Skip to main content
Log in

Computation of Smooth Manifolds Via Rigorous Multi-parameter Continuation in Infinite Dimensions

  • Published:
Foundations of Computational Mathematics Aims and scope Submit manuscript

Abstract

In this paper, we introduce a constructive rigorous numerical method to compute smooth manifolds implicitly defined by infinite-dimensional nonlinear operators. We compute a simplicial triangulation of the manifold using a multi-parameter continuation method on a finite-dimensional projection. The triangulation is then used to construct local charts and an atlas of the manifold in the infinite-dimensional domain of the operator. The idea behind the construction of the smooth charts is to use the radii polynomial approach to verify the hypotheses of the uniform contraction principle over a simplex. The construction of the manifold is globalized by proving smoothness along the edge of adjacent simplices. We apply the method to compute portions of a two-dimensional manifold of equilibria of the Cahn–Hilliard equation.

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
Fig. 2
Fig. 3
Fig. 4
Fig. 5
Fig. 6
Fig. 7
Fig. 8

Similar content being viewed by others

Notes

  1. By “patch centered at x,” we mean the neighborhood of simplices adjacent to x.

References

  1. Eugene Allgower and Kurt Georg, Simplicial and continuation methods for approximating fixed points and solutions to systems of equations, SIAM Rev., 22(1), pp. 28–85, (1980).

    Article  MathSciNet  MATH  Google Scholar 

  2. Jan Bouwe van den Berg, Jean-Philippe Lessard, and Konstantin Mischaikow, Global smooth solution curves using rigorous branch following, Math. Comp., 79(271), pp. 1565–1584, (2010).

  3. Jan Bouwe van den Berg, Jason D. Mireles-James, Jean-Philippe Lessard, and Konstantin Mischaikow, Rigorous numerics for symmetric connecting orbits: even homoclinics of the Gray-Scott equation, SIAM J. Math. Anal., 43(4), pp. 1557–1594, (2011).

  4. Wolf-Jürgen Beyn and Eusebius Doedel, Stability and multiplicity of solutions to discretizations of nonlinear ordinary differential equations, SIAM J. Sci. Statist. Comput., 2(1), pp. 107–120, (1981).

    Article  MathSciNet  MATH  Google Scholar 

  5. Maxime Breden, Jean-Philippe Lessard, and Matthieu Vanicat, Global Bifurcation Diagrams of Steady States of Systems of PDEs via Rigorous Numerics: a 3-Component Reaction-Diffusion System, Acta Appl. Math., 128, pp. 113–152, (2013).

    Article  MathSciNet  MATH  Google Scholar 

  6. M. L. Brodzik and W. C. Rheinboldt, The computation of simplicial approximations of implicitly defined two-dimensional manifolds, Comput. Math. Appl., 28(9), pp. 9–21, (1994).

    Article  MathSciNet  MATH  Google Scholar 

  7. M. L. Brodzik, The computation of simplicial approximations of implicitly defined p-dimensional manifolds, Comput. Math. Appl., 36(6), pp. 93–113, (1998).

    Article  MathSciNet  MATH  Google Scholar 

  8. J. W. Cahn and J. E. Hilliard, Free energy of a nonuniform system i. interfacial free energy, Journal of Chemical Physics, 28, pp. 258–267, (1958).

    Article  Google Scholar 

  9. Shui Nee Chow and Jack K. Hale, Methods of bifurcation theory, volume 251 of Grundlehren der Mathematischen Wissenschaften [Fundamental Principles of Mathematical Science]. Springer-Verlag, New York, 1982.

  10. Sarah Day, Yasuaki Hiraoka, Konstantin Mischaikow, and Toshi Ogawa, Rigorous numerics for global dynamics: a study of the Swift-Hohenberg equation, SIAM J. Appl. Dyn. Syst., 4(1), pp. 1–31, (2005).

    Article  MathSciNet  MATH  Google Scholar 

  11. L. Dieci and T. Eirola, Numerical dynamical systems, Lecture Notes, 2005.

  12. Luca Dieci, Maria Grazia Gasparo, and Alessandra Papini, Path following by svd, In Computational Science-ICCS 2006, pp. 677–684, Springer, 2006.

  13. Nelson Dunford and Jacob T. Schwartz, Linear operators. Part I. Wiley Classics Library. John Wiley & Sons Inc., New York, 1988. General theory, With the assistance of William G. Bade and Robert G. Bartle, Reprint of the 1958 original, A Wiley-Interscience Publication.

  14. Marcio Gameiro and Jean-Philippe Lessard, Analytic estimates and rigorous continuation for equilibria of higher-dimensional PDEs, J. Differential Equations, 249(9), pp. 2237–2268, (2010).

    Article  MathSciNet  MATH  Google Scholar 

  15. Marcio Gameiro and Jean-Philippe Lessard, Efficient Rigorous Numerics for Higher-Dimensional PDEs via One-Dimensional Estimates, SIAM J. Numer. Anal., 51(4), pp. 2063–2087, (2013).

