) \Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}{\textsf{f}}'](\textbf{x}),\\&\qquad - (\Psi _1[{\textsf{f}}']-\Psi _1[{\textsf{f}}''])(\Psi _2[{\textsf{f}}']-\Psi _1^2[{\textsf{f}}'])(\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}{\textsf{f}}'](\textbf{x}) \Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}{\textsf{f}}'](\textbf{x}),\\&\qquad - (\Psi _1[{\textsf{f}}']-\Psi _1[{\textsf{f}}''])(\Psi _2[{\textsf{f}}'']-\Psi _1^2[{\textsf{f}}''])(\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}{\textsf{f}}''](\textbf{x}) \Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}{\textsf{f}}'](\textbf{x}),\\ \end{aligned}$$

and

$$\begin{aligned}&(\partial _{\textbf{x}}^2\Psi _2[{\textsf{f}}']-\partial _{\textbf{x}}^2\Psi _2[{\textsf{f}}''])(\textbf{x},{\bar{\textbf{x}}})\\&\quad =(\Psi _1[{\textsf{f}}']-\Psi _1[{\textsf{f}}']\Psi _2[{\textsf{f}}'])(\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}^2({\textsf{f}}'-{\textsf{f}}'')](\textbf{x})\\&\qquad +(\Psi _2[{\textsf{f}}']-2\Psi _1^2[{\textsf{f}}'])(1-\Psi _2[{\textsf{f}}'])(\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}{\textsf{f}}'](\textbf{x})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}({\textsf{f}}'-{\textsf{f}}'')](\textbf{x})\\&\qquad + (\Psi _1[{\textsf{f}}']-\Psi _1[{\textsf{f}}''] )(\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}^2{\textsf{f}}''](\textbf{x})\\&\qquad + (\partial _{\textbf{x}}\Psi _1[{\textsf{f}}']-\partial _{\textbf{x}}\Psi _1[{\textsf{f}}''] )(\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}{\textsf{f}}''](\textbf{x})\\&\qquad - \Psi _1[{\textsf{f}}'](\Psi _2[{\textsf{f}}']-\Psi _2[{\textsf{f}}''])(\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}^2{\textsf{f}}''](\textbf{x})\\&\qquad - \Psi _1[{\textsf{f}}'](\partial _{\textbf{x}}\Psi _2[{\textsf{f}}']-\partial _{\textbf{x}}\Psi _2[{\textsf{f}}''] )(\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}{\textsf{f}}''](\textbf{x})\\&\qquad - (\Psi _2[{\textsf{f}}']-\Psi _1^2[{\textsf{f}}'])(\Psi _2[{\textsf{f}}']-\Psi _2[{\textsf{f}}''])(\textbf{x},{\bar{\textbf{x}}}) \Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}{\textsf{f}}'](\textbf{x}) \Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}{\textsf{f}}''](\textbf{x})\\&\qquad - \Psi _2[{\textsf{f}}''](\textbf{x},{\bar{\textbf{x}}})\left( \Psi _1[{\textsf{f}}']-\Psi _1[{\textsf{f}}''] \right) (\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}^2{\textsf{f}}''](\textbf{x})\\&\qquad - \Psi _2[{\textsf{f}}''](\textbf{x},{\bar{\textbf{x}}})(\partial _{\textbf{x}}\Psi _1[{\textsf{f}}']-\partial _{\textbf{x}}\Psi _1[{\textsf{f}}''] )(\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}{\textsf{f}}''](\textbf{x})\\&\qquad - (\Psi _1[{\textsf{f}}'']-\Psi _1[{\textsf{f}}'']\Psi _2[{\textsf{f}}''])\left( \Psi _1[{\textsf{f}}']-\Psi _1[{\textsf{f}}''] \right) (\textbf{x},{\bar{\textbf{x}}})\left( \Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}{\textsf{f}}''](\textbf{x})\right) ^2. \end{aligned}$$

The computations are very long and tedious but share lot of similarities. For this reason we shall focus only on \((\partial _{\textbf{x}}^2\Psi _2[{\textsf{f}}']-\partial _{\textbf{x}}^2\Psi _2[{\textsf{f}}''])(\textbf{x},{\bar{\textbf{x}}})\) to illustrate how the estimates work. The ideas is to write the above long expressions in groups of terms as

  • \((1-\Psi _2[{\textsf{f}}'])(\Psi _1[{\textsf{f}}'] \Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}^2({\textsf{f}}'-{\textsf{f}}'')](\textbf{x}) +(\Psi _2[{\textsf{f}}']-2\Psi _1^2[{\textsf{f}}'])\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}{\textsf{f}}'](\textbf{x})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}({\textsf{f}}'-{\textsf{f}}'')](\textbf{x})),\)

  • \((\Psi _1[{\textsf{f}}']-\Psi _1[{\textsf{f}}''] )(\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}^2{\textsf{f}}''](\textbf{x})- (\Psi _1[{\textsf{f}}'']-\Psi _1[{\textsf{f}}'']\Psi _2[{\textsf{f}}''])\left( \Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}{\textsf{f}}''](\textbf{x})\right) ^2 - \Psi _2[{\textsf{f}}''] \Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}^2{\textsf{f}}''](\textbf{x})),\)

  • \((\Psi _2[{\textsf{f}}']-\Psi _2[{\textsf{f}}''])\left( -\Psi _1[{\textsf{f}}']\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}^2{\textsf{f}}''](\textbf{x})- (\Psi _2[{\textsf{f}}']-\Psi _1^2[{\textsf{f}}'])\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}{\textsf{f}}'](\textbf{x}) \Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}{\textsf{f}}''](\textbf{x})\right) ,\)

  • \((\partial _{\textbf{x}}\Psi _1[{\textsf{f}}']-\partial _{\textbf{x}}\Psi _1[{\textsf{f}}''] )(1-\Psi _2[{\textsf{f}}''])\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}{\textsf{f}}''](\textbf{x}),\)

  • \((\partial _{\textbf{x}}\Psi _2[{\textsf{f}}']-\partial _{\textbf{x}}\Psi _2[{\textsf{f}}''] ) (- \Psi _1[{\textsf{f}}'] \Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}{\textsf{f}}''](\textbf{x})).\)

Remembering (131) and applying repeatedly Corollary 7.1, we get for \(1\le i \le 2 \) and \(|{\bar{\textbf{x}}}|\ll 1\) that

$$\begin{aligned} |(\partial _{\textbf{x}}^2\Psi _i[{\textsf{f}}']-\partial _{\textbf{x}}^2\Psi _i[{\textsf{f}}''])(\textbf{x},{\bar{\textbf{x}}})||{\bar{\textbf{x}}}|^{i}&\lesssim |{\bar{\textbf{x}}}|^{-1}\Vert {\textsf{f}}'-{\textsf{f}}''\Vert _{H^{4,3}(D_{\varepsilon })}+ |(\Psi _1[{\textsf{f}}']-\Psi _1[{\textsf{f}}''])(\textbf{x},{\bar{\textbf{x}}})|\\&\quad +|{\bar{\textbf{x}}}||(\Psi _2[{\textsf{f}}']-\Psi _2[{\textsf{f}}''])(\textbf{x},{\bar{\textbf{x}}})|\\&\quad +|{\bar{\textbf{x}}}||(\partial _{\textbf{x}}\Psi _1[{\textsf{f}}']-\partial _{\textbf{x}}\Psi _1[{\textsf{f}}''])(\textbf{x},{\bar{\textbf{x}}})|\\&\quad +|{\bar{\textbf{x}}}|^2|(\partial _{\textbf{x}}\Psi _2[{\textsf{f}}']-\partial _{\textbf{x}}\Psi _2[{\textsf{f}}''])(\textbf{x},{\bar{\textbf{x}}})|. \end{aligned}$$

Finally, applying i) and ii) in each of the terms of the above expressions we have proved iii). \(\square \)

At this point, we have all the ingredients to prove the main lemmas of the Appendix. The first one allows us to take derivatives into the kernel \(\partial _{\textbf{x}}^i K[g]\) as follows:

Lemma 7.4

Let \(g \in {\mathbb {B}}_{\delta }(H^{4,3}(D_{\varepsilon }))\) with \(0<\delta (\varepsilon )\ll 1\) small enough. The following bounds hold

  1. i)
    $$\begin{aligned} \sup _{\textbf{x}\in D_{\varepsilon }}\left( \sup _{{\bar{\textbf{x}}}\in D_\varepsilon (y)} |K[g](\textbf{x},{\bar{\textbf{x}}})|^2\right) \le C(\varepsilon ,\Vert g\Vert _{H^{4,3}(D_{\varepsilon })}), \end{aligned}$$
  2. ii)
    $$\begin{aligned} \sup _{\textbf{x}\in D_{\varepsilon }}\left( \int _{D_\varepsilon (y)}|\partial _{\textbf{x}} K[g](\textbf{x},{\bar{\textbf{x}}})|^2\hbox {d}{\bar{\textbf{x}}} \right) \le C(\varepsilon ,\Vert g\Vert _{H^{4,3}(D_{\varepsilon })}), \end{aligned}$$
  3. iii)
    $$\begin{aligned} \int _{D_{\varepsilon }}\left( \int _{D_\varepsilon (y)}\left| \partial _{\textbf{x}}^2 K[g](\textbf{x},{\bar{\textbf{x}}}) \right| ^2 |{\bar{\textbf{x}}}|^{2\gamma }\hbox {d}{\bar{\textbf{x}}} \right) \hbox {d}\textbf{x}\le C(\varepsilon ,\Vert g\Vert _{H^{4,3}(D_{\varepsilon })}), \end{aligned}$$
  4. iv)
    $$\begin{aligned} \int _{D_{\varepsilon }}\left( \int _{D_\varepsilon (y)}\left| \partial _{\textbf{x}}^3 K[g](\textbf{x},{\bar{\textbf{x}}}) \right| ^2 |{\bar{\textbf{x}}}|^{2}\hbox {d}{\bar{\textbf{x}}} \right) \hbox {d}\textbf{x}\le C(\varepsilon ,\Vert g\Vert _{H^{4,3}(D_{\varepsilon })}). \end{aligned}$$

Proof

As we can see the computations share lot of similarities. For this reason we shall focus only on one significant term iii) to illustrate how the estimates work. Remembering (129) we get

$$\begin{aligned} \int _{D_{\varepsilon }}\left( \int _{D_\varepsilon (y)}\left| \partial _{\textbf{x}}^2 K[g](\textbf{x},{\bar{\textbf{x}}}) \right| ^2 |{\bar{\textbf{x}}}|^{2\gamma }\hbox {d}{\bar{\textbf{x}}} \right) \hbox {d}\textbf{x}\le \sum _{i=1}^{2} \int _{D_{\varepsilon }}\left( \int _{D_\varepsilon (y)}\left| {\textsf{I}}_i[g](\textbf{x},{\bar{\textbf{x}}}) \right| ^2 |{\bar{\textbf{x}}}|^{2\gamma }\hbox {d}{\bar{\textbf{x}}} \right) \hbox {d}\textbf{x}\end{aligned}$$

Proceeding as usual, we split the integral into inner and outer regions. As \(g\in H^{4,3}(D_{\varepsilon })\subset C^{2}(D_{\varepsilon })\) is a continuous function on a bounded domain the outer integral is trivially bounded by some universal constant. To handle the remaining inner integral \(\{(\textbf{x},{\bar{\textbf{x}}})\in D_{\varepsilon }\times D_{\varepsilon }(y):|{\bar{\textbf{x}}}|\ll 1\}\) of each term, we use Corollary 7.2 as follows:

