Abstract
We show a global existence for the Cauchy problem with large initial data for the p-harmonic flow between two smooth, compact Riemannian manifolds. We devise new monotonicity type formulas of a local scaled energy and establish a partial regularity for the solution. The partial regularity obtained is almost optimal, comparing with that of the corresponding stationary case. The p-harmonic flow obtained also converges to a p-harmonic map along a certain time sequence tending to infinity.
Similar content being viewed by others
References
Ambrosio, L., Mantegazza, C.: Curvature and distance function from a manifold. J. Geom. Anal. 8(5), 723–748 (1998) (Dedicated to the memory of Fred Almgren)
Ambrosio, L., Soner, H.M.: A level set approach to the evolution of surfaces of any codimension. J. Differ. Geom. 43, 693–737 (1996)
Chang, K.-C.: Heat flow and boundary value problem for harmonic maps. Ann. Inst. Henri Poincaré, Anal. Non Linéaire 6(5), 363–395 (1989)
Chang, K.-C., Ding, W.-Y., Ye, R.: Finite-time blow up of the heat flow of harmonic maps from surfaces. J. Differ. Geom. 36(2), 507–515 (1992)
Chen, C.-N., Cheung, L.F., Choi, Y.S., Law, C.K.: On the blow-up of heat flow for conformal \(3\)-harmonic maps. Trans. AMS 354(12), 5087–5110 (2002)
Chen, Y.-M., Hong, M.-C., Hungerbuhler, N.: Heat flow of \(p\)-harmonic maps with values into spheres. Math. Z. 215, 25–35 (1994)
Chen, Y.-M., Ding, W.-Y.: Blow-up and global existence for heat flows of harmonic maps. Invent. Math. 99(3), 567–578 (1990)
Chen, Y.-M., Lin, F.H.: Evolution of harmonic maps with Dirichlet boundary conditions. Commun. Anal. Geom. 1(3–4), 327–346 (1993)
Chen, Y.-M., Struwe, M.: Existence and partial regularity results for the heat flow for harmonic maps. Math. Z. 201, 83–103 (1989)
Choe, H.J.: Hölder continuity of solutions of certain degenerate parabolic systems. Nonlinear Anal. 8(3), 235–243 (1992)
Coron, J.M., Ghidaglia, J.M.: Explosion en temps fini pour le flot des applications harmoniques. C. R. Acad. Sci. Paris Ser. I(308), 339–344 (1989)
DiBenedetto, E.: Degenerate Parabolic Equations. Springer, New York (1993)
Duzaar, F., Fuchs, M.: On removable singularities of \(p\)-harmonic maps. Ann. Inst. Henri Poincaré, Anal. Non Linéaire 7(5), 385–405 (1990)
Duzaar, F., Mingione, G.: The \(p\)-harmonic approximation and the regularity of \(p\)-harmonic maps. Calc. Var. Partial Differential Equations 20, 235–256 (2004)
Eells, J., Sampson, J.H.: Harmonic mappings of Riemannian manifolds. Am. J. Math. 86, 109–169 (1964)
Fardoun, A., Regbaoui, R.: Heat flow for \(p\)-harmonic maps between compact Riemannian manifolds. Indiana Univ. Math. J. 51(6), 1305–1320 (2002)
Giaquinta, M.: Multiple Integrals in the Calculus of Variations and Nonlinear Elliptic Systems. Annals of Mathematics Studies, vol. 105. Princeton University Press, Princeton (1983)
Giaquinta, M., Hildebrandt, S.: A priori estimates for harmonic mappings. J. Reine Angew. Math. 336, 123–164 (1982)
Giaquinta, M., Struwe, M.: On the partial regularity of weak solutions of nonlinear parabolic systems. Math. Z. 179, 437–451 (1982)
Giaquinta, M., Struwe, M.: An optimal regularity result for a class of quasilinear parabolic systems. Manuscr. Math. 36, 223–240 (1981)
Giusti, E.: Direct Methods in the Calculus of Variations. World Scientific, Singapore (2005)
Grotowski, J.F.: Finite time blow-up for the harmonic map heat flow. Calc. Var. Partial Differ. Equ. 1(2), 231–236 (1993)
Hamilton, R.: Harmonic Maps of Manifolds with Boundary \(m\)-Harmonic Flow. Lecture Notes in Mathematics, vol. 471, pp. 593–631. Springer, Berlin (1975)
Hungerbühler, N.: \(m\)-harmonic flow. Ann. Scuola Norm. Sup. Pisa CI. Sci. 24, 593–631 (1997)
Hungerbühler, N.: Global weak solutions of the \(p\)-harmonic flow into homogeneous spaces. Indiana Univ. Math. J. 45(1), 275–288 (1996)
Karim, C., Misawa, M.: Gradient Hölder regularity for nonlinear parabolic systems of \(p\)-Laplacian type. Differ. Integral Equ. 29(3–4), 201–228 (2016)