    Article  MathSciNet  MATH  Google Scholar 

  16. Marcio Gameiro, Jean-Philippe Lessard, and Konstantin Mischaikow, Validated continuation over large parameter ranges for equilibria of PDEs, Math. Comput. Simulation, 79(4), pp. 1368–1382, (2008).

    Article  MathSciNet  MATH  Google Scholar 

  17. Marcio Gameiro, Jean-Philippe Lessard, and Alessandro Pugliese, MATLAB codes, http://archimede.mat.ulaval.ca/jplessard/galepu.

  18. Michael E. Henderson, Multiple parameter continuation: computing implicitly defined k-manifolds, Internat. J. Bifur. Chaos Appl. Sci. Engrg., 12(3), pp. 451–476, (2002).

    Article  MathSciNet  MATH  Google Scholar 

  19. Jean-Philippe Lessard, Recent advances about the uniqueness of the slowly oscillating periodic solutions of Wright’s equation, J. Differential Equations, 248(5), pp. 992–1016, (2010).

    Article  MathSciNet  MATH  Google Scholar 

  20. R. Melville and D. S. Mackey, A new algorithm for two-dimensional numerical continuation, Comput. Math. Appl., 30(1), pp. 31–46, (1995).

    Article  MathSciNet  MATH  Google Scholar 

  21. Stanislaus Maier-Paape, Ulrich Miller, Konstantin Mischaikow, and Thomas Wanner, Rigorous numerics for the Cahn-Hilliard equation on the unit square, Rev. Mat. Complut., 21(2), pp. 351–426, (2008).

    Article  MathSciNet  MATH  Google Scholar 

  22. Michael Plum, Existence and enclosure results for continua of solutions of parameter-dependent nonlinear boundary value problems, J. Comput. Appl. Math., 60(1–2), pp. 187–200, (1995). Linear/nonlinear iterative methods and verification of solution (Matsuyama, 1993).

  23. Werner C. Rheinboldt, On the computation of multidimensional solution manifolds of parametrized equations, Numer. Math.,53(1-2), pp. 165–181, (1988).

  24. S.M. Rump, INTLAB - INTerval LABoratory, In Tibor Csendes, editor, Developments in Reliable Computing, pp. 77–104. Kluwer Academic Publishers, Dordrecht, 1999. http://www.ti3.tuhh.de/rump/intlab/.

Download references

Acknowledgments

The authors would like to thank the anonymous referees for helpful comments and suggestions. Part of this work was done while the third author was visiting the School of Mathematics of the Georgia Institute of Technology, whose hospitality and support are gratefully acknowledged.

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to Marcio Gameiro.

Additional information

Communicated by Richard Schwartz.

Marcio Gameiro was partially supported by FAPESP Grants 2013/07460-7 and 2010/00875-9, by CNPq Grant 306453/2009-6 and by CAPES Grant BEX 3979/10-5, Brazil. Alessandro Pugliese was supported in part by INdAM–GNCS.

Appendices

Appendix 1: Convolution estimates

In this Appendix, we provide the necessary convolution estimates required to construct the radii polynomials for the Cahn–Hilliard equation studied in Sect. 5. We decided to include all formulas and proofs so that the paper is self-contained. Note, however, that these analytic convolution estimates are taken directly from [15] for estimates concerning quadratic and cubic nonlinearities.

Consider a decay rate \(q \ge 2\), a computational parameter \(M \ge 6\) and define, for \(k \ge 3\),

$$\begin{aligned} \gamma _k = \gamma _k(q) \,\mathop {=}\limits ^{{{\mathrm{def}}}}\,2 \left[ \frac{k}{k-1}\right] ^q + \left[ \frac{4\ln (k-2)}{k} + \frac{\pi ^2-6}{3} \right] \left[ \frac{2}{k} + \frac{1}{2} \right] ^{q-2}. \end{aligned}$$
(47)

Lemma 6.1

For \(q \ge 2\) and \(k \ge 4\), we have

$$\begin{aligned} \sum _{k_1 = 1}^{k-1} \frac{k^q}{k_1^q (k-k_1)^q} \le \gamma _k. \end{aligned}$$

Proof

First observe that

$$\begin{aligned} \sum _{k_1 = 1}^{k-1} \frac{k^q}{k_1^q (k-k_1)^q}= & {} 2 {\left[ \frac{k}{k-1} \right] }^q + k^{q-1} \sum _{k_1 = 2}^{k-2} \frac{k}{k_1^q (k-k_1)^q} \\= & {} 2 {\left[ \frac{k}{k-1} \right] }^q + k^{q-1} \left[ \sum _{k_1 = 2}^{k-2} \frac{k-k_1}{k_1^q (k-k_1)^q} + \sum _{k_1 = 2}^{k-2} \frac{k_1}{k_1^q (k-k_1)^q} \right] \\= & {} 2 {\left[ \frac{k}{k-1} \right] }^q + k^{q-1} \left[ \sum _{k_1 = 2}^{k-2} \frac{1}{k_1^q (k-k_1)^{q-1}} + \sum _{k_1 = 2}^{k-2} \frac{1}{k_1^{q-1} (k-k_1)^q} \right] \\= & {} 2 {\left[ \frac{k}{k-1} \right] }^q + 2 \sum _{k_1 = 2}^{k-2} \frac{k^{q-1}}{k_1^{q-1} (k-k_1)^q}. \end{aligned}$$