\({\textsf{I}}_1)\):

As \(g\in H^{4,3}(D_{\varepsilon })\) we have that \(\partial _{\textbf{x}}g \in H^{3,2}(D_{\varepsilon })\subset C^1(D_{\varepsilon })\) and consequently we get

$$\begin{aligned}&\int _{D_{\varepsilon }}\left( \int _{|{\bar{\textbf{x}}}|\ll 1}\left| {\textsf{I}}_1[g](\textbf{x},{\bar{\textbf{x}}}) \right| ^2 |{\bar{\textbf{x}}}|^{2\gamma }\hbox {d}{\bar{\textbf{x}}} \right) \hbox {d}\textbf{x}\\&\quad = \int _{D_{\varepsilon }}\left( \int _{|{\bar{\textbf{x}}}|\ll 1}\left| {\tilde{{\textsf{I}}}}_1[g](\textbf{x},{\bar{\textbf{x}}})\right| ^2 \left| \Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}g](\textbf{x})\right| ^4 |{\bar{\textbf{x}}}|^{2\gamma }\hbox {d}{\bar{\textbf{x}}} \right) \hbox {d}\textbf{x}\\&\quad \le C(\varepsilon ,\Vert g \Vert _{H^{4,3}(D_{\varepsilon })})\left( \int _{|{\bar{\textbf{x}}}|\ll 1} |{\bar{\textbf{x}}}|^{2\gamma }\hbox {d}{\bar{\textbf{x}}}\right) , \end{aligned}$$
\({\textsf{I}}_2)\):

As \(g\in H^{4,3}(D_{\varepsilon })\) we have that \(\partial _{\textbf{x}}^2g \in H^{2,1}(D_{\varepsilon })\subset C(D_{\varepsilon })\) and consequently we get

$$\begin{aligned}&\int _{D_{\varepsilon }}\left( \int _{|{\bar{\textbf{x}}}|\ll 1}\left| {\textsf{I}}_2[g](\textbf{x},{\bar{\textbf{x}}}) \right| ^2 |{\bar{\textbf{x}}}|^{2\gamma }\hbox {d}{\bar{\textbf{x}}} \right) \hbox {d}\textbf{x}\\&\quad = \int _{D_{\varepsilon }}\left( \int _{|{\bar{\textbf{x}}}|\ll 1}\left| {\tilde{{\textsf{I}}}}_2[g](\textbf{x},{\bar{\textbf{x}}})\right| ^2 \left| \Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}^2 g](\textbf{x})\right| ^2 |{\bar{\textbf{x}}}|^{2\gamma }\hbox {d}{\bar{\textbf{x}}} \right) \hbox {d}\textbf{x}\\&\le C(\delta ,\Vert g \Vert _{H^{4,3}(D_{\varepsilon })})\left( \int _{|{\bar{\textbf{x}}}|\ll 1} \frac{1}{|{\bar{\textbf{x}}}|^{2(1-\gamma )}}\hbox {d}{\bar{\textbf{x}}}\right) . \end{aligned}$$

As \(0<\gamma <1,\) the last term is integrable and we have proved our goal. The rest of the terms i), ii) and iv) work in a similar way and we omit the details. \(\square \)

Now, we continue with an analog result for the difference \(\partial _{\textbf{x}}^i K[{\textsf{f}}']-\partial _{\textbf{x}}^iK[{\textsf{f}}'']\).

Lemma 7.5

Let \({\textsf{f}}', {\textsf{f}}'' \in {\mathbb {B}}_{\delta }(H^{4,3}(D_{\varepsilon }))\) with \(0<\delta (\varepsilon )\ll 1\) small enough. The following bounds hold

  1. i)
    $$\begin{aligned} \sup _{\textbf{x}\in D_{\varepsilon }}\left( \sup _{{\bar{\textbf{x}}}\in D_\varepsilon (y)} |{\mathcal {K}}[{\textsf{f}}',{\textsf{f}}''](\textbf{x},{\bar{\textbf{x}}})|^2\right) \le C(\varepsilon ) \Vert {\textsf{f}}'-{\textsf{f}}''\Vert _{H^{4,3}(D_{\varepsilon })}^2. \end{aligned}$$
  2. ii)
    $$\begin{aligned} \sup _{\textbf{x}\in D_{\varepsilon }}\left( \int _{D_\varepsilon (y)}|\partial _{\textbf{x}}{\mathcal {K}}[{\textsf{f}}',{\textsf{f}}''](\textbf{x},{\bar{\textbf{x}}})|^2\hbox {d}{\bar{\textbf{x}}} \right) \le C(\varepsilon ) \Vert {\textsf{f}}'-{\textsf{f}}''\Vert _{H^{4,3}(D_{\varepsilon })}^2. \end{aligned}$$
  3. iii)
    $$\begin{aligned} \int _{D_{\varepsilon }}\left( \int _{D_\varepsilon (y)}\left| \partial _{\textbf{x}}^2 {\mathcal {K}}[{\textsf{f}}',{\textsf{f}}''](\textbf{x},{\bar{\textbf{x}}}) \right| ^2 |{\bar{\textbf{x}}}|^{2\gamma }\hbox {d}{\bar{\textbf{x}}} \right) \hbox {d}\textbf{x}\le C(\varepsilon ) \Vert {\textsf{f}}'-{\textsf{f}}''\Vert _{H^{4,3}(D_{\varepsilon })}^2. \end{aligned}$$
  4. iv)
    $$\begin{aligned} \int _{D_{\varepsilon }}\left( \int _{D_\varepsilon (y)}\left| \partial _{\textbf{x}}^3 {\mathcal {K}}[{\textsf{f}}',{\textsf{f}}''](\textbf{x},{\bar{\textbf{x}}}) \right| ^2 |{\bar{\textbf{x}}}|^{2}\hbox {d}{\bar{\textbf{x}}} \right) \hbox {d}\textbf{x}\le C(\varepsilon ) \Vert {\textsf{f}}'-{\textsf{f}}''\Vert _{H^{4,3}(D_{\varepsilon })}^2. \end{aligned}$$

Proof of i). We start by remembering the kernel definition \({\mathcal {K}}[{\textsf{f}}',{\textsf{f}}'']=K[{\textsf{f}}']-K[{\textsf{f}}'']\) given by

$$\begin{aligned} {\mathcal {K}}[{\textsf{f}}',{\textsf{f}}''](\textbf{x},{\bar{\textbf{x}}})=\log \left[ \frac{\cosh \left( {\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}'](\textbf{x})\right) -\cos ({\bar{x}})}{\cosh \left( {\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}''](\textbf{x})\right) -\cos ({\bar{x}})}\right] . \end{aligned}$$

Adding and subtracting some appropriate term, one finds

$$\begin{aligned} \frac{\cosh \left( {\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}'](\textbf{x})\right) -\cos ({\bar{x}})}{\cosh \left( {\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}''](\textbf{x})\right) -\cos ({\bar{x}})}=1+\frac{\cosh \left( {\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}'](\textbf{x})\right) -\cosh \left( {\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}''](\textbf{x})\right) }{\cosh \left( {\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}''](\textbf{x})\right) -\cos ({\bar{x}})}, \end{aligned}$$

and using the standard logarithmic inequality \(1-x^{-1}\le \log x\le x-1\) for all \(x>0\), we get the lower and upper bounds

$$\begin{aligned} {\mathcal {K}}[{\textsf{f}}',{\textsf{f}}''](\textbf{x},{\bar{\textbf{x}}})&\le \frac{\cosh \left( {\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}'](\textbf{x})\right) -\cosh \left( {\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}''](\textbf{x})\right) }{\cosh \left( {\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}''](\textbf{x})\right) -\cos ({\bar{x}})},\\ {\mathcal {K}}[{\textsf{f}}',{\textsf{f}}''](\textbf{x},{\bar{\textbf{x}}})&\ge \frac{\cosh \left( {\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}'](\textbf{x})\right) -\cosh \left( {\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}''](\textbf{x})\right) }{\cosh \left( {\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}'](\textbf{x})\right) -\cos ({\bar{x}})}. \end{aligned}$$

As \({\textsf{f}}', {\textsf{f}}'' \in {\mathbb {B}}_{\delta }(H^{4,3}(D_{\varepsilon }))\) are arbitrary functions we can assume without loss of generality that

$$\begin{aligned} |{\mathcal {K}}[{\textsf{f}}',{\textsf{f}}''](\textbf{x},{\bar{\textbf{x}}})|\le \left| \frac{\cosh \left( {\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}'](\textbf{x})\right) -\cosh \left( {\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}''](\textbf{x})\right) }{\cosh \left( {\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}''](\textbf{x})\right) -\cos ({\bar{x}})}\right| , \end{aligned}$$

and using the trigonometric identity \(\cosh (a+b)=\cosh (a)\cosh (b)+\sinh (a)\sinh (b)\) we finally get

$$\begin{aligned}&\frac{\cosh \left( {\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}'](\textbf{x})\right) -\cosh \left( {\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}''](\textbf{x})\right) }{\cosh \left( {\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}''](\textbf{x})\right) -\cos ({\bar{x}})}\\&\quad =\cosh ({\bar{y}})\frac{\cosh \left( \Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}'](\textbf{x})\right) -\cosh \left( \Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}''](\textbf{x})\right) }{\cosh \left( {\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}''](\textbf{x})\right) -\cos ({\bar{x}})}\\&\qquad +\sinh ({\bar{y}})\frac{\sinh \left( \Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}'](\textbf{x})\right) -\sinh \left( \Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}''](\textbf{x})\right) }{\cosh \left( {\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}''](\textbf{x})\right) -\cos ({\bar{x}})}. \end{aligned}$$

Now, we will use the Taylor expansion of the trigonometric functions \(\cosh (\cdot )\) and \(\sinh (\cdot )\) together with the algebraic identity \(a^n-b^n=(a-b)\sum _{k=0}^{n-1}a^k b^{n-1-k}\) to obtain

$$\begin{aligned}&\cosh \left( \Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}'](\textbf{x})\right) -\cosh \left( \Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}''](\textbf{x})\right) \\&\quad =\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}'-{\textsf{f}}''](\textbf{x})\sum _{n=1}^{\infty }\sum _{k=0}^{2n-1} \frac{\left( \Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}'](\textbf{x})\right) ^{k}\left( \Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}''](\textbf{x})\right) ^{2n-1-k}}{(2n)!},\\&\sinh \left( \Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}'](\textbf{x})\right) -\sinh \left( \Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}''](\textbf{x})\right) \\&\quad =\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}'-{\textsf{f}}''](\textbf{x})\sum _{n=0}^{\infty }\sum _{k=0}^{2n} \frac{\left( \Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}'](\textbf{x})\right) ^{k}\left( \Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}''](\textbf{x})\right) ^{2n-k}}{(2n+1)!}. \end{aligned}$$

