Home Mathematics A Frobenius–Nirenberg theorem with parameter
Article
Licensed
Unlicensed Requires Authentication

A Frobenius–Nirenberg theorem with parameter

  • Xianghong Gong ORCID logo
Published/Copyright: January 17, 2018

Abstract

The Newlander–Nirenberg theorem says that a formally integrable complex structure is locally equivalent to the complex structure in the complex Euclidean space. We will show two results about the Newlander–Nirenberg theorem with parameter. The first extends the Newlander–Nirenberg theorem to a parametric version, and its proof yields a sharp regularity result as Webster’s proof for the Newlander–Nirenberg theorem. The second concerns a version of Nirenberg’s complex Frobenius theorem and its proof yields a result with a mild loss of regularity.

A Hölder norms for functions with parameter

The main purpose of the appendix is to derive some interpolation properties for Hölder norms defined in domains depending on a parameter. The interpolation properties were derived in Hörmander [18] and Gong and Webster [10].

We say that a domain D in m has the cone property if the following hold:

  1. Given two points p0,p1 in D, there exists a piecewise 𝒞1 curve γ(t) in D such that γ(0)=p0 and γ(1)=p1, |γ(t)|C*|p1-p0| for all t except finitely many values. The diameter of D is less than C*.

  2. For each point xD¯, D contains a cone V with vertex x, opening θ>C*-1 and height h>C*-1.

We will denote by C*(D) a constant C*>1 satisfying (i) and (ii). A constant Ca(D) may also depend on C*(D). In our applications, we will apply the inequalities to domains that are products of balls of which the radii are between two fixed numbers. Therefore, C*(D) and Ca(D) do not depend on D, which will be assumed in the appendix.

If D is a domain of the cone property, then the norms on D¯ satisfy

(A.1)|u|D;(1-θ)a+θbCa,b|u|D;a1-θ|u|D;bθ
|f1|a1+b1|f2|a2+b2Ca,b(|f1|a1+b1+b2|f2|a2+|f1|a1|f2|a2+b1+b2).

Here |fi|ai=|fi|Di;a. Throughout the appendix, we always assume that domains D,D,Di have the cone property.

Let |ui|ai=|ui|Di;ai. Then we have

(A.2)j=1m|uj|dj+ajCamj=1m|uj|dj+a1++amij|ui|di.

Let f={fλ}, g={gλ} be two families of functions on D¯ and D¯, respectively. To deal with two Hölder exponents in the variables x,t, we introduce the following notation:

|f|D;a,0|g|D;0,b:=|f|D;a,0|g|D;0,[b]+|f|D;[a],0|g|D;0,b,
QD,D;a,b*(f,g):=|f|D;a,0|g|D;0,b+|f|D;0,b|g|D;a,0

and

(A.3)QD,D;r,s(f,g):=j=0sQD,D;r-j,j+{s}*(f,g),[r][s],
Q^D,D;r,s(f,g):=fD;r,s|g|D;0,0+|f|D;0,0gD;r,s+QD,D;r,s(f,g).

For simplicity, the dependence of Q,Q*,Q^ on the domains D,D is not indicated when it is clear from the context.

Throughout the paper, by AB we mean that ACB for some constant C.

Lemma A.1.

Let ri,si,ai,bi, where 1im, be non-negative real numbers. Assume that (r1,,rm,a1,,am)N2m or (s1,,sm,b1,,bm)N2m. Let Di be a domain in Rni with the cone property and let fi={fiλ} be a family of functions on Di. Let

|fi|c,d=|fi|Di;c,d.

Assume that >m2. Then

(A.4)i=1m|fi|ri,siCr+sm{i|fi|r,si|f|0,0+ijQr,s*(fi,fj)li,j|f|0,0}

and

(A.5)i=1m|fi|ri+ai,si+biCr+s+a+bm{i|fi|ri+a,si+bi|f|r,s
+ij|fi|ri+a,si|fj|rj,sj+bi,j|f|r,s}.

Here a=ai, b=bi, etc. Assume further that risi and ai=bi=0 for all i. Then

(A.6)f1r1,s1fmrm,smCr,smijQ^r,s(fi,fj)i,j|f|0,0.

Proof.

Let ki,ji. Let xikiλijifi denote a partial derivative of fλ(x) of order ki in x and order ji in λ, evaluated at x=xi and λ=λi. Let k=ki and j=ji.