Using the above, we define

$$\begin{aligned} \phi _k^{(q)} := \sum _{k_1=2}^{k-2} \frac{k^{q-1}}{k_1^{q-1} (k-k_1)^q} = \frac{1}{2} \sum _{k_1=2}^{k-2} \frac{k^q}{k_1^q (k-k_1)^q}. \end{aligned}$$

We then obtain the following recurrence inequality

$$\begin{aligned} \phi _k^{(q)}= & {} \sum _{k_1=2}^{k-2} \frac{k^{q-1}}{k_1^{q-1} (k-k_1)^q} = k^{q-2} \sum _{k_1=2}^{k-2} \frac{(k-k_1) + k_1}{k_1^{q-1} (k-k_1)^q} \\= & {} \frac{1}{k} \sum _{k_1=2}^{k-2} \frac{k^{q-1}}{k_1^{q-1} (k-k_1)^{q-1}} + \sum _{k_1=2}^{k-2} \frac{k^{q-2}}{k_1^{q-2} (k-k_1)^q} \\\le & {} \frac{1}{k} \sum _{k_1=2}^{k-2} \frac{k^{q-1}}{k_1^{q-1} (k-k_1)^{q-1}} + \frac{1}{2} \sum _{k_1=2}^{k-2} \frac{k^{q-2}}{k_1^{q-2} (k-k_1)^{q-1}} \;\; = \;\; \left[ \frac{2}{k} + \frac{1}{2} \right] \phi _k^{(q-1)}. \end{aligned}$$

Applying the above inequality \(q-2\) times, we get

$$\begin{aligned} \phi _k^{(q)} \le \phi _k^{(2)} {\left[ \frac{2}{k} + \frac{1}{2} \right] }^{q-2}. \end{aligned}$$

Also

$$\begin{aligned} \phi _k^{(2)}= & {} \sum _{k_1=2}^{k-2} \frac{k}{k_1 (k-k_1)^2} = \sum _{k_1=2}^{k-2} \frac{1}{k_1 (k-k_1)} + \sum _{k_1=2}^{k-2} \frac{1}{(k-k_1)^2} \\= & {} \frac{1}{k} \left[ \sum _{k_1=2}^{k-2} \frac{1}{k_1} + \sum _{k_1=2}^{k-2} \frac{1}{k-k_1} \right] + \sum _{k_1=2}^{k-2} \frac{1}{(k-k_1)^2} \\= & {} \frac{2}{k} \sum _{k_1=2}^{k-2} \frac{1}{k_1} + \sum _{k_1=2}^{k-2} \frac{1}{k_1^2} \le \frac{2}{k} \ln {(k-2)} + \frac{\pi ^2}{6} -1. \end{aligned}$$

Using the above inequalities, we get

$$\begin{aligned} \sum _{k_1 = 1}^{k-1} \frac{k^q}{k_1^q (k-k_1)^q}= & {} 2 {\left[ \frac{k}{k-1} \right] }^q + 2 \phi _k^{(q)} \le 2 {\left[ \frac{k}{k-1} \right] }^q + 2 \phi _k^{(2)} {\left[ \frac{2}{k} + \frac{1}{2} \right] }^{q-2} \\\le & {} 2 {\left[ \frac{k}{k-1} \right] }^q + \left[ \frac{4 \ln {(k-2)}}{k} + \frac{\pi ^2 - 6}{3} \right] {\left[ \frac{2}{k} + \frac{1}{2} \right] }^{q-2} \;\; = \;\; \gamma _k. \end{aligned}$$

\(\square \)

Define the weights by

$$\begin{aligned} \omega _k^q := {\left\{ \begin{array}{ll} ~ 1, &{} \text {if} ~ k = 0 \\ |k|^q, &{} \text {if} ~ k \ne 0. \end{array}\right. } \end{aligned}$$

Lemma 6.2

(Quadratic estimates). Given a decay rate \(q \ge 2\) and \(M \ge 6\). For \(k \in \mathbb {Z}\), define the quadratic asymptotic estimates \(\alpha _k^{(2)} = \alpha _k^{(2)}(q,M)\) by

$$\begin{aligned} \alpha _k^{(2)} \,\mathop {=}\limits ^{{{\mathrm{def}}}}\,\left\{ \begin{array}{ll} \displaystyle 1 + 2 \sum _{k_1=1}^M \frac{1}{\omega _{k_1}^{2q}} + \frac{2}{M^{2q-1}(2q-1)}, &{} \text {for} ~ k=0 \\ \displaystyle \sum _{k_1 = 1}^{M} \frac{2 \omega _k^q}{\omega _{k_1}^q \omega _{k+k_1}^q} + \frac{2 \omega _k^q}{(k+M+1)^q M^{q-1} (q-1)} \\ \qquad \displaystyle +\, 2 + \sum _{k_1 = 1}^{k-1} \frac{\omega _k^q}{\omega _{k_1}^q \omega _{k-k_1}^q}, &{} \text {for} ~ 1 \le k \le M-1 \\ \displaystyle { 2 + 2 \sum _{k_1=1}^M \frac{1}{\omega _{k_1}^q} + \frac{2}{M^{q-1}(q-1)} + \gamma _M, } &{} \text {for} ~ k \ge M, \end{array} \right. \end{aligned}$$
(48)

and for \(k < 0\),

$$\begin{aligned} \alpha _k^{(2)} \,\mathop {=}\limits ^{{{\mathrm{def}}}}\,\alpha _{|k|}^{(2)}. \end{aligned}$$

