Skip to content
Licensed Unlicensed Requires Authentication Published by De Gruyter May 7, 2019

Convergence of the Chern–Moser–Beloshapka normal forms

  • Bernhard Lamel and Laurent Stolovitch

Abstract

In this article, we give a normal form for real-analytic, Levi-nondegenerate submanifolds of N of codimension d1 under the action of formal biholomorphisms. We find a very general sufficient condition on the formal normal form that ensures that the normalizing transformation to this normal form is holomorphic. In the case d=1 our methods in particular allow us to obtain a new and direct proof of the convergence of the Chern–Moser normal form.

Funding statement: Research of the authors was supported by ANR grant “ANR-14-CE34-0002-01” and the FWF (Austrian Science Foundation) grant I1776 for the international cooperation project “Dynamics and CR geometry”. Research of Bernhard Lamel was also supported by the Qatar Science Foundation, NPRP 7-511-1-098.

A Computations

We recall that Φp,0=Φ0,q=0. Therefore, (Q+Φ)l contains no terms (p,q) with p<l or q<l. As a consequence, we have

(A.1)(3.5)p,0=0,
(A.2)(3.5)p,1=ij<pDugp-jΦj,1+iDugp-1(u)Φ1,1,
(A.3)(3.5)2,2=iDug0(u)Φ2,2+iDug1(u)Φ1,2
+12Du2g0(u)(2Φ1,1Q+Φ1,12),
(A.4)(3.5)3,3=iDug0(u)Φ3,3+iDug1(u)Φ2,3+iDug2(u)Φ1,3
+12Du2g0(u)(2Φ2,2Q+{Φ2}3,3)
+12Du2g1(u)(2Φ1,2Q+{Φ2}2,3)
-i6Du3g0(u)(3Φ1,12Q+Φ1,13+3Φ1,1Q2),
(A.5)(3.5)3,2=iDug0(u)Φ3,2+iDug1(u)Φ2,2+iDug2(u)Φ1,2
+12Du2g0(u)(2Φ2,1Q+{Φ2}3,2)
+12Du2g1(u)(2Φ1,1Q+{Φ2}2,2),
(A.6)(3.5)3,1=iDug0(u)Φ3,1+iDug1(u)Φ2,1+iDug2(u)Φ1,1.

To obtain g¯3(z,u-iQ)-g¯3(z,u-iQ-iΦ), we just use the previous result and substitute gk in g¯k and i by -i. We have, using essentially the same computations,

(A.7)(3.6)p,1=(3.6)p,0=0,
(A.8)(3.6)2,2=Q(iDuf0(u)Φ2,1+iDuf1(u)Φ1,1,C¯z¯),
(A.9)(3.6)3,3=Q(iDuf0(u)Φ3,2+iDuf1(u)Φ2,2+iDuf2(u)Φ1,2,C¯z¯)
+12Q(Du2f0(u)(2Φ2,1Q+{Φ2}3,2)
+12Du2f1(u)(2Φ1,1Q+{Φ2}2,2),C¯z¯),
(A.10)(3.6)3,2=Q(iDuf0(u)Φ3,1+iDuf1(u)Φ2,1+iDuf2(u)Φ1,1,C¯z¯).

We have

Q(f2,f¯2)=k,l0ik+l(-1)lk!l!Q(Dukf2(z,u)(Q+Φ)k,Dulf¯2(z¯,u)(Q+Φ)l).