Computing the infinite sum of the geometric series \(\sum _{n=0}^{\infty }r^{2n}\) with

$$\begin{aligned} r:=\max \{\Vert {\textsf{f}}'\Vert _{H^{4,3}(D_{\varepsilon })},\Vert {\textsf{f}}''\Vert _{H^{4,3}(D_{\varepsilon })}\}\text {diam}(D_{\varepsilon }), \end{aligned}$$

we finally get (under condition \(\delta <\text {diam}(D_{\varepsilon })^{-1}\)) that

$$\begin{aligned}&\left| \cosh \left( \Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}'](\textbf{x})\right) -\cosh \left( \Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}''](\textbf{x})\right) \right| \\&\quad \le C(\varepsilon ) |{\bar{\textbf{x}}}|^2 \frac{ \max \{\Vert {\textsf{f}}'\Vert _{H^{4,3}(D_{\varepsilon })},\Vert {\textsf{f}}''\Vert _{H^{4,3}(D_{\varepsilon })}\} \Vert {\textsf{f}}'-{\textsf{f}}''\Vert _{H^{4,3}(D_{\varepsilon })}}{1-\max \{\Vert {\textsf{f}}'\Vert _{H^{4,3}(D_{\varepsilon })},\Vert {\textsf{f}}''\Vert _{H^{4,3}(D_{\varepsilon })}\} \text {diam}(D_{\varepsilon })} ,\\&\left| \sinh \left( \Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}'](\textbf{x})\right) -\sinh \left( \Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}''](\textbf{x})\right) \right| \\&\quad \le C(\varepsilon ) |{\bar{\textbf{x}}}| \frac{ \Vert {\textsf{f}}'-{\textsf{f}}''\Vert _{H^{4,3}(D_{\varepsilon })}}{1-\max \{\Vert {\textsf{f}}'\Vert _{H^{4,3}(D_{\varepsilon })},\Vert {\textsf{f}}''\Vert _{H^{4,3}(D_{\varepsilon })}\}\text {diam}(D_{\varepsilon })}, \end{aligned}$$

where we have applied the Sobolev embedding \(L^{\infty }(D_{\varepsilon })\hookrightarrow H^2(D_{\varepsilon }).\)

Now, it is not difficult to see that the following uniform bound holds

$$\begin{aligned} \sup _{\textbf{x}\in D_{\varepsilon }}\left( \sup _{{\bar{\textbf{x}}}\in D_\varepsilon (y)} \left| \frac{\cosh ({\bar{y}})|{\bar{\textbf{x}}}|^2+\sinh ({\bar{y}})|{\bar{\textbf{x}}}|}{\cosh ({\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}''](\textbf{x}))-\cos ({\bar{x}})}\right| ^2\right) \le C(\varepsilon ) \end{aligned}$$

thanks to the fact (see (128)) that for \(|{\bar{\textbf{x}}}|\ll 1\) we have

$$\begin{aligned} |\cosh ({\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}''](\textbf{x}))-\cos ({\bar{x}})|\ge \left( \tfrac{1}{2}-\Vert {\textsf{f}}''\Vert _{H^{3}(D_{\varepsilon })}\right) |{\bar{\textbf{x}}}|^2-C(\Vert {\textsf{f}}''\Vert _{H^{3}(D_{\varepsilon })}) |{\bar{\textbf{x}}}|^4. \end{aligned}$$

Then, taking \(0<\delta < \text {diam}(D_{\varepsilon })^{-1} \) small enough and combining all we have proved our goal. \(\square \)

Proof of ii). As \({\mathcal {K}}[{\textsf{f}}',{\textsf{f}}''](\textbf{x},{\bar{\textbf{x}}})=( K[{\textsf{f}}']-K[{\textsf{f}}''])(\textbf{x},{\bar{\textbf{x}}})\) and \(\partial _{\textbf{x}}K[g](\textbf{x},{\bar{\textbf{x}}})=\Psi _1[g](\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}g](\textbf{x})\), adding and subtracting some appropriate term we obtain

$$\begin{aligned} \partial _{\textbf{x}}{\mathcal {K}}[{\textsf{f}}',{\textsf{f}}''](\textbf{x},{\bar{\textbf{x}}})=\Psi _1[{\textsf{f}}'](\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}({\textsf{f}}'-{\textsf{f}}'')](\textbf{x})+(\Psi _1[{\textsf{f}}']-\Psi _1[{\textsf{f}}''])(\textbf{x},{\bar{\textbf{x}}}) \Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}{\textsf{f}}''](\textbf{x}). \nonumber \\ \end{aligned}$$
(134)

To bound the first term we use \({\textsf{f}}',{\textsf{f}}''\in {\mathbb {B}}_{\delta }(H^{4,3}(D_{\varepsilon }))\), which implies \(\partial _{\textbf{x}}({\textsf{f}}'-{\textsf{f}}'')\in H^{3,2}(D_\varepsilon )\subset C^{1}(D_{\varepsilon })\). As usual, splitting the integral and using Corollary 7.1 in the inner region we obtain directly that

$$\begin{aligned} \int _{D_\varepsilon (y)}|\Psi _1[{\textsf{f}}'](\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}({\textsf{f}}'-{\textsf{f}}'')](\textbf{x})|^2\hbox {d}{\bar{\textbf{x}}} \le C(\varepsilon ,\Vert {\textsf{f}}'\Vert _{H^{4,3}(D_{\varepsilon })})\Vert {\textsf{f}}'-{\textsf{f}}''\Vert _{H^{4,3}(D_{\varepsilon })}^2. \end{aligned}$$

For the second term of (134), using that \(\partial _{\textbf{x}}{\textsf{f}}''\in H^{3,2}(D_{\varepsilon })\subset C^{1}(D_{\varepsilon })\) and applying Lemma 7.3 we get

$$\begin{aligned}&\int _{D_\varepsilon (y)}|(\Psi _1[{\textsf{f}}']-\Psi _1[{\textsf{f}}''])(\textbf{x},{\bar{\textbf{x}}}) \Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}{\textsf{f}}''](\textbf{x})|^2\hbox {d}{\bar{\textbf{x}}} \\&\quad \le \Vert {\textsf{f}}''\Vert _{H^{4,3}(D_{\varepsilon })} \int _{D_\varepsilon (y)}|(\Psi _1[{\textsf{f}}']-\Psi _1[{\textsf{f}}''])(\textbf{x},{\bar{\textbf{x}}})|^2 |{\bar{\textbf{x}}}|^2\hbox {d}{\bar{\textbf{x}}}\\&\quad \le C(\varepsilon ,\Vert {\textsf{f}}'\Vert _{H^{4,3}(D_{\varepsilon })},\Vert {\textsf{f}}''\Vert _{H^{4,3}(D_{\varepsilon })}) \Vert {\textsf{f}}'-{\textsf{f}}''\Vert _{H^{4,3}(D_{\varepsilon })}^2. \end{aligned}$$

Finally, as the above holds for any \(\textbf{x}\in D_{\varepsilon }\), combining the above and taking the supremum over all the domain we have proved the desired inequality. \(\square \)

Proof of (iii)

Taking a derivative of (134) we obtain

$$\begin{aligned} \partial _{\textbf{x}}^2{\mathcal {K}}[{\textsf{f}}',{\textsf{f}}''](\textbf{x},{\bar{\textbf{x}}})&=\partial _{\textbf{x}}\Psi _1[{\textsf{f}}'](\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}({\textsf{f}}'-{\textsf{f}}'')](\textbf{x})+\Psi _1[{\textsf{f}}'](\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}^2({\textsf{f}}'-{\textsf{f}}'')](\textbf{x}) \nonumber \\&\quad +(\partial _{\textbf{x}}\Psi _1[{\textsf{f}}']-\partial _{\textbf{x}}\Psi _1[{\textsf{f}}''])(\textbf{x},{\bar{\textbf{x}}}) \Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}{\textsf{f}}''](\textbf{x})\nonumber \\&\quad +(\Psi _1[{\textsf{f}}']-\Psi _1[{\textsf{f}}''])(\textbf{x},{\bar{\textbf{x}}}) \Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}^2{\textsf{f}}''](\textbf{x}). \end{aligned}$$
(135)

To bound the first term of (135), as \({\textsf{f}}',{\textsf{f}}''\in {\mathbb {B}}_{\delta }(H^{4,3}(D_{\varepsilon }))\) we get \(\partial _{\textbf{x}}{\textsf{f}}', \partial _{\textbf{x}}({\textsf{f}}'-{\textsf{f}}'')\in H^{3,2}(D_\varepsilon )\subset C^{1}(D_{\varepsilon })\). Now, splitting the integral and using (126) together with Corollary 7.1 in the inner region we obtain

$$\begin{aligned}{} & {} \int _{D_{\varepsilon }}\left( \int _{|{\bar{\textbf{x}}}|\ll 1}\left| \partial _{\textbf{x}}\Psi _1[{\textsf{f}}'](\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}({\textsf{f}}'-{\textsf{f}}'')](\textbf{x}) \right| ^2 |{\bar{\textbf{x}}}|^{2\gamma }\hbox {d}{\bar{\textbf{x}}} \right) \hbox {d}\textbf{x}\\{} & {} \quad \le \frac{C(\Vert {\textsf{f}}' \Vert _{H^{4,3}(D_{\varepsilon })})}{ 1-C(\Vert {\textsf{f}}' \Vert _{H^{4,3}(D_{\varepsilon })}) } \Vert {\textsf{f}}'-{\textsf{f}}'' \Vert _{H^{4,3}(D_{\varepsilon })}^2. \end{aligned}$$

To bound the second term of (135) we use the same type of ideas. As \({\textsf{f}}',{\textsf{f}}''\in {\mathbb {B}}_{\delta }(H^{4,3}(D_{\varepsilon }))\) we get \(\partial _{\textbf{x}}^2({\textsf{f}}'-{\textsf{f}}'')\in H^{2,1}(D_\varepsilon )\subset C(D_{\varepsilon })\). Now, splitting the integral and applying Corollary 7.1 in the inner region we obtain

$$\begin{aligned}{} & {} \int _{D_{\varepsilon }}\left( \int _{|{\bar{\textbf{x}}}|\ll 1}\left| \Psi _1[{\textsf{f}}'](\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}^2({\textsf{f}}'-{\textsf{f}}'')](\textbf{x}) \right| ^2 |{\bar{\textbf{x}}}|^{2\gamma }\hbox {d}{\bar{\textbf{x}}} \right) \hbox {d}\textbf{x}\\{} & {} \quad \le \frac{C(\Vert {\textsf{f}}' \Vert _{H^{4,3}(D_{\varepsilon })})}{ 1-C(\Vert {\textsf{f}}' \Vert _{H^{4,3}(D_{\varepsilon })}) } \Vert {\textsf{f}}'-{\textsf{f}}'' \Vert _{H^{4,3}(D_{\varepsilon })}^2\left( \int _{|{\bar{\textbf{x}}}|\ll 1}\frac{1}{|{\bar{\textbf{x}}}|^{2(1-\gamma )}}\hbox {d}{\bar{\textbf{x}}}\right) , \end{aligned}$$

where the last integral is bounded thanks to the fact that \(0<\gamma <1.\) Finally, to bound the last two terms of (135) we use the fact that \(\partial _{\textbf{x}}{\textsf{f}}'' \in H^{3,2}(D_{\varepsilon })\subset C^1(D_{\varepsilon })\) and \(\partial _{\textbf{x}}^2{\textsf{f}}'' \in H^{2,1}(D_{\varepsilon })\subset C(D_{\varepsilon })\) to get