Then, for any \(k \in \mathbb {Z}\),

$$\begin{aligned} \sum _{\mathop {k_j \in \mathbb {Z}}\limits ^{k_1 + k_2 = k}} \frac{1}{\omega ^q_{k_1} \omega ^q_{k_2}} \le \frac{\alpha ^{(2)}_k}{\omega ^q_k}. \end{aligned}$$

Proof

For \(k=0\),

$$\begin{aligned} \sum _{\mathop {k_j \in \mathbb {Z}}\limits ^{k_1 + k_2 = 0}} \frac{1}{\omega ^q_{k_1} \omega ^q_{k_2}}= & {} 1 + 2 \sum _{k_1 = 1}^{M} \frac{1}{\omega _{k_1}^{2q}} + 2 \sum _{k_1 = M+1}^{\infty } \frac{1}{\omega _{k_1}^{2q}} \le 1 + 2 \sum _{k_1 = 1}^{M} \frac{1}{\omega _{k_1}^{2q}} + \int _M^\infty \frac{\mathrm{d}x}{x^{2q}} \\\le & {} 1 + 2 \sum _{k_1 = 1}^{M} \frac{1}{\omega _{k_1}^{2q}} + \frac{2}{M^{2q-1} (2q - 1)} = \frac{\alpha _0^{(2)}}{\omega _0^q}. \end{aligned}$$

For \(1 \le k \le M-1\), and recalling that the one-dimensional weights (2),

$$\begin{aligned}&\sum _{\mathop {k_j \in \mathbb {Z}}\limits ^{k_1 + k_2 = k}} \frac{1}{\omega ^q_{k_1} \omega ^q_{k_2}} \\&\quad = \frac{1}{\omega ^q_k} \left[ \sum _{k_1 = 1}^{M} \frac{2 \omega ^q_k}{\omega ^q_{k_1} \omega ^q_{k+k_1}} + \sum _{k_1 = M+1}^{\infty } \frac{2 \omega ^q_k}{\omega ^q_{k_1} \omega ^q_{k+k_1}} + \frac{2}{\omega ^q_0} + \sum _{k_1 = 1}^{k-1} \frac{\omega ^q_k}{\omega ^q_{k_1} \omega ^q_{k-k_1}} \right] \\&\quad \le \frac{1}{\omega ^q_k} \left[ \sum _{k_1 = 1}^{M} \frac{2 \omega _k^q}{\omega _{k_1}^q \omega _{k+k_1}^q} + \frac{2 \omega _k^q}{(k+M+1)^q} \int _M^\infty \frac{\mathrm{d}x}{x^q} + 2 + \sum _{k_1 = 1}^{k-1} \frac{\omega _k^q}{\omega _{k_1}^q \omega _{k-k_1}^q} \right] \\&\quad \le \frac{1}{\omega ^q_k} \left[ \sum _{k_1 = 1}^{M} \frac{2 \omega _k^q}{\omega _{k_1}^q \omega _{k+k_1}^q} + \frac{2 \omega _k^q}{(k+M+1)^q M^{q-1} (q-1)} + 2 + \sum _{k_1 = 1}^{k-1} \frac{\omega _k^q}{\omega _{k_1}^q \omega _{k-k_1}^q} \right] = \frac{\alpha _k^{(2)}}{\omega ^q_k}. \end{aligned}$$

Finally, for \(k \ge M\), one gets from Lemma 6.1 that

$$\begin{aligned} \sum _{\mathop {k_j \in \mathbb {Z}}\limits ^{k_1 + k_2 = k}} \frac{1}{\omega ^q_{k_1} \omega ^q_{k_2}}= & {} \frac{1}{\omega ^q_k} \left[ 2 \sum _{k_1 = 1}^{\infty } \frac{\omega ^q_k}{\omega ^q_{k_1} \omega ^q_{k+k_1}} + \frac{2}{\omega ^q_0} + \sum _{k_1 = 1}^{k-1} \frac{\omega ^q_k}{\omega ^q_{k_1} \omega ^q_{k-k_1}} \right] \\\le & {} \frac{1}{\omega ^q_k} \left[ 2 \sum _{k_1=1}^M \frac{1}{\omega _{k_1}^q} + 2 \sum _{k_1=M+1}^\infty \frac{1}{\omega _{k_1}^q} + \frac{2}{\omega ^q_0} + \gamma _k \right] \\\le & {} \frac{1}{\omega ^q_k} \left[ 2 \sum _{k_1=1}^M \frac{1}{\omega _{k_1}^q} + 2 \int _{M}^\infty \frac{\mathrm{d}x}{x^q} + \frac{2}{\omega ^q_0} + \gamma _M \right] \\\le & {} \frac{1}{\omega ^q_k} \left[ \sum _{k_1=1}^M \frac{2}{\omega _{k_1}^q} + \frac{2}{M^{q-1}(q-1)} + 2 + \gamma _M \right] = \frac{\alpha _k^{(2)}}{\omega ^q_k}. \end{aligned}$$