The function Dukfj(z,u)(Q+Φ)k (respectively Dulf¯j(z¯,u)(Q+Φ)l) has only terms (p,q) with pj+k and qk (respectively pl and ql+j). It follows that the function Q(Dukfj(z,u)(Q+Φ)k,Dulf¯j(z,u)(Q+Φ)l) contains only terms (p,q) with pj+k+l and qj+k+l. We have

(A.11)Q(f2,f¯2)p,0=Q(fp,f¯0),
(A.12)Q(f2,f¯2)p,1=Q(fp,f¯1)+iQ(Dfp-1(Q+Φ1,1)+Dufp-jΦj,1,f¯0)
-iQ(fp-1,Duf¯0(Q+Φ1,1)),
(A.13)Q(f2,f¯2)2,2=Q(f2,f¯2)+iQ(Df1(Q+Φ1,1),f¯1)
-iQ(f1,Df¯1(Q+Φ1,1))
-12(Q(f0,Du2f¯0(Q+Φ1,1)2)
+Q(Du2f0(Q+Φ1,1)2,f¯0))
-Q(Duf0(u)(Q+Φ1,1),Duf¯0(u)(Q+Φ1,1)),
(A.14)Q(f2,f¯2)3,3=Q(f3,f¯3)+iQ(Df0Φ3,1+Df1Φ2,1+Df2(Q+Φ1,1),f¯2)
-iQ(f2,Duf¯0Φ1,3+Duf¯1Φ1,2+Duf¯2(Q+Φ1,1))
+Q(i(Duf0Φ3,2+Duf1Φ2,2+Duf2(Q+Φ1,1)
-12(Du2f0(Q+Φ1,1)Φ2,1
+Du2f1(Q+Φ1,1)2)),f¯1)
+Q(f1,-i(Duf¯0Φ2,3+Duf¯1Φ2,2
+Duf¯2(Q+Φ1,1)
-12(Du2f¯0(Q+Φ1,1)Φ1,2
+D2f¯1(Q+Φ1,1)2)))
+Q(-i3Du3f0(Q+Φ1,1)3
+-12(Du2f0(Q+Φ1,1)Φ2,2
+Du2f1(Q,Φ1,1)Φ1,2),f¯0)
+Q(f0,i3Du3f¯0(Q+Φ1,1)3
+-12(Du2f¯0(Q+Φ1,1)Φ2,2
+Du2f¯1(Q,Φ1,1)Φ2,1))
+Q(-i(Duf0Φ3,3+Duf1Φ2,3+Duf2Φ2,3),f¯0)
+Q(f0,i(Duf¯0Φ3,3+Duf¯1Φ3,2+Duf¯2Φ3,2))
+-i2Q(Duf0(u)(Q+Φ1,1),Du2f¯0(u)(Q+Φ1,1)2)
+i2Q(Du2f0(u)(Q+Φ1,1)2,Duf¯0(u)(Q+Φ1,1))
+Q(iDuf1(z,u)(Q+Φ1,1),Duf¯1(z¯,u)(Q+Φ1,1)),
(A.15)Q(f2,f¯2)3,2=Q(f3,f¯2)-iQ(f2,Duf¯1(Q+Φ1,1)+Duf¯0Φ1,2)
-iQ(f1,Duf¯0(Q+Φ1,1)2+Duf¯1Φ2,1)
-iQ(f0,Duf¯1Φ3,1+Duf¯0Φ3,2)
+Q(Duf1(z,u)(Q+Φ1,1),Duf¯0(u)(Q+Φ1,1))
-12Q(Du2f1(Q+Φ1,1)2,f¯0).

We have, for α,βn and γd,

(A.16)Φ~3(f,f¯,12(g+g¯))-Φ~3(Cz,C¯z¯,su)
=|α|+|β|+|γ|=kk11α!β!γ!kΦ~3zαz¯βuγ(Cz,C¯z¯,su)f2αf¯2β(12(g3+g¯3))γ.

Hence, the (p,q)-term of Φ~3(f,f¯,12(g+g¯))-Φ~3(Cz,C¯z¯,su) is a sum of terms of the form

{kΦ~3zαz¯βuγ(Cz,C¯z¯,su)}p1,q1{f2α}p2,q2{f¯2β}p3,q3{(12(g3+g¯3))γ}p4,q4

with

i=14pi=p,i=14qi=q.

Let us first compute {f2α}p2,q2 with p2,q23. In the following computations, f,g are considered as vector valued functions except when computing fα,(g+g¯)γ, where f,g are considered as scalar functions and α,γ as an integers.