$$\begin{aligned}{} & {} \int _{D_{\varepsilon }}\left( \int _{D_\varepsilon (y)}\left| (\partial _{\textbf{x}}\Psi _1[{\textsf{f}}']-\partial _{\textbf{x}}\Psi _1[{\textsf{f}}''])(\textbf{x},{\bar{\textbf{x}}}) \Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}{\textsf{f}}''](\textbf{x}) \right| ^2 |{\bar{\textbf{x}}}|^{2\gamma }\hbox {d}{\bar{\textbf{x}}} \right) \hbox {d}\textbf{x}\\{} & {} \quad \le \Vert {\textsf{f}}'' \Vert _{H^{4,3}(D_{\varepsilon })} \int _{D_{\varepsilon }}\left( \int _{D_\varepsilon (y)}\left| (\partial _{\textbf{x}}\Psi _1[{\textsf{f}}']-\partial _{\textbf{x}}\Psi _1[{\textsf{f}}''])(\textbf{x},{\bar{\textbf{x}}}) \right| ^2 |{\bar{\textbf{x}}}|^{2(1+\gamma )}\hbox {d}{\bar{\textbf{x}}} \right) \hbox {d}\textbf{x}\, \end{aligned}$$

and

$$\begin{aligned}{} & {} \int _{D_{\varepsilon }}\left( \int _{D_\varepsilon (y)}\left| (\Psi _1[{\textsf{f}}']-\Psi _1[{\textsf{f}}''])(\textbf{x},{\bar{\textbf{x}}}) \Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}^2{\textsf{f}}''](\textbf{x}) \right| ^2 |{\bar{\textbf{x}}}|^{2\gamma }\hbox {d}{\bar{\textbf{x}}} \right) \hbox {d}\textbf{x}\\{} & {} \quad \le \Vert {\textsf{f}}'' \Vert _{H^{4,3}(D_{\varepsilon })}\int _{D_{\varepsilon }}\left( \int _{D_\varepsilon (y)}\left| (\Psi _1[{\textsf{f}}']-\Psi _1[{\textsf{f}}''])(\textbf{x},{\bar{\textbf{x}}}) \right| ^2 |{\bar{\textbf{x}}}|^{2\gamma }\hbox {d}{\bar{\textbf{x}}} \right) \hbox {d}\textbf{x}. \end{aligned}$$

Using auxiliary Lemma 7.3 on the last terms of the above expressions we have proved our goal. \(\square \)

Proof of (iv)

As before, taking a derivative of (135) we obtain

$$\begin{aligned} \partial _{\textbf{x}}^3{\mathcal {K}}[{\textsf{f}}',{\textsf{f}}''](\textbf{x},{\bar{\textbf{x}}})&=\partial _{\textbf{x}}^2\Psi _1[{\textsf{f}}'](\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}({\textsf{f}}'-{\textsf{f}}'')](\textbf{x})+2\partial _{\textbf{x}}\Psi _1[{\textsf{f}}'](\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}^2({\textsf{f}}'-{\textsf{f}}'')](\textbf{x}) \nonumber \\&\quad +\Psi _1[{\textsf{f}}'](\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}^3({\textsf{f}}'-{\textsf{f}}'')](\textbf{x})\nonumber \\&\quad +2\left( \partial _{\textbf{x}}\Psi _1[{\textsf{f}}']-\partial _{\textbf{x}}\Psi _1[{\textsf{f}}'']\right) (\textbf{x},{\bar{\textbf{x}}}) \Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}^2{\textsf{f}}''](\textbf{x}) \nonumber \\&\quad +\left( \Psi _1[{\textsf{f}}']-\Psi _1[{\textsf{f}}'']\right) (\textbf{x},{\bar{\textbf{x}}}) \Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}^3{\textsf{f}}''](\textbf{x})\nonumber \\&\quad +\left( \partial _{\textbf{x}}^2\Psi _1[{\textsf{f}}']-\partial _{\textbf{x}}^2\Psi _1[{\textsf{f}}'']\right) (\textbf{x},{\bar{\textbf{x}}}) \Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}{\textsf{f}}''](\textbf{x}). \end{aligned}$$
(136)

In first place, due to the relations (126) and (127), we have

$$\begin{aligned} \partial _{\textbf{x}}^2\Psi _1[{\textsf{f}}']&=[\Psi _1[{\textsf{f}}'](1-\Psi _2[{\textsf{f}}'])-2\Psi _1[{\textsf{f}}'](\Psi _2[{\textsf{f}}']-\Psi _1^2[{\textsf{f}}'])]\left( \Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}{\textsf{f}}'](\textbf{x})\right) ^2\\&\quad +(\Psi _2[{\textsf{f}}']-\Psi _1^2[{\textsf{f}}'])\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}^2{\textsf{f}}']. \end{aligned}$$

Remembering (131) and applying repeatedly Corollary 7.1, we get that the first three terms of (136) can be handled in the following way:

$$\begin{aligned}&\sum _{i=1}^{3}\int _{D_{\varepsilon }}\left( \int _{D_\varepsilon (y)}\left| \partial _{\textbf{x}}^{3-i}\Psi _1[{\textsf{f}}'](\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}^{i}({\textsf{f}}'-{\textsf{f}}'')](\textbf{x}) \right| ^2 |{\bar{\textbf{x}}}|^{2}\hbox {d}{\bar{\textbf{x}}} \right) \hbox {d}\textbf{x}\\&\quad \le C(\varepsilon ) \Vert {\textsf{f}}'-{\textsf{f}}''\Vert _{H^{4,3}(D_{\varepsilon })}^2. \end{aligned}$$

The remaining three terms of (136) follows by auxiliary Lemma 7.3. Notice that using (131) we get

$$\begin{aligned}{} & {} \int _{D_{\varepsilon }}\left( \int _{D_\varepsilon (y)}\left| (\Psi _1[{\textsf{f}}']-\Psi _1[{\textsf{f}}''])(\textbf{x},{\bar{\textbf{x}}}) \Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}^3{\textsf{f}}''](\textbf{x}) \right| ^2 |{\bar{\textbf{x}}}|^{2}\hbox {d}{\bar{\textbf{x}}} \right) \hbox {d}\textbf{x}\\{} & {} \quad \le \Vert {\textsf{f}}''\Vert _{H^{4,3}(D_{\varepsilon })} \int _{D_{\varepsilon }}\left( \sup _{{\bar{\textbf{x}}}\in D_\varepsilon (y)}\left| (\Psi _1[{\textsf{f}}']-\Psi _1[{\textsf{f}}''])(\textbf{x},{\bar{\textbf{x}}}) \right| ^2 |{\bar{\textbf{x}}}|^{2} \right) \hbox {d}\textbf{x},\\{} & {} \int _{D_{\varepsilon }}\left( \int _{D_\varepsilon (y)}\left| (\partial _{\textbf{x}}\Psi _1[{\textsf{f}}']-\partial _{\textbf{x}}\Psi _1[{\textsf{f}}''])(\textbf{x},{\bar{\textbf{x}}}) \Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}^2{\textsf{f}}''](\textbf{x}) \right| ^2 |{\bar{\textbf{x}}}|^{2}\hbox {d}{\bar{\textbf{x}}} \right) \hbox {d}\textbf{x}\\{} & {} \quad \le \Vert {\textsf{f}}''\Vert _{H^{4,3}(D_{\varepsilon })} \int _{D_{\varepsilon }}\left( \int _{D_\varepsilon (y)}\left| (\partial _{\textbf{x}}\Psi _1[{\textsf{f}}']-\partial _{\textbf{x}}\Psi _1[{\textsf{f}}''])(\textbf{x},{\bar{\textbf{x}}}) \right| ^2 |{\bar{\textbf{x}}}|^{2}\hbox {d}{\bar{\textbf{x}}} \right) \hbox {d}\textbf{x}, \end{aligned}$$

and

$$\begin{aligned}{} & {} \int _{D_{\varepsilon }}\left( \int _{D_\varepsilon (y)}\left| (\partial _{\textbf{x}}^2\Psi _1[{\textsf{f}}']-\partial _{\textbf{x}}^2\Psi _1[{\textsf{f}}''])(\textbf{x},{\bar{\textbf{x}}}) \Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}{\textsf{f}}''](\textbf{x}) \right| ^2 |{\bar{\textbf{x}}}|^{2}\hbox {d}{\bar{\textbf{x}}} \right) \hbox {d}\textbf{x}\\{} & {} \quad \le \Vert {\textsf{f}}''\Vert _{H^{4,3}(D_{\varepsilon })} \int _{D_{\varepsilon }}\left( \int _{D_\varepsilon (y)}\left| (\partial _{\textbf{x}}^2\Psi _1[{\textsf{f}}']-\partial _{\textbf{x}}^2\Psi _1[{\textsf{f}}''])(\textbf{x},{\bar{\textbf{x}}}) \right| ^2 |{\bar{\textbf{x}}}|^{4}\hbox {d}{\bar{\textbf{x}}} \right) \hbox {d}\textbf{x}. \end{aligned}$$

Using auxiliary Lemma 7.3 on the last terms of the above expressions we have proved our goal. \(\square \)

Lemma 7.6

Let \({\textsf{f}}\in {\mathbb {B}}_{\delta }(H^{4,3}(D_{\varepsilon }))\) with \(0<\delta (\varepsilon )\ll 1\) small enough and \(h\in H^{4,3}(D_{\varepsilon })\) with \(\Vert h\Vert _{H^{4,3}(D_{\varepsilon })}=1.\) For \(0<\tau \ll 1\), the following bound holds

  1. i)
    $$\begin{aligned} \sup _{\textbf{x}\in D_{\varepsilon }}\left( \sup _{{\bar{\textbf{x}}}\in D_{\varepsilon }(y)}\left| \frac{{\mathcal {K}}[{\textsf{f}}+\tau h,{\textsf{f}}](\textbf{x},{\bar{\textbf{x}}})}{\tau }-\Psi _1[{\textsf{f}}](\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[h](\textbf{x})\right| ^2 \right) \le C(\varepsilon )\tau ^2. \end{aligned}$$
  2. ii)
    $$\begin{aligned} \sup _{\textbf{x}\in D_{\varepsilon }}\left( \int _{D_{\varepsilon }(y)}\left| \partial _{\textbf{x}}\left\{ \frac{{\mathcal {K}}[{\textsf{f}}+\tau h,{\textsf{f}}](\textbf{x},{\bar{\textbf{x}}})}{\tau }-\Psi _1[{\textsf{f}}](\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[h](\textbf{x})\right\} \right| ^2\hbox {d}{\bar{\textbf{x}}} \right) \le C(\varepsilon )\tau ^2. \end{aligned}$$
  3. iii)
    $$\begin{aligned} \int _{D_{\varepsilon }}\left( \int _{D_\varepsilon (y)}\left| \partial _{\textbf{x}}^2\left\{ \frac{{\mathcal {K}}[{\textsf{f}}+\tau h,{\textsf{f}}](\textbf{x},{\bar{\textbf{x}}})}{\tau }-\Psi _1[{\textsf{f}}](\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[h](\textbf{x})\right\} \right| ^2 |{\bar{\textbf{x}}}|^{2\gamma }\hbox {d}{\bar{\textbf{x}}} \right) \hbox {d}\textbf{x}\le C(\varepsilon )\tau ^2. \end{aligned}$$
  4. iv)
    $$\begin{aligned} \int _{D_{\varepsilon }}\left( \int _{D_\varepsilon (y)}\left| \partial _{\textbf{x}}^3\left\{ \frac{{\mathcal {K}}[{\textsf{f}}+\tau h,{\textsf{f}}](\textbf{x},{\bar{\textbf{x}}})}{\tau }-\Psi _1[{\textsf{f}}](\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[h](\textbf{x})\right\} \right| ^2 |{\bar{\textbf{x}}}|^{2}\hbox {d}{\bar{\textbf{x}}} \right) \hbox {d}\textbf{x}\le C(\varepsilon )\tau ^2. \end{aligned}$$