\(\square \)

Lemma 6.3

For any \(k \in \mathbb {Z}\) with \(|k| \ge M \ge 6\), we have that \(\alpha _k^{(2)} \le \alpha _M^{(2)}\).

Proof

For \(k \ge 6\), the fact that \( \frac{\ln (k-1)}{k+1} \le \frac{\ln (k-2)}{k} \) implies that \(\gamma _{k+1}(q) \le \gamma _{k}(q)\). By definition of \(\alpha _k^{(2)}\), for \(|k| \ge M\), one gets that \(\alpha _k^{(2)} \le \alpha _M^{(2)}\). \(\square \)

Lemma 6.4

(Cubic estimates). Given \(q \ge 2\) and \(M \ge 6\). Let

$$\begin{aligned} \Sigma _a^* := \sum _{k_1=1}^{M-1} \frac{\alpha _{k_1}^{(2)} M^q }{\omega _{k_1}^q \bigl ( M - k_1 \bigr )^{q} } + \alpha _M^{(2)} \left( \gamma _M - \sum _{k_1=1}^{M-1} \frac{1}{\omega _{k_1}^q} \right) , \end{aligned}$$

\(\tilde{\alpha }_{M}^{(2)} := \max \left\{ \alpha _k^{(2)} \mid k = 0, \ldots , M \right\} \), \(\Sigma _b^* := \tilde{\alpha }_{M}^{(2)} \gamma _M \) and \(\Sigma ^* := \min \left\{ \Sigma _a^*, \Sigma _b^* \right\} \). Define the cubic asymptotic estimates \(\alpha _k^{(3)} = \alpha _k^{(3)}(s,M)\) by

$$\begin{aligned} \alpha _k^{(3)} \! \,\mathop {=}\limits ^{{{\mathrm{def}}}}\,\! \left\{ \!\! \begin{array}{ll} \displaystyle { \alpha _0^{(2)} + 2 \sum _{k_1=1}^{M-1} \frac{\alpha _{k_1}^{(2)}}{\omega _{k_1}^{2q}} + \frac{2 \alpha _M^{(2)} }{(M-1)^{2q-1}(2q-1)}, } &{} \mathrm{for} ~ k=0 \\ \displaystyle { \sum _{k_1=1}^{M-k} \! \frac{\alpha _{k+k_1}^{(2)} \omega _k^q }{\omega _{k_1}^q \omega _{k+k_1}^q} + \frac{\alpha _M^{(2)} \omega _k^q }{(M+1)^q(M-k)^{q-1}(q-1)} + \sum _{k_1=1}^{k-1} \frac{\alpha _{k_1}^{(2)} \omega _k^q}{\omega _{k_1}^q \omega _{k-k_1}^q} }\\ \displaystyle \qquad + \sum _{k_1=1}^M \frac{\alpha _{k_1}^{(2)} \omega _k^q}{\omega _{k_1}^q \omega _{k+k_1}^q} + \frac{\alpha _M^{(2)} \omega _k^q }{(M+k+1)^qM^{q-1}(q-1)} + \alpha _{k}^{(2)}+ \alpha _0^{(2)},\\ \qquad \qquad &{} \mathrm{for} ~ 1 \le k \le M\!-\!1 \\ \displaystyle { \alpha _M^{(2)} \sum _{k_1=1}^M \frac{1}{\omega _{k_1}^q} + \frac{2 \alpha _M^{(2)}}{M^{q-1}(q-1)} + \Sigma ^* + \sum _{k_1=1}^M \frac{\alpha _{k_1}^{(2)}}{\omega _{k_1}^q} + \alpha _M^{(2)} + \alpha _0^{(2)}, } &{} \mathrm{for} ~ k \ge M \end{array} \right. \end{aligned}$$
(49)

and for \(k < 0\),

$$\begin{aligned} \alpha _k^{(3)} \,\mathop {=}\limits ^{{{\mathrm{def}}}}\,\alpha _{|k|}^{(3)}. \end{aligned}$$

Then, for any \(k \in \mathbb {Z}\),

$$\begin{aligned} \sum _{\mathop {k_j \in \mathbb {Z}}\limits ^{k_1 + k_2 + k_3 = k}} \frac{1}{\omega ^q_{k_1} \omega ^q_{k_2} \omega ^q_{k_3}} \le \frac{\alpha ^{(3)}_k}{\omega ^q_k}. \end{aligned}$$

Moreover, \(\alpha _k^{(3)} \le \alpha _M^{(3)}\), for all \(k \ge M\).