In the sums below, the terms appear with some positive multiplicity that we do not write since we are only interested in a lower bound of vanishing order of the terms. From these computations, we easily obtain {f¯2α}p2,q2 in the following way: replace fk by f¯k in the formula defining {fα}p,q in order to obtain the sum {f¯α}q,p. Furthermore, we have

{kΦ~3zαz¯βuγ}p1,q1=kΦ~p1+|α|,q1+|β|zαz¯βuγ.

Let us set as notation

Re(g):=g+g¯2
=g(z,u+i(Q(z,z¯)+Φ(z,z¯,u)))+g¯(z¯,u-i(Q(z,z¯)+Φ(z,z¯,u))2.

B Big Denominators theorem for nonlinear systems of PDEs

In this section we recall one of the main results of article [24] about local analytic solvability of some nonlinear systems of PDEs that have the “Big Denominators property”.

B.1 The problem

Let r* and 𝐦=(m1,,mr)r a fixed multiindex. Let us denote 𝔸nk (respectively (𝔸nk)>d, 𝔸nk^, (𝔸nk)(i) ) the space of k-tuples of germs at 0n (or n) of analytic functions (respectively vanishing at order d at the origin, formal power series maps, homogeneous polynomials of degree i) of n variables. Let us set

r,𝐦0:=(𝔸n)m1×(𝔸n)m2××(𝔸n)mr.

Given F=(F1,,Fr)r,𝐦0 and x(n,0), let us denote

jx𝐦F:=(jxm1F1,,jxmrFr),J𝐦r,𝐦0:={(x,jx𝐦F):x(n,0),Fr,𝐦0}.

Definition 9.

A map 𝒯:r,𝐦0𝔸ns is a differential analytic map of order m at the point 0𝔸nk if there exists an analytic map germ

W:(J𝐦r,𝐦0,0)s

such that 𝒯(F)(x)=W(x,jx𝐦F) for any xn close to 0 and any function germ Fr,𝐦0 such that j0mF is close to 0.

Denote by

v=(x1,,xn,uj,α),1jr,α=(α1,,αn)n,|α|mj,

the local coordinates in J𝐦𝔸nr, where uj,α corresponds to the partial derivative |α|x1α1xnαn of the j-th component of a vector function F𝔸nr. As usual, we have set

|α|=α1++αn.

Definition 10.

Let q be a nonnegative integer. Let 𝒯:r,𝐦0𝔸ns be a map.

  1. We shall say that it increases the order at the origin (respectively strictly) by q if for all (F,G)(r,𝐦0)2 then

    ord0(𝒯(F)-𝒯(G))ord0(F-G)+q,

    (respectively > instead of ).

  2. Assume that 𝒯 is an analytic differential map of order 𝐦 defined by a map germ

    W:(J𝐦r,𝐦0,0)s

    as in Definition 9. We call it regular if, for any formal map F=(F1,,Fr)^r,𝐦0, then

    ord0(Wiuj,α(x,F))pj,|α|,

    where

    (B.1)pj,|α|=max(0,|α|+q+1-mj).

    We have set F:=(|α|Fixα, 1ir, 0|α|mi).

Let us consider linear maps

  1. 𝒮:r,𝐦0𝔸ns that increases the order by q and is homogeneous, i.e. we have the inclusion 𝒮(r,𝐦(i))(𝔸ns)(q+i).

  2. π:𝔸nsImage(𝒮)𝔸ns is a projection onto Image(𝒮).

Let us consider a differential analytic map of order 𝐦, 𝒯:r,𝐦0𝔸ns.

We consider the equation

(B.2)𝒮(F)=π(𝒯(F)).

In [24], we gave a sufficient condition on the triple (𝒮,𝒯,π) under which equation (B.2) has a solution Fr,𝐦0; this condition is called the Big Denominators property of the triple (𝒮,𝒯,π) defined below.

B.2 Big Denominators. Main theorem

We can define the Big Denominators property of the triple (𝒮,𝒯,π) in equation (B.2).