Proof of (i)

Just applying the standard logarithmic inequality \(1-x^{-1}\le \log x\le x-1\) for all \(x>0\), we get the lower and upper bounds

$$\begin{aligned} {\mathcal {K}}[{\textsf{f}}+\tau h,{\textsf{f}}](\textbf{x},{\bar{\textbf{x}}})&\le \frac{\cosh \left( {\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}+\tau h](\textbf{x})\right) -\cosh \left( {\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}](\textbf{x})\right) }{\cosh \left( {\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}](\textbf{x})\right) -\cos ({\bar{x}})},\\ {\mathcal {K}}[{\textsf{f}}+\tau h,{\textsf{f}}](\textbf{x},{\bar{\textbf{x}}})&\ge \frac{\cosh \left( {\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}+\tau h](\textbf{x})\right) -\cosh \left( {\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}](\textbf{x})\right) }{\cosh \left( {\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}+\tau h](\textbf{x})\right) -\cos ({\bar{x}})}. \end{aligned}$$

Therefore, we have

$$\begin{aligned}{} & {} \left| \frac{{\mathcal {K}}[{\textsf{f}}+\tau h,{\textsf{f}}](\textbf{x},{\bar{\textbf{x}}})}{\tau }-\Psi _1[{\textsf{f}}](\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[h](\textbf{x})\right| \\{} & {} \quad \le \max _{0\le \xi \le \tau }\left| \frac{1}{\tau }\left( \frac{\cosh \left( {\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}+\tau h](\textbf{x})\right) -\cosh \left( {\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}](\textbf{x})\right) }{\cosh \left( {\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}+\xi h](\textbf{x})\right) -\cos ({\bar{x}})}\right) -\Psi _1[{\textsf{f}}](\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[h](\textbf{x})\right| . \end{aligned}$$

Now, using the trigonometric identity \(\cosh (a+b)=\cosh (a)\cosh (b)+\sinh (a)\sinh (b)\) on the above expression, we get

$$\begin{aligned}{} & {} \frac{\cosh \left( {\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}+\tau h](\textbf{x})\right) -\cosh \left( {\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}](\textbf{x})\right) }{\cosh \left( {\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}+\xi h](\textbf{x})\right) -\cos ({\bar{x}})}\\{} & {} \quad =\cosh ({\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}](x))\frac{\cosh \left( \Delta _{{\bar{\textbf{x}}}}[\tau h](\textbf{x})\right) -1}{\cosh \left( {\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}+\xi h](\textbf{x})\right) -\cos ({\bar{x}})}\\{} & {} \qquad +\sinh ({\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}](x))\frac{\sinh \left( \Delta _{{\bar{\textbf{x}}}}[\tau h](\textbf{x})\right) }{\cosh \left( {\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}+\xi h](\textbf{x})\right) -\cos ({\bar{x}})}, \end{aligned}$$

and consequently

$$\begin{aligned}{} & {} \left| \frac{{\mathcal {K}}[{\textsf{f}}+\tau h,{\textsf{f}}](\textbf{x},{\bar{\textbf{x}}})}{\tau }-\Psi _1[{\textsf{f}}](\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[h](\textbf{x})\right| \\{} & {} \quad \le \left| \frac{\cosh \left( \Delta _{{\bar{\textbf{x}}}}[\tau h](\textbf{x})\right) -1}{\tau }\right| \max _{0\le \xi \le \tau }\left| \frac{\cosh ({\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}](x))}{\cosh \left( {\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}+\xi h](\textbf{x})\right) -\cos ({\bar{x}})}\right| \\{} & {} \qquad +\left| \frac{\sinh \left( \Delta _{{\bar{\textbf{x}}}}[\tau h](\textbf{x})\right) }{\tau }-\Delta _{{\bar{\textbf{x}}}}[h](\textbf{x})\right| \max _{0\le \xi \le \tau }\left| \frac{\sinh ({\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}](x))}{\cosh \left( {\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}+\xi h](\textbf{x})\right) -\cos ({\bar{x}})}\right| . \end{aligned}$$

Applying the Taylor expansion of the trigonometric functions \(\sinh (\cdot ), \cosh (\cdot )\) together with the fact that \(\Vert h\Vert _{H^3(D_{\varepsilon })}=1\) we get the bounds

$$\begin{aligned} \left| \frac{\cosh \left( \Delta _{{\bar{\textbf{x}}}}[\tau h](\textbf{x})\right) -1}{\tau }\right| \le C(\varepsilon )\tau |{\bar{\textbf{x}}}|^2,\\ \left| \frac{\sinh \left( \Delta _{{\bar{\textbf{x}}}}[\tau h](\textbf{x})\right) }{\tau }-\Delta _{{\bar{\textbf{x}}}}[h](\textbf{x})\right| \le C(\varepsilon )\tau ^2 |{\bar{\textbf{x}}}|^3. \end{aligned}$$

In addition, we have

$$\begin{aligned}{} & {} \max \left\{ \left| \frac{\cosh ({\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}](x))}{\cosh \left( {\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}+\xi h](\textbf{x})\right) -\cos ({\bar{x}})}\right| ,\left| \frac{\sinh ({\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}](x))}{\cosh \left( {\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}+\xi h](\textbf{x})\right) -\cos ({\bar{x}})}\right| \right\} \\{} & {} \quad \le \frac{C(\Vert {\textsf{f}}\Vert _{H^{3}(D_{\varepsilon })})}{\left( \tfrac{1}{2}-\Vert {\textsf{f}}+\xi h\Vert _{H^{3}(D_{\varepsilon })}\right) |{\bar{\textbf{x}}}|^2-C(\Vert {\textsf{f}}+\xi h\Vert _{H^{3}(D_{\varepsilon })}) |{\bar{\textbf{x}}}|^4}, \end{aligned}$$

which give us

$$\begin{aligned}{} & {} \left| \frac{{\mathcal {K}}[{\textsf{f}}+\tau h,{\textsf{f}}](\textbf{x},{\bar{\textbf{x}}})}{\tau }-\Psi _1[{\textsf{f}}](\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[h](\textbf{x})\right| \\{} & {} \quad \le \max _{0\le \xi \le \tau }\frac{C(\varepsilon ,\Vert {\textsf{f}}\Vert _{H^{3}(D_{\varepsilon })}) \tau }{\left( \tfrac{1}{2}-\Vert {\textsf{f}}+\xi h\Vert _{H^{3}(D_{\varepsilon })}\right) -C(\Vert {\textsf{f}}+\xi h\Vert _{H^{3}(D_{\varepsilon })}) |{\bar{\textbf{x}}}|^2}. \end{aligned}$$

Coming back to the starting point, as \({\textsf{f}}\in {\mathbb {B}}_{\delta }(X(D_{\varepsilon }))\) and \(\Vert h\Vert _{H^{4,3}(D_{\varepsilon })}=1,\) taking \(0<\delta (\varepsilon ),\tau \ll 1\) small enough we have proved our desired inequality. \(\square \)

Proof of (ii)

As \({\mathcal {K}}({\textsf{f}}',{\textsf{f}}'')(\textbf{x},{\bar{\textbf{x}}})=\left( K[{\textsf{f}}']-K[{\textsf{f}}'']\right) (\textbf{x},{\bar{\textbf{x}}})\) and \(\partial _{\textbf{x}}K[g](\textbf{x},{\bar{\textbf{x}}})=\Psi _1[g](\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}g](\textbf{x})\), adding and subtracting some appropriate term we obtain

$$\begin{aligned} \partial _{\textbf{x}}{\mathcal {K}}[{\textsf{f}}',{\textsf{f}}''](\textbf{x},{\bar{\textbf{x}}})=\Psi _1[{\textsf{f}}'](\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}({\textsf{f}}'-{\textsf{f}}'')](\textbf{x})+\left( \Psi _1[{\textsf{f}}']-\Psi _1[{\textsf{f}}'']\right) (\textbf{x},{\bar{\textbf{x}}}) \Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}{\textsf{f}}''](\textbf{x}). \end{aligned}$$

So, it is just a matter of algebra to get

$$\begin{aligned} \partial _{\textbf{x}}\left\{ \frac{{\mathcal {K}}[{\textsf{f}}+\tau h,{\textsf{f}}](\textbf{x},{\bar{\textbf{x}}})}{\tau }-\Psi _1[{\textsf{f}}](\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[h](\textbf{x})\right\} = {\textsf{A}}[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}})+{\textsf{B}}[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}}), \nonumber \\ \end{aligned}$$
(137)

where both new functionals are given respectively by

$$\begin{aligned} {\textsf{A}}[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}})&:=\left[ \frac{(\Psi _1[{\textsf{f}}+\tau h]-\Psi _1[{\textsf{f}}])(\textbf{x},{\bar{\textbf{x}}})}{\tau }-(\Psi _2[{\textsf{f}}]-\Psi _1^2[{\textsf{f}}])(\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[h](\textbf{x})\right] \Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}{\textsf{f}}](\textbf{x}),\nonumber \\ {\textsf{B}}[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}})&:=(\Psi _1[{\textsf{f}}+\tau h]-\Psi _1[{\textsf{f}}])(\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}h](\textbf{x}). \end{aligned}$$
(138)

Directly from auxiliary Lemma 7.3, as \(h\in H^{4,3}(D_{\varepsilon })\) implies \(\partial _{\textbf{x}}h\in H^{3,2}(D_{\varepsilon })\subset C^{1}(D_{\varepsilon })\) and by hypothesis \(\Vert h\Vert _{H^{4,3}(D_{\varepsilon })}=1,\) we get

$$\begin{aligned} \sup _{\textbf{x}\in D_{\varepsilon }} \int _{D_{\varepsilon }(y)}\left| {\textsf{B}}[{\textsf{f}},h]\right| ^2\hbox {d}{\bar{\textbf{x}}}\le C(\varepsilon )\tau ^2. \end{aligned}$$

To obtain something similar for \({\textsf{A}}[{\textsf{f}},h]\) we need to work a little bit more. In first place, thanks to the definition of \(\Psi _1[\cdot ](\textbf{x},{\bar{\textbf{x}}})\), adding and subtracting some appropriate term we have

$$\begin{aligned}{} & {} \left( \Psi _1[{\textsf{f}}+\tau h]-\Psi _1[{\textsf{f}}]\right) (\textbf{x},{\bar{\textbf{x}}})\\{} & {} \quad =\frac{\sinh ({\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}+\tau h](\textbf{x}))-\sinh ({\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}](\textbf{x}))}{\cosh ({\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}](\textbf{x}))-\cos ({\bar{x}})} -\sinh ({\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}+\tau h](\textbf{x}))\\{} & {} \qquad \frac{\cosh ({\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}+\tau h](\textbf{x}))-\cosh ({\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}](\textbf{x}))}{\left[ \cosh ({\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}+\tau h](\textbf{x}))-\cos ({\bar{x}})\right] \left[ \cosh ({\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}](\textbf{x}))-\cos ({\bar{x}})\right] }, \end{aligned}$$

and using on it some trigonometric identities, we obtain

$$\begin{aligned}{} & {} \left( \Psi _1[{\textsf{f}}+\tau h]-\Psi _1[{\textsf{f}}]\right) (\textbf{x},{\bar{\textbf{x}}})\\{} & {} \quad =\left[ \Psi _2[{\textsf{f}}](\textbf{x},{\bar{\textbf{x}}})-\Psi _1^2[{\textsf{f}}](\textbf{x},{\bar{\textbf{x}}})\cosh \left( \tau \Delta _{{\bar{\textbf{x}}}}[h](\textbf{x})\right) \right] \sinh \left( \tau \Delta _{{\bar{\textbf{x}}}}[h](\textbf{x})\right) +\Psi _1^2[{\textsf{f}}](\textbf{x},{\bar{\textbf{x}}})\\{} & {} \qquad \frac{\cosh ({\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}+\tau h](\textbf{x})-\cosh ({\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}](\textbf{x})}{\cosh ({\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}+\tau h](\textbf{x})-\cos ({\bar{x}})}\sinh \left( \tau \Delta _{{\bar{\textbf{x}}}}[h](\textbf{x})\right) \cosh \left( \tau \Delta _{{\bar{\textbf{x}}}}[h](\textbf{x})\right) . \end{aligned}$$