Proof

In what follows, the estimates are obtained similarly as in the proof of Lemma 6.2 with the difference that we often use the fact \(\alpha _k^{(2)} \le \alpha _M^{(2)}\), for all \(k \ge M\) (see, e.g., Remark A.1 in [14]). For \(k=0\),

$$\begin{aligned} \sum _{\mathop {k_j \in \mathbb {Z}}\limits ^{k_1 + k_2 + k_3 = 0}} \frac{1}{\omega ^q_{k_1} \omega ^q_{k_2} \omega ^q_{k_3}}\le & {} \alpha _0^{(2)} + 2 \sum _{k_1 = 1}^{M-1} \frac{\alpha _{k_1}^{(2)}}{\omega _{k_1}^{2q}} + \frac{2 \alpha _M^{(2)}}{(M - 1)^{2q-1} (2q-1)} = \frac{\alpha ^{(3)}_0}{\omega ^q_0}. \end{aligned}$$

For \(k > 0\),

$$\begin{aligned} \sum _{\mathop {k_j \in \mathbb {Z}}\limits ^{k_1 + k_2 + k_3 = k}} \frac{1}{\omega ^q_{k_1}\omega ^q_{k_2} \omega ^q_{k_3}}\le & {} \sum _{k_1 = 1}^{\infty } \left[ \frac{1}{\omega ^q_{k_1}} \frac{\alpha _{k+k_1}^{(2)}}{\omega ^q_{k+k_1}} \right] + \sum _{k_1 = 1}^{k-1} \left[ \frac{1}{\omega ^q_{k_1}} \frac{\alpha _{k-k_1}^{(2)}}{\omega ^q_{k-k_1}} \right] + \sum _{k_1 = 1}^{\infty } \left[ \frac{1}{\omega ^q_{k+k_1}} \frac{\alpha _{k_1}^{(2)}}{\omega ^q_{k_1}} \right] \\&+\, \frac{1}{\omega ^q_0} \frac{\alpha _{k}^{(2)}}{\omega ^q_{k}} + \frac{1}{\omega ^q_k} \frac{\alpha _{0}^{(2)}}{\omega ^q_{0}}. \end{aligned}$$

Consider \(k \in \{ 1, \ldots , M-1 \}\). Since \(\alpha _k^{(2)} \le \alpha _M^{(2)}\), for all \(k \ge M\) by Lemma 6.3, we have

$$\begin{aligned} \sum _{k_1=1}^{\infty } \frac{\alpha _{k+k_1}^{(2)} }{\omega _{k_1}^q \omega _{k+k_1}^q} \le \frac{1}{\omega _k^q} \left[ \sum _{k_1=1}^{M-k} \! \frac{\alpha _{k+k_1}^{(2)} \omega _k^q }{\omega _{k_1}^q \omega _{k+k_1}^q} + \frac{\alpha _M^{(2)} \omega _k^q }{(M+1)^q(M-k)^{q-1}(q-1)} \right] \! . \end{aligned}$$

Similarly,

$$\begin{aligned} \sum _{k_1=1}^{\infty } \frac{\alpha _{k_1}^{(2)}}{\omega _{k_1}^q \omega _{k+k_1}^q} \le \frac{1}{\omega _k^q} \left[ \sum _{k_1=1}^M \frac{\alpha _{k_1}^{(2)} \omega _k^q}{\omega _{k_1}^q \omega _{k+k_1}^q} + \frac{\alpha _M^{(2)} \omega _k^q }{(M+k+1)^qM^{q-1}(q-1)} \right] . \end{aligned}$$

From the definition of \(\alpha _k^{(3)}\) for \(k \in \{1, \ldots ,M-1\}\), one gets that

$$\begin{aligned} \sum _{\mathop {k_j \in \mathbb {Z}}\limits ^{k_1 + k_2 + k_3 = k}} \frac{1}{\omega ^q_{k_1}\omega ^q_{k_2} \omega ^q_{k_3}} \le \frac{ \alpha _k^{(3)}}{\omega _k^q}. \end{aligned}$$

For \(k \ge M\), using again that \(\alpha _k^{(2)} \le \alpha _M^{(2)}\) by Lemma 6.3, one gets that

$$\begin{aligned} \sum _{k_1=1}^{\infty } \frac{\alpha _{k+k_1}^{(2)} }{\omega _{k_1}^q \omega _{k+k_1}^q} \le \frac{1}{\omega _k^q} \left[ \alpha _M^{(2)} \sum _{k_1=1}^M \frac{1}{\omega _{k_1}^q} + \frac{\alpha _M^{(2)}}{M^{q-1}(q-1)} \right] . \end{aligned}$$