Definition 11.

The triple of maps (𝒮,𝒯,π) of form (10) has Big Denominators property of order m if there exists an nonnegative integer q such that the following holds:

  1. 𝒯 is an regular analytic differential map of order 𝐦 that strictly increases the order by q and j0q-1𝒯(0)=0, i.e. 𝒯(F)(x)=W(x,jx𝐦F) for any xn close to 0 and any function germ Fr,𝐦0 such that j0mF is close to 0 and ord0(W(x,0))q.

  2. 𝒮:r,𝐦0𝔸ns is linear, increases the order by q and is homogeneous, i.e.

    𝒮(r,𝐦(i))(𝔸ns)(q+i).
  3. The linear map π:𝔸nsImage(𝒮)𝔸ns is a projection.

  4. The map 𝒮 admits right-inverse 𝒮-1:Image(S)𝔸nr such that the composition 𝒮-1π satisfies: there exists C>0 such that for any G𝔸ns of order >q, one has for all 1jr, and all integers i,

    (B.3)(𝒮j-1π(G))(i+mj)CG(i+q)(i+mj+q)(i+q+1),

    where 𝒮i-1 denotes the i-th component of 𝒮-1, 1ir.

Remark 12.

Let i0 and let F=(F1,,Fk)(𝔸nk)(i). Let

Fj=Fj,αxα,

where the sum is taken over all j=1,,k and all multiindices α=(α1,,αn) such that |α|=α1++αn=i. The norm F used in (B.3) is either

Fj=|α|=i|Fj,α|,F=max(F1,,Fk)

or the modified Fisher–Belitskii norm

Fj2=|α|=iα!|α|!|Fj,α|2,F2=F12++Fk2.

Remark 13.

In practice, for each i, there is a decomposition into direct sums

r,𝐦(i)=LiKi

with 𝒮|Li is a bijection onto its range. The chosen right inverse is then the one with zero component along Ki. For instance, the case of the modified Fisher–Belitskii norm, Ki:=ker𝒮i* is the natural one, where 𝒮i* denotes the adjoint of 𝒮i with respect to the scalar product.

Theorem 14 ([24, Theorem 7]).

Let us consider a system of analytic nonlinear PDEs such as equation (B.2):

(B.4)𝒮(F)=π(W(x,jx𝐦F)).

If the triple (S,T,π) has the Big Denominators property of order m, according to Definition 11, then the equation has an analytic solution FFr,m0.

Remark 15.

The precise statement of [24, Theorem 7] holds for Fr,𝐦>0 and where the order of W(x,0) at the origin is greater than q. The shift by 1 (i.e. Fr,𝐦0 and where the order of W(x,0) at the origin is greater than or equal to q) of the above statement, does not affect its proof.

B.3 Application

In this subsection we shall devise the strictly increasing condition in more detail. We look for a formal solution F0=i0F(i) to (B.4). As above, F(i) stands for (F1(m1+i),,Fr(mr+i)). We define

𝒮(F(i+1)):=[πW(x,jx𝐦j0iF(j))](i+q+1).

Here [G](i) denotes the homogeneous part of degree i of G in the Taylor expansion at the origin. Therefore F:=iF(i) is a solution of (B.4) if

(B.5)ord0(W(x,jx𝐦j0F(j))-W(x,jx𝐦j0iF(j)))>i+q+1.

Indeed, we would have

𝒮(i0F(i))=i0[πW(x,jx𝐦j0iF(j))](i+q+1)
=i0[πW(x,jx𝐦j0F(j))](i+q+1)
=πW(x,jx𝐦F).

We emphasize that condition (B.5) just means that W strictly increases the order by q as defined in Definition 10. Let us look closer to that condition. Let us denote

Fi:=j0iF(j)andF>i:=j>iF(j).

Let us Taylor expand W(x,jx𝐦F) at Fi. We thus have

W(x,jx𝐦F)-W(x,jx𝐦Fi)=Wuj,α((x,jx𝐦Fi))|α|Fj>ixα
+12Wuj,αuj,α((x,jx𝐦Fi))|α|Fj>ixα|α|Fj>ixα
+.