Now, dividing the above expression by \(\tau \) and subtracting the term \((\Psi _2[{\textsf{f}}]-\Psi _1^2[{\textsf{f}}])(\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[h](\textbf{x})\) we obtain that

$$\begin{aligned}{} & {} \frac{(\Psi _1[{\textsf{f}}+\tau h]-\Psi _1[{\textsf{f}}])(\textbf{x},{\bar{\textbf{x}}})}{\tau }-(\Psi _2[{\textsf{f}}]-\Psi _1^2[{\textsf{f}}])(\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[h](\textbf{x})\\{} & {} \quad =\Psi _2[{\textsf{f}}](\textbf{x},{\bar{\textbf{x}}})\left( \frac{\sinh (\tau \Delta _{{\bar{\textbf{x}}}}[h](\textbf{x}))}{\tau } - \Delta _{{\bar{\textbf{x}}}}[h](\textbf{x})\right) \\{} & {} \qquad -\Psi _1^2[{\textsf{f}}](\textbf{x},{\bar{\textbf{x}}})\left( \cosh (\tau \Delta _{{\bar{\textbf{x}}}}[h](\textbf{x}))\frac{\sinh (\tau \Delta _{{\bar{\textbf{x}}}}[h](\textbf{x}))}{\tau } - \Delta _{{\bar{\textbf{x}}}}[h](\textbf{x})\right) \\{} & {} \qquad +\Psi _1^2[{\textsf{f}}](\textbf{x},{\bar{\textbf{x}}}) \frac{\cosh ({\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}+\tau h](\textbf{x})-\cosh ({\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}](\textbf{x})}{\cosh ({\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}+\tau h](\textbf{x})-\cos ({\bar{x}})}\\{} & {} \qquad \sinh \left( \tau \Delta _{{\bar{\textbf{x}}}}[h](\textbf{x})\right) \cosh \left( \tau \Delta _{{\bar{\textbf{x}}}}[h](\textbf{x})\right) . \end{aligned}$$

Finally, using the Taylor expansion of the trigonometric functions \(\sinh (\cdot ), \cosh (\cdot )\) and (131) give us the desired inequality

$$\begin{aligned} \sup _{\textbf{x}\in D_{\varepsilon }} \int _{D_{\varepsilon }(y)}\left| {\textsf{A}}[{\textsf{f}},h]\right| ^2\hbox {d}{\bar{\textbf{x}}}\le C(\varepsilon )\tau ^2. \end{aligned}$$
(139)

\(\square \)

Proof of (iii)

Recalling (137), we have

$$\begin{aligned} \partial _{\textbf{x}}^2\left\{ \frac{{\mathcal {K}}[{\textsf{f}}+\tau h,{\textsf{f}}](\textbf{x},{\bar{\textbf{x}}})}{\tau }-\Psi _1[{\textsf{f}}](\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[h](\textbf{x})\right\} = \partial _{\textbf{x}}{\textsf{A}}[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}})+\partial _{\textbf{x}}{\textsf{B}}[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}}), \nonumber \\ \end{aligned}$$
(140)

where, using relation \(\partial _{\textbf{x}}\Psi _1[g](\textbf{x},{\bar{\textbf{x}}})={\tilde{{\textsf{I}}}}_1[g](\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}g](\textbf{x})\) whit \({\tilde{{\textsf{I}}}}_1[g]=\left( \Psi _2[g]\right. \left. -(\Psi _1[g])^2\right) \), we get

$$\begin{aligned} \partial _{\textbf{x}}{\textsf{A}}[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}})= & {} ({\tilde{{\textsf{I}}}}_1[{\textsf{f}}+\tau h]-{\tilde{{\textsf{I}}}}_1[{\textsf{f}}])(\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}{\textsf{f}}](\textbf{x})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}h](\textbf{x})\\{} & {} \quad +\frac{(\Psi _1[{\textsf{f}}+\tau h]-\Psi _1[{\textsf{f}}])(\textbf{x},{\bar{\textbf{x}}})}{\tau }\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}^2{\textsf{f}}](\textbf{x})\\{} & {} \quad -{\tilde{{\textsf{I}}}}_1[{\textsf{f}}](\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}^2{\textsf{f}}]\Delta _{{\bar{\textbf{x}}}}[h](\textbf{x})\\{} & {} \quad +\frac{({\tilde{{\textsf{I}}}}_1[{\textsf{f}}+\tau h]-{\tilde{{\textsf{I}}}}_1[{\textsf{f}}])(\textbf{x},{\bar{\textbf{x}}})}{\tau }\left( \Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}{\textsf{f}}](\textbf{x})\right) ^2\\{} & {} \quad -\left( \partial _{\textbf{x}}\Psi _2[{\textsf{f}}]-2\Psi _1[{\textsf{f}}]\partial _{\textbf{x}}\Psi _1[{\textsf{f}}]\right) (\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}{\textsf{f}}](\textbf{x})\Delta _{{\bar{\textbf{x}}}}[h](\textbf{x}), \end{aligned}$$

and

$$\begin{aligned} \partial _{\textbf{x}}{\textsf{B}}[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}})= & {} ({\tilde{{\textsf{I}}}}_1[{\textsf{f}}+\tau h]-{\tilde{{\textsf{I}}}}_1[{\textsf{f}}])(\textbf{x},{\bar{\textbf{x}}}) \Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}{\textsf{f}}](\textbf{x})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}h](\textbf{x})\\{} & {} +(\Psi _1[{\textsf{f}}+\tau h]-\Psi _1[{\textsf{f}}])(\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}^2 h](\textbf{x})\\{} & {} + \tau {\tilde{{\textsf{I}}}}_1[{\textsf{f}}+\tau h](\textbf{x},{\bar{\textbf{x}}})\left( \Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}} h](\textbf{x})\right) ^2. \end{aligned}$$

By definition of \({\tilde{{\textsf{I}}}}_1[\cdot ]\) and relations (126) and (127), the previous expressions can be rewritten in a more manageable way as

$$\begin{aligned} \partial _{\textbf{x}}{\textsf{A}}[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}})&=({\textsf{A}}_1'+{\textsf{A}}_2'+{\textsf{A}}_3'+{\textsf{A}}_4'+{\textsf{A}}_5')[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}}), \end{aligned}$$
(141)
$$\begin{aligned} \partial _{\textbf{x}}{\textsf{B}}[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}})&=({\textsf{B}}_1'+{\textsf{B}}_2'+{\textsf{B}}_3'+{\textsf{B}}_4')[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}}), \end{aligned}$$
(142)

with

$$\begin{aligned} {\textsf{A}}_1'[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}})&:=\left[ \frac{(\Psi _1[{\textsf{f}}+\tau h]-\Psi _1[{\textsf{f}}])(\textbf{x},{\bar{\textbf{x}}})}{\tau }-(\Psi _2[{\textsf{f}}]-\Psi _1^2[{\textsf{f}}])(\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[h](\textbf{x})\right] \\&\qquad \Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}^2{\textsf{f}}](\textbf{x}),\\ {\textsf{A}}_2'[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}})&:=(\Psi _2[{\textsf{f}}+\tau h]-\Psi _2[{\textsf{f}}])(\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}{\textsf{f}}](\textbf{x})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}h](\textbf{x}),\\ {\textsf{A}}_3'[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}})&:=-(\Psi _1[{\textsf{f}}+\tau h]-\Psi _1[{\textsf{f}}])(\Psi _1[{\textsf{f}}+\tau h]+\Psi _1[{\textsf{f}}])(\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}{\textsf{f}}](\textbf{x})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}h](\textbf{x}),\\ {\textsf{A}}_4'[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}})&:=\left[ \frac{(\Psi _2[{\textsf{f}}+\tau h]-\Psi _2[{\textsf{f}}])(\textbf{x},{\bar{\textbf{x}}})}{\tau } -\left( \Psi _1[{\textsf{f}}]-\Psi _1[{\textsf{f}}]\Psi _2[{\textsf{f}}]\right) (\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[h](\textbf{x})\right] \\&\qquad \left( \Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}{\textsf{f}}](\textbf{x})\right) ^2,\\ {\textsf{A}}_5'[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}})&:=\bigg [-(\Psi _1[{\textsf{f}}+\tau h]+\Psi _1[{\textsf{f}}])(\textbf{x},{\bar{\textbf{x}}})\frac{(\Psi _1[{\textsf{f}}+\tau h]-\Psi _1[{\textsf{f}}])(\textbf{x},{\bar{\textbf{x}}})}{\tau }\\&\qquad +2\Psi _1[{\textsf{f}}](\textbf{x},{\bar{\textbf{x}}}) (\Psi _2[{\textsf{f}}]-\Psi _1^2[{\textsf{f}}])(\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[h](\textbf{x})\bigg ]\left( \Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}{\textsf{f}}](\textbf{x})\right) ^2, \end{aligned}$$

and

$$\begin{aligned} {\textsf{B}}_1'[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}})&:=(\Psi _2[{\textsf{f}}+\tau h]-\Psi _2[{\textsf{f}}])(\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}{\textsf{f}}](\textbf{x})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}h](\textbf{x}),\\ {\textsf{B}}_2'[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}})&:=-(\Psi _1[{\textsf{f}}+\tau h]-\Psi _1[{\textsf{f}}])(\Psi _1[{\textsf{f}}+\tau h]\\&\qquad +\Psi _1[{\textsf{f}}])(\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}{\textsf{f}}](\textbf{x})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}h](\textbf{x}),\\ {\textsf{B}}_3'[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}})&:= \tau (\Psi _2[{\textsf{f}}+\tau h]-\Psi _1^2[{\textsf{f}}+\tau h])(\textbf{x},{\bar{\textbf{x}}})\left( \Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}} h](\textbf{x})\right) ^2,\\ {\textsf{B}}_4'[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}})&:=(\Psi _1[{\textsf{f}}+\tau h]-\Psi _1[{\textsf{f}}])(\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}^2 h](\textbf{x}). \end{aligned}$$