We will prove the inequality by estimating derivatives pointwise and Hölder ratios of derivatives. We apply (A.2) for domains Di to obtain

(A.7)|xikiλijifi||λijifi|ki|λijifi|ki|λjf|0.

Estimating each term via pointwise derivatives, we can write

(A.8)|λijifi|ki|λjf|0=|xikλijifi|i|λjf(x)|

for some partial derivative k with order kk. While applying (A.2) for the domain I, we see that the last term is bounded by the right-hand side of (A.4). Thus, we have verified (A.4) when ri,sj are integers.

Let βi={ri} and ki=[ri], or βi=0 and ki[ri]. We estimate the product of Hölder ratios

βi>0|yikiλijifi-xikiλijifi||yi-xi|βi×βi=0|xikiλijifi||λijifi|ri|λjf|0.

We estimate each term in the last sum by considering pointwise derivatives and Hölder ratios. The former can be estimated by computations similar to (A.7)–(A.8). For the Hölder ratio, applying (A.2) for I, we can bound

|yikλijifi-xikλijifi||yi-xi|β×i|λjf|0

by the right-hand side of (A.4).

To verify (A.5), it suffices to consider the case with (s1,,sm,b1,,bm)2m. Furthermore, it is easy to reduce it to the case with si=0, which we now assume. Now let kiri+ai and jibi. We apply (A.2) to the domains Di to obtain

(A.9)|xikiλijifi||kr|λjf|ki|ki>ri|λijifi|ki+a|i,k>r|λjf|a

with a=|k>a(k-r)a. Again we estimate each term via pointwise derivative. We can write

{|kr|λjf|k}|λijifi|[ki+a]|i,k>a|λjf|[a]=|xikλijifi|i|xkλjf|.

While applying (A.2) for the domain I, we see that the last term is bounded by the right-hand side of (A.5). For Hölder ratios on the right-hand side of (A.9), let βi={ai+r} and ki=[ai+r], or βi=0 and ki[ai+r]; and for i, let β={a} or β=0 and let k[a]. We estimate the product of Hölder ratios

i|βi>0|yikiλijifi-xikiλijifi||yi-xi|βi×i|βi=0|xikiλijifi||λijifi|ri|λjf|0.

We need to estimate each term in the last sum by considering pointwise derivatives and Hölder ratios. Applying (A.2) to Di, we can bound |λijifi|[r]i|λjf|0 by the right-hand side of (A.5). Let β={r} and k=[r]. Applying now (A.2) to the interval I, we can bound

|yikλijifi-xikλijifi||yi-xi|β×i|λjf|0

by the right-hand side of (A.5). We have verified (A.5) and hence (A.6). ∎

By the product rule and Lemma A.1, we obtain the following.

Lemma A.2.

Let D be a domain in Rn with the cone property, and let m,m be such that 1m<m<. With all norms on D, we have

|{f1λfmλ}|r,s+Qr,s*({imfiλ},{i>mfiλ})
Cr,sm{i|fi|r,ski|fk|0,0+ijQr,s*(fi,fj)ki,j|fk|D;0,0}.

Let us use Lemmas A.1 and A.2 to verify the following.

Proposition A.3.

Let r,s[0,). Let ϕi be Cr+s+1 functions in [ai,bi]. Let D be as in Lemma A.2. Let fiC*r,s(D¯) and fiλ(D)[ai,bi]. Suppose that 0[ai,bi] and ϕi(0)=0. Let 1m<m. With all norms on D, we have

(A.10)|ϕ1(f1)|r,sCr,s(|f1|r,s+|f1|r,0|f1|0,s)

and

(A.11)|ϕ1(f1)ϕm(fm)|r,s+Qr,s*(imϕi(fi),i>mϕi(fi))
Cr,sm{i|fi|r,ski|fk|0,0+ijQr,s*(fi,fj)ki,j|fk|0,0},

where Cr,s also depends on |ϕi|[ai,bi];[r]+[s]+1.

Proof.

Since ϕi𝒞1 and ϕi(0)=0, we get |ϕi(fi)|D;α,β|fi|α,β for α,β[0,1]. Applying the chain rule and Lemma A.1, we get

|ϕi(fi)|r,sr1++rm=r,s1++sm=b|fi|r,s|fi|r,s+|fi|r,0|fi|0,s

with all ri,sj are integers, except for possible one. We can verify (A.11) similarly by using Lemma A.1. ∎

When applying the chain rule, we need to count derivatives efficiently.

Definition A.4.