We recall that ord0Fj>i>mj+i and when considering a coordinate uj,α, we have |α|mj. Hence, we have

ord0|α|Fj>ixα>mj+i-|α|.

In order that the first derivative part of this Taylor expansion satisfies (B.5), it is sufficient that

ord0Wuj,α((x,jx𝐦Fi))|α|-mj+q+1.

This is nothing but the regularity condition as defined in Definition 10. Let us consider the other terms in the Taylor expansion. We have, for instance,

ord0|α|Fj>ixα|α|Fj>ixαmj+i+1-|α|+mj+i+1-|α|.

If i+1>q, then not only the second but also any higher order derivative part of this Taylor expansion satisfies (B.5).

Corollary 16.

If q=0 and if the system is regular, it strictly increases the order by 0.

Acknowledgements

We thank the referees for their very careful reading that helped us improve our text.

References

[1] M. S. Baouendi, P. Ebenfelt and L. P. Rothschild, CR automorphisms of real analytic manifolds in complex space, Comm. Anal. Geom. 6 (1998), no. 2, 291–315. 10.4310/CAG.1998.v6.n2.a3Search in Google Scholar

[2] M. S. Baouendi, P. Ebenfelt and L. P. Rothschild, Real submanifolds in complex space and their mappings, Princeton Math. Ser. 47, Princeton University Press, Princeton 1999. 10.1515/9781400883967Search in Google Scholar

[3] M. Beals, C. Fefferman and R. Grossman, Strictly pseudoconvex domains in 𝐂n, Bull. Amer. Math. Soc. (N. S.) 8 (1983), no. 2, 125–322. 10.1090/S0273-0979-1983-15087-5Search in Google Scholar

[4] G. R. Belitskii, Invariant normal forms of formal series, Funct. Anal. Appl. 13 (1979), no. 1, 46–47. 10.1007/BF01076439Search in Google Scholar

[5] V. Beloshapka, Construction of the normal form of the equation of a surface of high codimension, Math. Not. Acad. Sci. USSR 48 (1990), no. 2, 721–725. 10.1007/BF01262601Search in Google Scholar

[6] A. Čap and J. Slovák, Parabolic geometries. I, Math. Surveys Monogr. 154, American Mathematical Society, Providence 2009. 10.1090/surv/154Search in Google Scholar

[7] E. Cartan, Sur la géométrie pseudo-conforme des hypersurfaces de l’espace de deux variables complexes II, Ann. Sc. Norm. Super. Pisa Cl. Sci. (2) 1 (1932), no. 4, 333–354. Search in Google Scholar

[8] E. Cartan, Sur la géométrie pseudo-conforme des hypersurfaces de l’espace de deux variables complexes, Ann. Mat. Pura Appl. (2) 11 (1933), no. 1, 17–90. 10.1007/BF02417822Search in Google Scholar

[9] S. S. Chern and J. K. Moser, Real hypersurfaces in complex manifolds, Acta Math. 133 (1974), 219–271. 10.1142/9789812812834_0027Search in Google Scholar

[10] V. V. Ežov and G. Schmalz, Normal form and two-dimensional chains of an elliptic CR manifold in 𝐂4, J. Geom. Anal. 6 (1996), no. 4, 495–529. 10.1007/BF02921621Search in Google Scholar

[11] V. V. Ežov and G. Schmalz, Special normal form of a hyperbolic CR-manifold in 𝐂4, Ann. Polon. Math. 70 (1998), 99–107. 10.4064/ap-70-1-99-107Search in Google Scholar

[12] E. Fischer, Über die Differentiationsprozesse der Algebra, J. reine angew. Math. 148 (1918), 1–78. 10.1515/crll.1918.148.1Search in Google Scholar