Using Lemma 6.1,

$$\begin{aligned} \sum _{k_1=1}^{k-1} \frac{\alpha _{k_1}^{(2)}}{\omega _{k_1}^q \omega _{k-k_1}^q}= & {} \sum _{k_1=1}^{M-1} \frac{\alpha _{k_1}^{(2)}}{\omega _{k_1}^q \omega _{k-k_1}^q} + \frac{1}{\omega _k^q} \sum _{k_1=M}^{k-1} \frac{\omega _k^q \alpha _{k_1}^{(2)}}{\omega _{k_1}^q \omega _{k-k_1}^q} \\\le & {} \frac{1}{\omega _k^q} \sum _{k_1=1}^{M-1} \frac{\alpha _{k_1}^{(2)}}{\omega _{k_1}^q \bigl (1-\frac{k_1}{k}\bigr )^{q}} + \frac{\alpha _M^{(2)}}{\omega _k^q} \sum _{k_1=M}^{k-1} \frac{\omega _k^q}{\omega _{k_1}^q \omega _{k-k_1}^q} \\\le & {} \frac{1}{\omega _k^q} \left[ \sum _{k_1=1}^{M-1} \frac{\alpha _{k_1}^{(2)}}{\omega _{k_1}^q \bigl ( 1 \!-\! \frac{k_1}{M} \bigr )^{q} } \!+\! \alpha _M^{(2)} \left( \sum _{k_1=1}^{k-1} \frac{\omega _k^q}{\omega _{k_1}^q \omega _{k-k_1}^q} \!-\! \sum _{k_1=1}^{M-1} \frac{\omega _k^q}{\omega _{k_1}^q \omega _{k-k_1}^q} \right) \right] \\\le & {} \frac{1}{\omega _k^q} \left[ \sum _{k_1=1}^{M-1} \frac{\alpha _{k_1}^{(2)} M^q }{\omega _{k_1}^q \bigl ( M - k_1 \bigr )^{q} } + \alpha _M^{(2)} \left( \gamma _M - \sum _{k_1=1}^{M-1} \frac{1}{\omega _{k_1}^q} \right) \right] = \frac{1}{\omega _k^q} \Sigma _a^* . \end{aligned}$$

Hence,

$$\begin{aligned} \sum _{k_1=1}^{k-1} \frac{\alpha _{k_1}^{(2)}}{\omega _{k_1}^q \omega _{k-k_1}^q} \le \frac{\tilde{\alpha }_{M}^{(2)} }{\omega _k^q} \gamma _M = \frac{1}{\omega _k^q} \Sigma _b^*. \end{aligned}$$

Recalling that \(\Sigma ^* = \min \left\{ \Sigma _a^*, \Sigma _b^* \right\} \), one gets that \( \displaystyle \sum \nolimits _{k_1=1}^{k-1} \frac{\alpha _{k_1}^{(2)}}{\omega _{k_1}^q \omega _{k-k_1}^q} \le \frac{1}{\omega _k^q} \Sigma ^*\). Also,

$$\begin{aligned} \sum _{k_1=1}^{\infty } \frac{\alpha _{k_1}^{(2)}}{\omega _{k_1}^q \omega _{k+k_1}^q} \le \frac{1}{\omega _k^q} \left[ \sum _{k_1=1}^M \frac{\alpha _{k_1}^{(2)}}{\omega _{k_1}^q} + \frac{\alpha _M^{(2)}}{M^{q-1} (q-1)} \right] . \end{aligned}$$

Combining the above inequalities, we get, for the case \(k \ge M\),

$$\begin{aligned}&\sum _{\mathop {k_j \in \mathbb {Z}}\limits ^{k_1 + k_2 + k_3 = k}} \frac{1}{\omega ^q_{k_1}\omega ^q_{k_2} \omega ^q_{k_3}} \\&\quad \quad \le \frac{1}{\omega _k^q} \left[ \alpha _M^{(2)} \sum _{k_1=1}^M \frac{1}{\omega _{k_1}^q} + \frac{2 \alpha _M^{(2)}}{M^{q-1}(q-1)} + \Sigma ^* + \sum _{k_1=1}^M \frac{\alpha _{k_1}^{(2)}}{\omega _{k_1}^q} \!+\! \alpha _M^{(2)} + \alpha _0^{(2)} \right] \!=\! \dfrac{\alpha _k^{(3)}}{\omega _k^q}. \end{aligned}$$

\(\square \)

Lemma 6.5

For any \(k \in \mathbb {Z}\) with \(|k| \ge M \ge 6\), we have that \(\alpha _k^{(3)} \le \alpha _M^{(3)}\).

Proof

For \(k \ge 6\), the fact that \( \frac{\ln (k-1)}{k+1} \le \frac{\ln (k-2)}{k} \) implies that \(\gamma _{k+1}(q) \le \gamma _{k}(q)\). By definition of \(\alpha _k^{(3)}\), for \(|k| \ge M\), one gets that \(\alpha _k^{(3)} \le \alpha _M^{(3)}\). \(\square \)