Let k1 and kj0. Let Fλ be a family of mappings from D to D. Let {uλ} be a family of functions on D and let {fiλ},,{fmλ} be families of functions on D. Define 𝒫k,j({uλFλ};{fλ}) to be the linear space spanned by functions of the form

(A.12){K0λj0uλ}Fλ,{K0λj0uλ}Fλ1lKλjfnλ,1l<,

with

(A.13)|K|+j1,0,0jj,

and

(A.14)|K0|+j0+1(|K|+j-1)k,i.e. 0(|K|+j-1)<k.

Analogously, define 𝒫k,j({fλ}) to be the linear space spanned by

(A.15)1lKλjfnλ,1l<,

with |K|+j1, jj, (|K|+j-1)<k.

By counting efficiently, we count one less for the order of derivative Kλjfnλ. Also the l in (A.12) and (A.15) will have an upper bound depending on k,j.

It is easy to see that if {aλ}𝒫k0,j0({fλ}) and {vλ}𝒫k1,j1({uλFλ};{fλ}), with k01, k11, then

{aλvλ}𝒫k0+k1-1,j0+j1({uλFλ};{fλ})

by (A.14). We express it as

(A.16)𝒫k0,j0({fλ})×𝒫k1,j1({uλFλ};{fλ})𝒫k0+k1-1,j0+j1({uλFλ};{fλ})

for k0,k11. Also, if Fλ=I+fλ, then

𝒫k,j({uλFλ};{Fλ})=𝒫k,j({uλFλ};{fλ}).

Lemma A.5.

If 1|K|+jk, then

Kλj{uλFλ}𝒫k,j({uλFλ};{Fλ}).

Proof.

Let y=Fλ(x). Let i be a derivative of order i in x. We use the chain rule. Applying or λ one by one, we can verify that k-jλj(uλGλ) is a linear combination of functions

(A.17)(λj0K0uλ)Gλ1im,ji>0λjiKiGniλm<i|K0|,|Ki|>0KiGniλ.

Here 0j=j, and j+1|K|=k. When |K0|=0, it is clear that (A.13)–(A.14) hold. When |K0|>0, we have

0(j+|K|-1)=k-|K0|<k.

Thus (A.17) is in 𝒫k,j({uλFλ};{Fλ}).∎

Define

Q~r,s(f,g):=Qr-1,s(f,g),
Q~r,s(f,~g):=Q~r,s(f,g)+Qr-1,s-1(f,λg):=Q~r,s(~g,f),
Q~r,s(~f,~g):=Q~r,s(f,~g)+Qr-1,s-1(λf,g)+Qr-1,s-2(λf,λg),

where Qr-1,s=0 for r<s+1, Qr-1,s-1=0 for s<1, and Qr-1,s-2=0 for s<2.

Lemma A.6.

Let r,s satisfy (2.3). Let D,D be bounded domains in Rm,Rn respectively, which have the cone property. Let Gλ:DD be of class Cr,s(D¯).

  1. Assume that |G|D;1,0<2. Then

    (A.18)|{uλGλ}|D;α,0C|u|D;α,0,0α<1,
    (A.19)|{uλGλ}|D;r,0Cr(|u|D;r,0+|u|D;0,0|G|D;r-1,0),r1.

  2. Assume that |G|D;1,02 and s1. Then

    {uλGλ}D;r,sCr{uD;r,s+Q~D,D;r,s(u,~G)
    +|u|D;1,0(|G|D;r,s+Q~D,D;r,s(~G,~G))}.

Proof.

(i) Inequality (A.18) is immediate and (A.19) is in [18] and [10].

(ii) Let α={r} and β={s}. Let r=k+α with k1. Let y=Gλ(x). Let iuλ(y) be a partial derivatives in y of order i. We know that

k-jλj(uλGλ)𝒫k,j({uλGλ};{Gλ}),

i.e. it is a linear combination of functions vλ(x) of the form (A.17). We can express vλ as

vλ=(λj0muλ)Gλ1imλjikiλGniλm<imkiGniλ,

where j=j0++jm+m, and by (A.14) we have j+m+kik.

When m=0, it is immediate that by |G|1<2, we obtain

|v|α,β|u|α,j+β.

Suppose that m1. Computing the Hölder norms, we get

|v|α,β|u|m-1+α,j0+β|λGni|ki,ji|Gni|ki,0
+|u|j0+m-1,j0|λGn|m+α,j+βi,i,m|λGni|ki,jii>m|Gni|ki,0
+|u|j0+m-1,j0|Gn|m+α,j+βim|λGni|ki,jii,i,>m|Gni|ki,0.