Note that the only new type of term that appear is \({\textsf{A}}_4'[{\textsf{f}},h].\) The rest of terms can be easily handle using auxiliary Lemma 7.3, Corollary 7.1 and (131). To sum up, our proof reduces to check that

$$\begin{aligned} \int _{D_{\varepsilon }}\left( \int _{D_\varepsilon (y)}\left| {\textsf{A}}_4'[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}})\right| ^2 |{\bar{\textbf{x}}}|^{2\gamma }\hbox {d}{\bar{\textbf{x}}} \right) \hbox {d}\textbf{x}\le C(\varepsilon ,\Vert {\textsf{f}}\Vert _{H^{3}(D_{\varepsilon })})\tau ^2. \end{aligned}$$
(143)

By (131), as \({\textsf{f}}\in H^{4,3}(D_{\varepsilon })\) we have that \(\partial _{\textbf{x}}{\textsf{f}}\in H^{3,2}(D_{\varepsilon })\subset C^{1}(D_{\varepsilon })\) and we just need to prove the same type of estimate for the term

$$\begin{aligned}{} & {} \int _{D_{\varepsilon }}\left( \int _{D_\varepsilon (y)}\left| \frac{(\Psi _2[{\textsf{f}}+\tau h]-\Psi _2[{\textsf{f}}])(\textbf{x},{\bar{\textbf{x}}})}{\tau }-\left( \Psi _1[{\textsf{f}}]-\Psi _1[{\textsf{f}}]\Psi _2[{\textsf{f}}]\right) (\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[h](\textbf{x})\right| ^2 |{\bar{\textbf{x}}}|^{4+2\gamma }\hbox {d}{\bar{\textbf{x}}} \right) \hbox {d}\textbf{x}. \end{aligned}$$

Thanks to the definition of \(\Psi _2[\cdot ](\textbf{x},{\bar{\textbf{x}}})\), adding and subtracting some appropriate term we have

$$\begin{aligned}{} & {} \left( \Psi _2[{\textsf{f}}+\tau h]-\Psi _2[{\textsf{f}}]\right) (\textbf{x},{\bar{\textbf{x}}})\\{} & {} \quad =\frac{\cosh ({\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}+\tau h](\textbf{x}))-\cosh ({\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}](\textbf{x}))}{\cosh ({\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}](\textbf{x}))-\cos ({\bar{x}})} -\cosh ({\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}+\tau h](\textbf{x}))\\{} & {} \qquad \frac{\cosh ({\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}+\tau h](\textbf{x}))-\cosh ({\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}](\textbf{x}))}{\left[ \cosh ({\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}+\tau h](\textbf{x}))-\cos ({\bar{x}})\right] \left[ \cosh ({\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}](\textbf{x}))-\cos ({\bar{x}})\right] }, \end{aligned}$$

and using on it some trigonometric identities, we obtain

$$\begin{aligned}{} & {} \left( \Psi _2[{\textsf{f}}+\tau h]-\Psi _2[{\textsf{f}}]\right) (\textbf{x},{\bar{\textbf{x}}})\\{} & {} \quad =\left[ \Psi _1[{\textsf{f}}](\textbf{x},{\bar{\textbf{x}}})-\Psi _1[{\textsf{f}}]\Psi _2[{\textsf{f}}](\textbf{x},{\bar{\textbf{x}}})\cosh \left( \tau \Delta _{{\bar{\textbf{x}}}}[h](\textbf{x})\right) \right] \sinh \left( \tau \Delta _{{\bar{\textbf{x}}}}[h](\textbf{x})\right) \\{} & {} \qquad -\Psi _1^2[{\textsf{f}}](\textbf{x},{\bar{\textbf{x}}})\sinh ^2(\tau \Delta _{{\bar{\textbf{x}}}}[h](\textbf{x})) +2\Psi _1[{\textsf{f}}]\Psi _2[{\textsf{f}}](\textbf{x},{\bar{\textbf{x}}})\sinh (\tau \Delta _{{\bar{\textbf{x}}}}[h](\textbf{x})) \\{} & {} \qquad \quad \sinh ^2\left( \tfrac{\tau }{2}\Delta _{{\bar{\textbf{x}}}}[h](\textbf{x})\right) \\{} & {} \qquad +2\Psi _2[{\textsf{f}}](\textbf{x},{\bar{\textbf{x}}})\left[ 1-\Psi _2[{\textsf{f}}](\textbf{x},{\bar{\textbf{x}}})\cosh (\tau \Delta _{{\bar{\textbf{x}}}}[h](\textbf{x}))\right] \sinh ^2\left( \tfrac{\tau }{2}\Delta _{{\bar{\textbf{x}}}}[h](\textbf{x})\right) \\{} & {} \qquad +\Psi _1[{\textsf{f}}+\tau h](\textbf{x},{\bar{\textbf{x}}})\left[ \frac{\cosh ({\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}+\tau h](\textbf{x}))-\cosh ({\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}](\textbf{x}))}{\cosh ({\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}](\textbf{x}))-\cos ({\bar{x}})}\right] ^2. \end{aligned}$$

Now, dividing the above expression by \(\tau \) and subtracting the term \((\Psi _1[{\textsf{f}}]-\Psi _1[{\textsf{f}}]\Psi _2[{\textsf{f}}])(\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[h](\textbf{x})\) we obtain that

$$\begin{aligned}{} & {} \frac{(\Psi _2[{\textsf{f}}+\tau h]-\Psi _2[{\textsf{f}}])(\textbf{x},{\bar{\textbf{x}}})}{\tau } -\left( \Psi _1[{\textsf{f}}]-\Psi _1[{\textsf{f}}]\Psi _2[{\textsf{f}}]\right) (\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[h](\textbf{x})\\{} & {} \quad =\Psi _1[{\textsf{f}}](\textbf{x},{\bar{\textbf{x}}})\left( \frac{\sinh \left( \tau \Delta _{{\bar{\textbf{x}}}}[h](\textbf{x})\right) }{\tau }-\Delta _{{\bar{\textbf{x}}}}[h](\textbf{x}) \right) \\{} & {} \qquad -\Psi _1[{\textsf{f}}]\Psi _2[{\textsf{f}}](\textbf{x},{\bar{\textbf{x}}})\left( \cosh \left( \tau \Delta _{{\bar{\textbf{x}}}}[h](\textbf{x})\right) \frac{\sinh \left( \tau \Delta _{{\bar{\textbf{x}}}}[h](\textbf{x})\right) }{\tau }-\Delta _{{\bar{\textbf{x}}}}[h](\textbf{x}) \right) \\{} & {} \qquad -\Psi _1^2[{\textsf{f}}](\textbf{x},{\bar{\textbf{x}}})\sinh ^2(\tau \Delta _{{\bar{\textbf{x}}}}[h](\textbf{x})) \\{} & {} \qquad +2\Psi _1[{\textsf{f}}]\Psi _2[{\textsf{f}}](\textbf{x},{\bar{\textbf{x}}})\sinh (\tau \Delta _{{\bar{\textbf{x}}}}[h](\textbf{x})) \sinh ^2\left( \tfrac{\tau }{2}\Delta _{{\bar{\textbf{x}}}}[h](\textbf{x})\right) \\{} & {} \qquad +2\Psi _2[{\textsf{f}}](\textbf{x},{\bar{\textbf{x}}})\left[ 1-\Psi _2[{\textsf{f}}](\textbf{x},{\bar{\textbf{x}}})\cosh (\tau \Delta _{{\bar{\textbf{x}}}}[h](\textbf{x}))\right] \sinh ^2\left( \tfrac{\tau }{2}\Delta _{{\bar{\textbf{x}}}}[h](\textbf{x})\right) \\{} & {} \qquad +\Psi _1[{\textsf{f}}+\tau h](\textbf{x},{\bar{\textbf{x}}})\left[ \frac{\cosh ({\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}+\tau h](\textbf{x}))-\cosh ({\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}](\textbf{x}))}{\cosh ({\bar{y}}+\Delta _{{\bar{\textbf{x}}}}[{\textsf{f}}](\textbf{x}))-\cos ({\bar{x}})}\right] ^2. \end{aligned}$$

Finally, using the Taylor expansion of the trigonometric functions \(\sinh (\cdot ), \cosh (\cdot )\) and (131) give us the desired inequality (143). \(\square \)

Proof of (iv)

Recalling (140) together with (141) and (142), we have

$$\begin{aligned}{} & {} \partial _{\textbf{x}}^3\left\{ \frac{{\mathcal {K}}[{\textsf{f}}+\tau h,{\textsf{f}}](\textbf{x},{\bar{\textbf{x}}})}{\tau }-\Psi _1[{\textsf{f}}](\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[h](\textbf{x})\right\} \nonumber \\{} & {} \quad = \sum _{i=1}^{5}\partial _{\textbf{x}}{\textsf{A}}_i'[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}})+\sum _{i=1}^{4}\partial _{\textbf{x}}{\textsf{B}}_i'[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}}), \end{aligned}$$
(144)

where,

and

$$\begin{aligned}{} & {} \partial _{\textbf{x}}{\textsf{B}}_1'[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}})=(\partial _{\textbf{x}}\Psi _2[{\textsf{f}}+\tau h]-\partial _{\textbf{x}}\Psi _2[{\textsf{f}}])(\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}{\textsf{f}}](\textbf{x})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}h](\textbf{x})\\{} & {} \qquad +(\Psi _2[{\textsf{f}}+\tau h]-\Psi _2[{\textsf{f}}])(\textbf{x},{\bar{\textbf{x}}})\left[ \Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}^2{\textsf{f}}](\textbf{x})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}h](\textbf{x})+\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}{\textsf{f}}](\textbf{x})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}^2h](\textbf{x})\right] \\{} & {} \quad =:{\textsf{B}}_{11}''[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}})+{\textsf{B}}_{12}''[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}}),\\{} & {} \partial _{\textbf{x}}{\textsf{B}}_2'[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}})=-(\partial _{\textbf{x}}\Psi _1[{\textsf{f}}+\tau h]-\partial _{\textbf{x}}\Psi _1[{\textsf{f}}])(\Psi _1[{\textsf{f}}+\tau h]+\Psi _1[{\textsf{f}}])(\textbf{x},{\bar{\textbf{x}}})\\{} & {} \qquad \Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}{\textsf{f}}](\textbf{x})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}h](\textbf{x})\\{} & {} \qquad \qquad \qquad \qquad \qquad -(\Psi _1[{\textsf{f}}+\tau h]-\Psi _1[{\textsf{f}}])(\partial _{\textbf{x}}\Psi _1[{\textsf{f}}+\tau h]\\{} & {} \qquad \qquad \qquad \qquad \qquad +\partial _{\textbf{x}}\Psi _1[{\textsf{f}}])(\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}{\textsf{f}}](\textbf{x})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}h](\textbf{x})\\{} & {} \qquad \qquad \qquad \qquad \qquad -(\Psi _1[{\textsf{f}}+\tau h]-\Psi _1[{\textsf{f}}])(\Psi _1[{\textsf{f}}+\tau h]\\{} & {} \qquad \qquad \qquad \qquad \qquad +\Psi _1[{\textsf{f}}])(\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}^2{\textsf{f}}](\textbf{x})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}h](\textbf{x})\\{} & {} \qquad \qquad \qquad \qquad \qquad -(\Psi _1[{\textsf{f}}+\tau h]-\Psi _1[{\textsf{f}}])(\Psi _1[{\textsf{f}}+\tau h]\\{} & {} \qquad \qquad \qquad \qquad \qquad +\Psi _1[{\textsf{f}}])(\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}{\textsf{f}}](\textbf{x})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}^2h](\textbf{x})\\{} & {} \quad =:{\textsf{B}}_{21}''[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}})+{\textsf{B}}_{22}''[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}})\\{} & {} \qquad \qquad \qquad \qquad \qquad \quad +{\textsf{B}}_{23}''[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}})+{\textsf{B}}_{24}''[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}}),\\{} & {} \partial _{\textbf{x}}{\textsf{B}}_3'[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}})= \tau (\partial _{\textbf{x}}\Psi _2[{\textsf{f}}+\tau h]\\{} & {} \qquad \qquad \qquad \qquad \qquad -2\Psi _1[{\textsf{f}}+\tau h]\partial _{\textbf{x}}\Psi _1[{\textsf{f}}+\tau h])(\textbf{x},{\bar{\textbf{x}}})\left( \Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}} h](\textbf{x})\right) ^2\\{} & {} \qquad \qquad \qquad \qquad \qquad +\tau (\Psi _2[{\textsf{f}}+\tau h]-\Psi _1^2[{\textsf{f}}+\tau h])(\textbf{x},{\bar{\textbf{x}}})2 \Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}} h](\textbf{x})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}^2 h](\textbf{x})\\{} & {} \quad =:{\textsf{B}}_{31}''[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}})+{\textsf{B}}_{32}''[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}}),\\{} & {} \partial _{\textbf{x}}{\textsf{B}}_4'[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}})= (\partial _{\textbf{x}}\Psi _1[{\textsf{f}}+\tau h]-\partial _{\textbf{x}}\Psi _1[{\textsf{f}}])(\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}^2 h](\textbf{x})\\{} & {} \qquad \qquad \qquad \qquad \qquad +(\Psi _1[{\textsf{f}}+\tau h]-\Psi _1[{\textsf{f}}])(\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}^3 h](\textbf{x})\\{} & {} \quad =:{\textsf{B}}_{41}''[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}})+{\textsf{B}}_{42}''[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}}). \end{aligned}$$