[13] X. Gong and J. Lebl, Normal forms for CR singular codimension-two Levi-flat submanifolds, Pacific J. Math. 275 (2015), no. 1, 115–165. 10.2140/pjm.2015.275.115Search in Google Scholar

[14] X. Gong and L. Stolovitch, Real submanifolds of maximum complex tangent space at a CR singular point. I, Invent. Math. 206 (2016), no. 2, 293–377. 10.1007/s00222-016-0654-8Search in Google Scholar

[15] X. Huang, Local equivalence problems for real submanifolds in complex spaces, Real methods in complex and CR geometry, Lecture Notes in Math. 1848, Springer, Berlin (2004), 109–163. 10.1007/978-3-540-44487-9_3Search in Google Scholar

[16] X. Huang and W. Yin, A Bishop surface with a vanishing Bishop invariant, Invent. Math. 176 (2009), no. 3, 461–520. 10.1007/s00222-008-0167-1Search in Google Scholar

[17] X. Huang and W. Yin, Flattening of CR singular points and analyticity of the local hull of holomorphy. I, Math. Ann. 365 (2016), no. 1–2, 381–399. 10.1007/s00208-015-1228-6Search in Google Scholar

[18] X. Huang and W. Yin, Flattening of CR singular points and analyticity of the local hull of holomorphy. II, Adv. Math. 308 (2017), 1009–1073. 10.1016/j.aim.2016.12.008Search in Google Scholar

[19] H. Jacobowitz, An introduction to CR structures, Math. Surveys Monogr. 32, American Mathematical Society, Providence 1990. 10.1090/surv/032Search in Google Scholar

[20] E. Lombardi and L. Stolovitch, Normal forms of analytic perturbations of quasihomogeneous vector fields: Rigidity, invariant analytic sets and exponentially small approximation, Ann. Sci. Éc. Norm. Supér. (4) 43 (2010), no. 4, 659–718. 10.24033/asens.2131Search in Google Scholar

[21] J. K. Moser and S. M. Webster, Normal forms for real surfaces in 𝐂2 near complex tangents and hyperbolic surface transformations, Acta Math. 150 (1983), no. 3–4, 255–296. 10.1007/BF02392973Search in Google Scholar

[22] H. Rüssmann, Über die Normalform analytischer Hamiltonscher Differentialgleichungen in der Nähe einer Gleichgewichtslösung, Math. Ann. 169 (1967), 55–72. 10.1007/BF01399531Search in Google Scholar

[23] G. Schmalz and A. Spiro, Explicit construction of a Chern–Moser connection for CR manifolds of codimension two, Ann. Mat. Pura Appl. (4) 185 (2006), no. 3, 337–379. 10.1007/s10231-005-0156-6Search in Google Scholar

[24] L. Stolovitch, Big denominators and analytic normal forms. With an appendix of Michail Zhitomirskii, J. reine angew. Math. 710 (2016), 205–249. 10.1515/crelle-2013-0111Search in Google Scholar

[25] N. Tanaka, On the pseudo-conformal geometry of hypersurfaces of the space of n complex variables, J. Math. Soc. Japan 14 (1962), 397–429. 10.2969/jmsj/01440397Search in Google Scholar

[26] A. G. Vitushkin, Holomorphic mappings and the geometry of surfaces, Current problems in mathematics. Fundamental directions. Vol. 7, Itogi Nauki i Tekhniki, Akad. Nauk SSSR, Moscow (1985), 167–226, 258. Search in Google Scholar

[27] A. G. Vitushkin, Real-analytic hypersurfaces of complex manifolds, Uspekhi Mat. Nauk 40 (1985), no. 2(242), 3–31, 237. 10.1070/RM1985v040n02ABEH003556Search in Google Scholar

Received: 2018-05-24
Revised: 2019-03-15
Published Online: 2019-05-07
Published in Print: 2020-08-01

© 2021 Walter de Gruyter GmbH, Berlin/Boston

Downloaded on 19.4.2024 from https://www.degruyter.com/document/doi/10.1515/crelle-2019-0004/html
Scroll to top button