Lemma 6.6

Given \(q \ge 2\) and \(6 \le {\bar{M}}\le M\), define for \(0 \le k \le {\bar{M}}-1\)

$$\begin{aligned} \varepsilon _{k}^{(3)} = \varepsilon _{k}^{(3)}(q, {\bar{M}}, M):= & {} \sum _{k_1={\bar{M}}}^{M-k} \frac{\alpha _{k+k_1}^{(2)}}{ \omega ^q_{k_1} \omega ^q_{k+k_1}} + \sum _{k_1={\bar{M}}}^{M+k} \frac{\alpha _{k_1-k}^{(2)}}{\omega ^q_{k_1} \omega ^q_{k_1-k}}\nonumber \\&+\, \frac{ \alpha _{M}^{(2)}}{(M+1)^q(q-1)} \left[ \frac{1}{(M-k)^{q-1}} + \frac{1}{(M+k)^{q-1}} \right] \nonumber \\ \end{aligned}$$
(50)

and for \(k < 0\)

$$\begin{aligned} \varepsilon _{k}^{(3)}(q,{\bar{M}},M) := \varepsilon _{|k|}^{(3)}(q,{\bar{M}},M). \end{aligned}$$

Fix \(0 \le |k| \le {\bar{M}}-1\) and \(\ell \in \{1,2,3\}\). Then, we have that

$$\begin{aligned} \sum _{\mathop {\max \{ |k_1|, \ldots , |k_{\ell }| \} \ge {\bar{M}}}\limits ^{k_1+k_2 +k_3=k}} \frac{1}{\omega ^q_{k_1}\omega ^q_{k_2} \omega ^q_{k_3}} \le \ell \varepsilon _k^{(3)}. \end{aligned}$$

Proof

We have that

$$\begin{aligned} \sum _{\mathop {\max \{ |k_1|, \ldots , |k_{\ell }| \} \ge {\bar{M}}}\limits ^{k_1+k_2 +k_3=k}} \frac{1}{\omega ^q_{k_1}\omega ^q_{k_2} \omega ^q_{k_3}} \le \ell \sum _{\mathop { |k_1| \ge {\bar{M}}}\limits ^{k_1+k_2 +k_3=k}} \frac{1}{\omega ^q_{k_1}\omega ^q_{k_2} \omega ^q_{k_3}}, \end{aligned}$$

and

$$\begin{aligned} \sum _{\mathop { |k_1| \ge {\bar{M}}}\limits ^{k_1+k_2 +k_3=k}} \frac{1}{\omega ^q_{k_1}\omega ^q_{k_2} \omega ^q_{k_3}}= & {} \sum _{k_1=-\infty }^{-{\bar{M}}} \frac{1}{\omega ^q_{k_1}} \sum _{k_2 + k_3=k-k_1} \frac{1}{\omega ^q_{k_2}\omega ^q_{k_2} \omega ^q_{k_3}} \\&+ \sum _{k_1={\bar{M}}}^{\infty } \frac{1}{\omega ^q_{k_1}} \sum _{k_2 + k_3 = k-k_1} \frac{1}{\omega ^q_{k_2}\omega ^q_{k_2} \omega ^q_{k_3}} \\\le & {} \sum _{k_1={\bar{M}}}^{\infty } \left[ \frac{\alpha _{k+k_1}^{(2)}}{ \omega ^q_{k_1} \omega ^q_{k+k_1}} + \frac{\alpha _{k_1-k}^{(2)}}{\omega ^q_{k_1} \omega ^q_{k_1-k}} \right] \\\le & {} \sum _{k_1={\bar{M}}}^{M-k} \frac{\alpha _{k+k_1}^{(2)}}{ \omega ^q_{k_1} \omega ^q_{k+k_1}} + \alpha _{M}^{(2)} \sum _{k_1=M-k+1}^{\infty } \frac{1}{ \omega ^q_{k_1} \omega ^q_{k+k_1}} \\&+ \sum _{k_1={\bar{M}}}^{M+k} \frac{\alpha _{k_1-k}^{(2)}}{\omega ^q_{k_1} \omega ^q_{k_1-k}} + \alpha _{M}^{(2)} \sum _{k_1=M+k+1}^{\infty } \frac{1}{\omega ^q_{k_1} \omega ^q_{k_1-k}} \\\le & {} \sum _{k_1={\bar{M}}}^{M-k} \frac{\alpha _{k+k_1}^{(2)}}{ \omega ^q_{k_1} \omega ^q_{k+k_1}} + \sum _{k_1={\bar{M}}}^{M+k} \frac{\alpha _{k_1-k}^{(2)}}{\omega ^q_{k_1} \omega ^q_{k_1-k}}\\&+ \frac{ \alpha _{M}^{(2)}}{(M+1)^q(q-1)} \left[ \frac{1}{(M-k)^{q-1}} + \frac{1}{(M+k)^{q-1}} \right] =\varepsilon _k^{(3)}. \end{aligned}$$

\(\square \)

Appendix 2: Coefficients Used to Define the Bounds Y and Z

The coefficients are presented in Tables 3, 4, 5 and 6

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Gameiro, M., Lessard, JP. & Pugliese, A. Computation of Smooth Manifolds Via Rigorous Multi-parameter Continuation in Infinite Dimensions. Found Comput Math 16, 531–575 (2016). https://doi.org/10.1007/s10208-015-9259-7

Download citation

  • Received:

  • Revised:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s10208-015-9259-7

Keywords

Mathematics Subject Classification

Navigation