Taking into account all the above we have proved that

$$\begin{aligned}{} & {} \int _{D_{\varepsilon }}\left( \int _{D_\varepsilon (y)}\left| \partial _{\textbf{x}}^3\left\{ \frac{{\mathcal {K}}[{\textsf{f}}+\tau h,{\textsf{f}}](\textbf{x},{\bar{\textbf{x}}})}{\tau }-\Psi _1[{\textsf{f}}](\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[h](\textbf{x})\right\} \right| ^2 |{\bar{\textbf{x}}}|^{2}\hbox {d}{\bar{\textbf{x}}} \right) \hbox {d}\textbf{x}\\{} & {} \quad \lesssim \sum _{i=1}^{5} \int _{D_{\varepsilon }}\left( \int _{D_\varepsilon (y)}\left| \partial _{\textbf{x}}{\textsf{A}}_i'[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}})\right| ^2 |{\bar{\textbf{x}}}|^{2}\hbox {d}{\bar{\textbf{x}}} \right) \hbox {d}\textbf{x}\\{} & {} \qquad + \sum _{i=1}^{4} \int _{D_{\varepsilon }}\left( \int _{D_\varepsilon (y)}\left| \partial _{\textbf{x}}{\textsf{B}}_i'[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}})\right| ^2 |{\bar{\textbf{x}}}|^{2}\hbox {d}{\bar{\textbf{x}}} \right) \hbox {d}\textbf{x}, \end{aligned}$$

where each of the terms that appear can be decomposed as follows:

$$\begin{aligned} \partial _{\textbf{x}}{\textsf{A}}_1'[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}})&={\textsf{A}}_{11}''[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}})+{\textsf{A}}_{12}''[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}}),\\ \partial _{\textbf{x}}{\textsf{A}}_2'[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}})&={\textsf{A}}_{21}''[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}})+{\textsf{A}}_{22}''[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}})+{\textsf{A}}_{23}''[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}}),\\ \partial _{\textbf{x}}{\textsf{A}}_3'[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}})&={\textsf{A}}_{31}''[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}})+{\textsf{A}}_{32}''[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}})+{\textsf{A}}_{33}''[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}})+{\textsf{A}}_{34}''[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}}),\\ \partial _{\textbf{x}}{\textsf{A}}_4'[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}})&={\textsf{A}}_{41}''[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}})+{\textsf{A}}_{42}''[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}}),\\ \partial _{\textbf{x}}{\textsf{A}}_5'[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}})&={\textsf{A}}_{51}''[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}})+{\textsf{A}}_{52}''[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}}), \end{aligned}$$

and

$$\begin{aligned} \partial _{\textbf{x}}{\textsf{B}}_1'[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}})&={\textsf{B}}_{11}''[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}})+{\textsf{B}}_{12}''[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}}),\\ \partial _{\textbf{x}}{\textsf{B}}_2'[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}})&={\textsf{B}}_{21}''[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}})+{\textsf{B}}_{22}''[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}})+{\textsf{B}}_{23}''[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}})+{\textsf{B}}_{24}''[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}}),\\ \partial _{\textbf{x}}{\textsf{B}}_3'[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}})&={\textsf{B}}_{31}''[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}})+{\textsf{B}}_{32}''[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}}),\\ \partial _{\textbf{x}}{\textsf{B}}_4'[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}})&={\textsf{B}}_{41}''[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}})+{\textsf{B}}_{42}''[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}}). \end{aligned}$$

For the sake of brevity, we shall present here the complete details for the term \({\textsf{A}}_{51}''[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}})\) and the other terms can be dealt using the previous lemmas via straightforward variations. To sum up, our goal reduces to check that

$$\begin{aligned} \int _{D_{\varepsilon }}\left( \int _{D_\varepsilon (y)}\left| {\textsf{A}}_{51}''[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}})\right| ^2 |{\bar{\textbf{x}}}|^{2}\hbox {d}{\bar{\textbf{x}}} \right) \hbox {d}\textbf{x}\le C(\varepsilon )\tau ^2. \end{aligned}$$
(145)

Notice that \({\textsf{A}}_{51}''[{\textsf{f}},h]\) can be written in a more manageable way (adding and subtracting terms) as

$$\begin{aligned} {\textsf{A}}_{51}''[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}})= & {} 2(\Psi _1[{\textsf{f}}]-\Psi _1[{\textsf{f}}+\tau h])(\textbf{x},{\bar{\textbf{x}}}) \partial _{\textbf{x}}\\{} & {} \left\{ \frac{(\Psi _1[{\textsf{f}}+\tau h]-\Psi _1[{\textsf{f}}])(\textbf{x},{\bar{\textbf{x}}})}{\tau } \right\} \left( \Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}{\textsf{f}}](\textbf{x})\right) ^2\\{} & {} -2 \partial _{\textbf{x}}\Psi _1[{\textsf{f}}](\textbf{x},{\bar{\textbf{x}}}) \left\{ \frac{(\Psi _1[{\textsf{f}}+\tau h] -\Psi _1[{\textsf{f}}])(\textbf{x},{\bar{\textbf{x}}})}{\tau } \right. \\{} & {} \left. -(\Psi _2[{\textsf{f}}]-\Psi _1^2[{\textsf{f}}])(\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[h](\textbf{x})\right\} \left( \Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}{\textsf{f}}](\textbf{x})\right) ^2\\{} & {} -2 \Psi _1[{\textsf{f}}](\textbf{x},{\bar{\textbf{x}}})\partial _{\textbf{x}}\left\{ \frac{(\Psi _1[{\textsf{f}}+\tau h]-\Psi _1[{\textsf{f}}])(\textbf{x},{\bar{\textbf{x}}})}{\tau }-(\Psi _2[{\textsf{f}}]\right. \\{} & {} \left. -\Psi _1^2[{\textsf{f}}])(\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[h](\textbf{x})\right\} \left( \Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}{\textsf{f}}](\textbf{x})\right) ^2. \end{aligned}$$

Now, each one of the above terms can be easily handled. Using (131) and applying repeatedly Lemma 7.3 we get that the first term is bounded as required. For the second term, we only need to note that it can be written, remembering (138), as

$$\begin{aligned}{} & {} 2 \partial _{\textbf{x}}\Psi _1[{\textsf{f}}](\textbf{x},{\bar{\textbf{x}}}) \left\{ \frac{(\Psi _1[{\textsf{f}}+\tau h]-\Psi _1[{\textsf{f}}])(\textbf{x},{\bar{\textbf{x}}})}{\tau } \right. \\{} & {} \quad \left. - (\Psi _2[{\textsf{f}}]-\Psi _1^2[{\textsf{f}}])(\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[h](\textbf{x})\right\} \left( \Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}{\textsf{f}}](\textbf{x})\right) ^2\\{} & {} \quad =2 \partial _{\textbf{x}}\Psi _1[{\textsf{f}}](\textbf{x},{\bar{\textbf{x}}}) {\textsf{A}}[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}}) \Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}{\textsf{f}}](\textbf{x}). \end{aligned}$$

Combining (126) with Corollary 7.2 and (131) we obtain the bound \(2 |\partial _{\textbf{x}}\Psi _1[{\textsf{f}}](\textbf{x},{\bar{\textbf{x}}}) \Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}{\textsf{f}}](\textbf{x})|\le C(\varepsilon ).\) After that, as an immediate consequence of (139) we get the required bound for the second term. For the latter, we proceed in the same spirit as before, trying to split that term into previously studied terms. Notice that

$$\begin{aligned}{} & {} 2 \Psi _1[{\textsf{f}}](\textbf{x},{\bar{\textbf{x}}})\partial _{\textbf{x}}\left\{ \frac{(\Psi _1[{\textsf{f}}+\tau h]-\Psi _1[{\textsf{f}}])(\textbf{x},{\bar{\textbf{x}}})}{\tau }-(\Psi _2[{\textsf{f}}]-\Psi _1^2[{\textsf{f}}])(\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[h](\textbf{x})\right\} \\{} & {} \qquad \left( \Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}{\textsf{f}}](\textbf{x})\right) ^2\\{} & {} \quad =2 \Psi _1[{\textsf{f}}](\textbf{x},{\bar{\textbf{x}}})\partial _{\textbf{x}}{\textsf{A}}[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}}) \Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}{\textsf{f}}](\textbf{x})-2 \Psi _1[{\textsf{f}}](\textbf{x},{\bar{\textbf{x}}}){\textsf{A}}[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}}) \Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}^2{\textsf{f}}](\textbf{x})\\{} & {} \quad =2 \Psi _1[{\textsf{f}}](\textbf{x},{\bar{\textbf{x}}}) \left\{ {\textsf{A}}_1'[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}})+{\textsf{A}}_4'[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}})+{\textsf{A}}_5'[{\textsf{f}},h](\textbf{x},{\bar{\textbf{x}}})\right\} \Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}{\textsf{f}}](\textbf{x}). \end{aligned}$$

Using Corollary 7.1 and (131) we obtain the bound \(2 |\Psi _1[{\textsf{f}}](\textbf{x},{\bar{\textbf{x}}})\Delta _{{\bar{\textbf{x}}}}[\partial _{\textbf{x}}{\textsf{f}}](\textbf{x})|\le C(\varepsilon )\). Finally, working as we did to get inequality (143) we get the required bound for each of the above terms. Therefore, combining all we have proved our goal. \(\square \)

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

Castro, Á., Lear, D. Traveling Waves Near Couette Flow for the 2D Euler Equation. Commun. Math. Phys. 400, 2005–2079 (2023). https://doi.org/10.1007/s00220-023-04636-6

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s00220-023-04636-6

Navigation