Leone, C., Misawa, M., Verde, A.: A global existence result for the heat flow of higher dimensional \(H\)-systems. J. Math. Pures Appl. (9) 97(3), 282–294 (2012)
Leone, C., Misawa, M., Verde, A.: The regularity for nonlinear parabolic systems of \(p\)-Laplacian type with critical growth. J. Differ. Equ. 256, 2807–2845 (2014)
Misawa, M.: Approximation of \(p\)-harmonic maps by the penalized equation. Nonlinear Anal. 47, 1069–1080 (2011)
Misawa, M.: Local Hölder regularity of gradients for evolutional \(p\)-Laplacian systems. Ann. Mat. Pura Appl. (IV) 181, 389–405 (2002)
Misawa, M.: Existence and regularity results for the gradient flow for \(p\)-harmonic maps. Electron J. Differ. Equ. 36, 1–17 (1998)
Misawa, M.: On the \(p\)-harmonic flow into spheres in the singular case. Nonlinear Anal. Ser. A Theory Methods 50(4), 485–494 (2002)
Misawa, M.: Local regularity and compactness for the \(p\)-harmonic map heat flows. Adv. Calc. Var. (2017). https://doi.org/10.1515/acv-2016-0064
Misawa, M.: Regularity for the evolution of \(p\)-harmonic maps. J. Differ. Equ. 264, 1716–1749 (2018)
Saloff-Coste, L.: Aspects of Sobolev-Type Inequalities. Lecture Note Series, vol. 289. London Mathematical Society
Schoen, R.: Analytic aspects of the harmonic map problem. In: Chern, S.S. (ed.) Seminar on Nonlinear Partial Differential Equations. MSRI Publications, vol. 2, pp. 321–358. Springer, New-York (1984)
Struwe, M.: On the Hölder continuity of bounded weak solutions of quasilinear parabolic systems. Manuscr. Math. 35, 125–145 (1981)
Struwe, M.: On the evolution of harmonic maps of Riemannian surfaces. Comment. Math. Helv. 60(4), 558–581 (1985)
Struwe, M.: On the evolution of harmonic maps in higher dimensions. J. Differ. Geom. 28, 485–502 (1988)
Wang, C.: Limits of solutions to the generalized Ginzburg–Landau functional. Commun. Partial Differ. Equ. 27(5–6), 877–906 (2002)
Acknowledgements
I would like to record here my sincere thanks to the referee for kindly reading this long paper and giving some corrections.
Funding
The work by M. Masashi was partially supported by the Grant-in-Aid for Scientific Research (C) No. 15K04962 and No. 18K03375 at Japan Society for the Promotion of Science.
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflicts of interest
The author declares that it has no conflict of interest.
Additional information
Communicated by M. Struwe.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
The work by M. Masashi was partially supported by the Grant-in-Aid for Scientific Research (C) No. 15K04962 at Japan Society for the Promotion of Science.
6. Appendix
6. Appendix
1.1 Appendix A A global existence and regularity of a weak solution of (2.1).
Proof od Lemma 6
We use the Galerkin method and the monotonicity trick for p-Laplace operator to solve the Cauchy problem (2.1). The proof is standard and we can refer to [6, Theorem 1.5 and its proof, pp. 29–31].
Regularity of a weak solution. Let \(u = u_{K, \, \epsilon }\) be a weak solution of (2.1). The lower-order term is bounded by the definition of \(\chi \) as
and thus, we can apply the Hölder regularity result for the evolutionary p-Laplace operator in [12, Theorem \(1.1^\prime \), p. 256] (also see [26]) to find that the solution u and its gradient are locally Hölder continuous on \({\mathcal{M}}_\infty \). We also have that the second derivative is integrable : \(\big (\epsilon + |D u|^2\big )^{\frac{p - 2}{2}} |D^2 u|^2\) is locally integrable in \({\mathcal{M}}_\infty \) and that the gradient Du is locally bounded in \({\mathcal{M}}_\infty \) (see [12, Proposition 3.1, p. 223; Theorem 5.1 , p. 238]. Then, expanding the principal part of the p-Laplace operator, the solution u is also satisfies the linear parabolic systems with Hölder continuous coefficients and lower-order terms almost everywhere. Thus, it follows from the Schauder regularity theory that u, Du, \(D^2 u\) and \(\partial _t u\) are locally Hölder continuous in \({\mathcal{M}}_\infty \). \(\square \)
1.2 Appendix B Energy inequality and maximum principle
We present the proof of Lemmata 7 and 8.
Proof of Lemma 7
The energy inequality (3.1) is shown to be valid in the proof of Lemma 6. However, as a priori estimates for regular solutions of (2.1), we naturally multiply (2.1) by \(\partial _t u \, \sqrt{|g|}\) and integrate by parts on space variable in \({\mathcal{M}}_T\) for any \(T > 0\). \(\square \)
Proof of Lemma 8
We multiply (2.1) by \(\sqrt{|g|} u \big (|u|^2 - H^2\big )_+\) and integrate in \({\mathcal{M}}_\infty \), where \(\big (f\big )_+\) is the positive part of a function f. Since the support of \(\chi ^\prime \) is in \(\mathcal{O}_{2 \delta _{\mathcal{N}}} ({\mathcal{N}}) \subset B (H)\), \(\chi ^\prime (\mathrm{dist}^2 (u, \, {\mathcal{N}}))\) is zero in \(\mathrm{I}\!\mathrm{R}^l {\setminus } B (H, \, 0)\). Also \(u_0 \in {\mathcal{N}} \subset B (H)\). Hence, we have
and thus, \(|u (t)| \le H\) in \({\mathcal{M}}\) and any \(t \ge 0\). \(\square \)
1.3 Appendix C Proof of the Bochner estimate.
Proof of Lemma 9
In the proof, for brevity, let the regularized p-energy density be
In the general case in \({\mathcal{M}}\), the terms containing the spatial derivative of \(g_{\alpha \beta }\) only appear and are bounded by \(C \, (\epsilon + |D u|^2)^{p - 1}\). In fact, in (6.5) below, by a direct computation, we have the terms with derivatives of the metric
which are bounded by such terms as
with a positive constant C depending only on \(\mathcal{M}\), p and m. Here the 1st term is controllable lower-order one. By Cauchy’s inequality with \(c > 0\), the 2nd term are estimated above as
of which the 1st term with a small \(c >0\) in the right hand side is abosrbed into the squared 2nd derivative term of the solution in (6.5) below, and the 2nd term is a controllable one. The controllable terms \(C \, f\) above are multiplied by \((p f)^{1 - \frac{2}{p}}\) in (6.6) below, and thus, becomes \(C \, f^{2 (1 - \frac{1}{p})}\).
Hereafter, we assume that the metric \(g = \left( g_{\alpha \beta }\right) \) is the identity matrix.
Since u, Du and \(D^2 u\) are continuous in \(\mathrm{I}\!\mathrm{R}^m_\infty \), it holds in the distribution sense that
Hereafter the summation convention over repeated indices is used. Since \(\chi \big ({ \text{ dist }}^2 (u, \, {\mathcal{N}})\big ) = 2 (\delta _{\mathcal{N}})^2\) for \(u \in \mathrm{I}\!\mathrm{R}_l {\setminus } \mathcal{O}_{2 \delta _{\mathcal{N}}}\), \(D \Big (D_u \chi \big ({ \text{ dist }}^2 (u, \, {\mathcal{N}})\big )\Big ) = 0\) if \( \text{ dist } (u, \, {\mathcal{N}}) > 2 \delta _{\mathcal{N}}\) and then, we have (3.4) by (2.1). We treat the case that \( \text{ dist } (u, \, {\mathcal{N}}) \le 2 \delta _{\mathcal{N}}\). Noting that \(\chi \big ({ \text{ dist }}^2 (u, \, {\mathcal{N}})\big )\) is smooth, by a direct calculation we have
where the arguments in \(\chi ^\prime \) are omitted. By (6.3) and (6.4) with (2.1), we have
Furthermore, multiplying (6.5) by \((p \, f)^{1 - \frac{2}{p}}\), we obtain
where we use the fact that
By differentiation of the penalty term
and (6.4), we have
where in particular, multiplying (2.1) by the derivative of penalty term we compute
By the support of \(\chi ^{\prime \prime }\), we have
and thus, the 2nd terms in the 2nd line of (6.6) and the 3rd line of (6.7), and the 3rd term in the 3rd line of (6.8) are estimated above by
with a positive constant C depending only on p and \(\chi \), because
By Schwarz’s and Cauchy’s inequalities, the terms in the 3rd lines of (6.6) and in the 4th line of (6.7), (6.8) are bounded by
where by a positive constant \(C ({\mathcal{N}})\) depending on a bound for the curvature of \({\mathcal{N}}\), we have the boundedness for any \(u \in {\mathcal{N}}\)
of which the validity will be shown later.
The terms in the 5th line of (6.7) are bounded by
Gathering (6.9), (6.10) and (6.12) in (6.6) and (6.7), respectively, we obtain
and thus, from (6.13), the desired inequality (3.4) is obtained, if the constant \(C_0\) is so large that
We present the proof of (6.11). We follow the argument as in [2, Theorems 3.1 and 3.2, their proofs, pp. 704–707] (also refer to [1, Theorem 2.2]). \(\square \)
Lemma 21
There exists a positive constant C depending only on a bound of curvatures of \(\mathcal{N}\) such that, for any \(u \in \mathcal{O}_{2 \delta _{\mathcal{N}}}\) and \(q \in \mathrm{I}\!\mathrm{R}^l \cong \mathcal{T}_u \mathrm{I}\!\mathrm{R}^l\),
Proof
For any \(u \in \mathcal{O}_{2 \delta _{\mathcal{N}}}\) such that \(u \notin {\mathcal{N}}\), we make parallel transformation with the direction \(\big (u - \pi _{\mathcal{N}} (u)\big )/\big |u - \pi _{\mathcal{N}} (u)\big |\) and follow the following argument. Therefore, we treat the case that \(u \in {\mathcal{N}}\) and thus, \(\pi _{\mathcal{N}} (u) = u\). For any \(v \in \mathcal{O}_{2 \delta _{\mathcal{N}}}\) let us put
We know that the squared distance function \(\eta (v)\) is smooth on \(v \in \mathcal{O}_{2 \delta _{\mathcal{N}}}\). Let \(q \in \mathrm{I}\!\mathrm{R}^l\) be any vector in \(\mathrm{I}\!\mathrm{R}^l\) and be fixed. Under the orthogonal decomposition of \(\mathrm{I}\!\mathrm{R}^l\) with respect to the tangent space \(\mathcal{T}_u {\mathcal{N}}\) at \(u \in {\mathcal{N}}\), \(\mathrm{I}\!\mathrm{R}^l = \mathcal{T}_u {\mathcal{N}} \oplus (\mathcal{T}_u {\mathcal{N}})^\bot \), we set as
where p is the unit normal vector along the normal component of q and f(t) is the distance to \(\mathcal{N}\) measured along p. Then, we compute as
and, also, for any \(v \in \mathcal{O}_{2 \delta _{\mathcal{N}}}\), \(v \notin {\mathcal{N}}\),
Thus, letting, for any \(t \in (0, \, 2 \delta _{\mathcal{N}}]\),
we have
where by (6.17) we have, as \(t \searrow 0\),
and the 1st convergence is valid because \(D_{v^i} D_{v^j} \eta (u) = \left. D_{v^i} D_{v^j} \eta (v)\right| _{v = u}\) is the orthogonal projection on \((\mathcal{T}_u {\mathcal{N}})^\bot \) (see [2, Theorem 3.1, p.704]). Therefore, from l’Hospital’s theorem, we obtain
where we use that \(\displaystyle { p \cdot D_v d (u + t p) = \big |p\big |^2 = 1 }\) and \(\displaystyle { D_v (p \cdot D_u d (u + t p)) = 0. }\) Thus, we have
where in the last inequality the positive constant C depends only on a bound of curvatures of \(\mathcal{N}\) (see [2, Remark 3.3; Theorem 3.5, its proof, pp. 707–709]). \(\square \)
1.4 Appendix D Proof of the gradient boundedness.
Here we demonstrate the proof of Lemma 10, relying on Moser’s iteration method as usual. Such estimate has been originally done for the evolutionary p-Laplacian system with controllable growth lower-order terms, by DiBenedetto, developing the intrinsic scaling transformation to the evolutionary p-Laplace operator (refer to [10, 12]).
However, the emphasis here is to make localization by use of the cut-off function \(\mathcal C\).
Proof of Lemma 10
In the following we use the same notation as in Lemma 10.
By use of a scaling transformation intrinsic to the evolutionary p-Laplace operator
we now rewrite (3.4) in Lemma 9 by the scaled solution v on \(Q (1, \, 1) : = Q (1, \, 1) (0)\),
satisfying in \(Q (1, \, 1)\)
We put the notation
As in “Appendix C”, we assume that the metric \((g_{\alpha \beta })\) is the identity matrix and, in the general case in \({\mathcal{M}}\), the terms with the derivative of \(g_{\alpha \beta } (x_0 + r y)\) appear and are bounded by \(C \, f (v)^{2 (1 - \frac{1}{p})}\), as in (6.1) and (6.2), where we note that \(D_y g^{\alpha \beta } (x_0 + r y) = r D g^{\alpha \beta } (x)\) and \(r \le 1\). We proceed to the same computation as (6.6) and (6.7), where the quantities appeared are transformed to the corresponding ones, respectively, defined by the scaled solution v as above. Now we will look at the transformed estimation for the scaled solution v.
By the support of \(\chi ^{\prime \prime }\), we have
and thus, the corresponding terms for the scaled solution v to 2nd terms in the 2nd line of (6.6) and the 3rd line of (6.7), and the 3rd term in the 3rd line of (6.8) are estimated above by
with a positive constant C depending only on p and \(\chi \), because, by the definition of \(r_0\),
By Schwarz’s and Cauchy’s inequality, the terms in the 3rd line of (6.6) and in the 4th line of (6.7), (6.8) are estimated as
where by the definition of \(r_0\) as before, and a positive constant \(C ({\mathcal{N}})\) depending on a bound for the curvature of \({\mathcal{N}}\), we compute as
The terms in 5th line of (6.7) are bounded by
Gathering all of the estimations above yields
Finally we make Moser’s iteration estimate by (6.24) and scaling back to have (3.6). Now, taking care of localization by the cut off function \(\mathcal C\), we proceed to the estimations.
Let \(B (\rho ) = B (\rho , \, 0)\) be a ball in \(\mathrm{I}\!\mathrm{R}^m\) with radius \(\rho \le \min \{1, \, R_{\mathcal{M}}/2, \, T^{1/\lambda _0}\}\) and center of origin. Let \(0< r < \rho \). We use local parabolic cyllinders \(Q (r^2, \, r) = (- r^2, 0) \times B (r)\) and \(Q (\rho ^2, \, \rho ) = (- \rho ^2, 0) \times B (\rho )\), Let \(\eta \) be a smooth real-valued function on \(\mathrm{I}\!\mathrm{R}^m\) such that \(0 \le \eta \le 1\), the support of \(\eta \) is contained in \(B (\rho )\) and \(\eta = 1\) on B(r). Let \(\sigma = \sigma (t)\) be a smooth real-valued function on \(\mathrm{I}\!\mathrm{R}\) such that \(0 \le \sigma \le 1\), \(\sigma = 1\) on \([- r^2, \, \infty )\) and \(\sigma = 0\) on \((- \infty , \, - \rho ^2]\). We denote by the original notation the scaled function under (6.19). Put
and also write as \(z = (s, \, y) \in Q (1, 1)\) and \(d z = d {\mathcal{M}} d s\).
Put \(w = e (v)\) in the Bochner type estimate (6.24). Let \(\alpha \) be nonnegative number and use the test function \(w^\alpha \eta ^2 \sigma \, \mathcal{C}^q \, \sqrt{|g|}\) in the weak form of (6.24). After a routine computation we have the so-called reverse Poincaré inequality
where we compute as
Applying the Sobolev embedding \(W^{1, 2}_0 (B (\rho )) \rightarrow L^{2 m/(m - 2)} (B (\rho ))\) we have
which is combined with (6.25) and yields
By Hölder’s inequality and (6.26) we compute as
where we use a simple inequality valid for \(\alpha \ge 0\)
and also estimate the derivative of \(\mathcal{C}\) as
because, by the range Q(1, 1) of \((s, \, y)\) and the condition (3.5) of \(r_0\),
and so, we have the estimations
We arrange some terms in an appropriate way to have
Here let \(\{\rho _k\}\) be a sequence of radii, defined as
and \(\{\alpha _k\}\) be a sequence of exponents
We choose \(r = \rho _{k + 1}\), \(\rho = \rho _k\) and \(\alpha = \alpha _k\) in (6.27) and make routine computation to have
which is computed by sequences (6.28) and (6.29) as
An iterative application of (6.31) yields, as \(k \rightarrow \infty \),
where we use the relation of exponents
and the limit as \(k \rightarrow \infty \)
Finally, scaling back in (6.32) yields the desired estimate (3.6). \(\square \)
Rights and permissions
About this article
Cite this article
Misawa, M. Global existence and partial regularity for the p-harmonic flow. Calc. Var. 58, 54 (2019). https://doi.org/10.1007/s00526-019-1500-9
Received:
Accepted:
Published:
DOI: https://doi.org/10.1007/s00526-019-1500-9