We use Lemma A.1 and obtain

|v|α,β|u|r-1,s+|u|1,0|G|r,s+Qr-1,s(u,G)+Qr-1,s-1(u,λG)
+|u|1,0{Qr-1,s(G,G)+Qr-1,s-1(G,λG)+Qr-1,s-2(λG,λG)},

as desired. ∎

Let Dρ=Bρ2n×BρM×BρL with N=2n+K+L, and set ||ρ;r,s=||Dρ;r,s and

Q~ρ,ρ~;r,s=Q~Dρ,Dρ~;r,s.

Lemma A.7.

Let 1r<, let 0<ρ<, 0<θ<12 and ρi=(1-θ)iρ, and let Fλ=I+fλ be mappings from Dρ into RN. Assume that FC1,0(Dρ¯) and

fλ(0)=0,|f|ρ;1,0θCN.

Then the following statements hold:

  1. The mappings Fλ are injective in Dρ. There exist unique Gλ=I+gλ satisfying

    Gλ:Dρ1Dρ,FλGλ=Ion Dρ1.

    Furthermore, Fλ:DρD(1-θ)-1ρ and GλFλ=I in Dρ2.

  2. Suppose that 14<ρ<2. Assume further that F𝒞r,0(Dρ). Then {gλ}𝒞r,0(Dρ1) and

    (A.20)|g|ρ1;r,0Cr|f|ρ;r,0,|uG|ρ1;r,0Cr(|u|ρ;r,0+|u|ρ;1,0|f|ρ;r,0).

  3. Suppose that 14<ρ<2. Assume further that r,s satisfy (2.3) with s1, |f|ρ;1<C, and f𝒞r,s(Dρ). Then for u𝒞r,s(Dρ),

    (A.21)gρ1;r,sCr(fρ;r,s+Q~ρ,ρ;r,s(f,~f)+|f|ρ;1,0Q~ρ,ρ;r,s(~f,~f))

    and

    (A.22)uλGλρ1;r,sCr(uρ;r,s+Q~ρ,ρ1;r,s(u,~f)+|u|ρ;1,0Q~ρ,ρ;r,s(~f,~f)).

Proof.

We may reduce the proof to the case ρ=1 by using dilations xρ-1Fλ(ρx) and xρ-1Gλ(ρx). The proof of (i) can be obtained easily by applying the contraction mapping theorem to

gλ(x)=-fλ(x+gλ(x)).

Statement (ii) is a special case of (iii). Thus, let us verify (iii). Assume that r>1. Differentiating the above identity, we separate terms of the highest order derivatives of gλ from the rest to get the identities

(Kλj)glλ+m(ymflλ)Gλ(Kλj)gmλ=ElKjλ.

Here ElKjλ are the linear combinations of the functions in 𝒫|K|+j,j({fλGλ},{gλ}) of the form

(A.23)(K0λj0)flλ,vλ:=(M0λj0flλ)Gλ1mMλjgnλ

with m|M0|, |M|+j1, and jj. Furthermore,

(A.24)|M|+j<|K|+j,>0,0(|M|+j-1)<|K|+j.

We now verify that Kλjgλ is a finite sum of

(A.25)hKjλ:={A(f)K1λj1fn1Kmλjmfnm}Gλ,

where A(1f) is a polynomial in (det(fλ))-1 and fλ. Moreover, ji1 or |Ki|+ji2, and

(A.26)0(|K|+j-1)<|K|+j,0jj.

The assertion is trivial when |K|+j=1. Assume that it holds for |K|+j<N. By (A.16), we can see that the {vλ} in (A.23) has the form (A.25). This shows that vλ is of the form (A.25) and (A.26). The claim has been verified.

The estimation of |{hKjλ}|{r},{s} is the same as in the proof of Lemma A.6. Indeed, when |M0|=0, we have

|{hKjλ}|{r},{s}fr,s

for |K|+j[r] and j[s]. When |M0|>0, we have

|{hKjλ}|{r},{s}Q~r,s(f,~f)+|f|1,0Q~r,s(~f,~f).

This verifies (A.21).

To verify (A.22), as in (A.23)–(A.24) we note that kλj(uλGλ(x)) is a linear combination of functions in 𝒫k+j,j({uλGλ};{Gλ}) of the form

(A.27)v~λ=(M0λj0uλ)Gλ1mMλjgnλ

with m|M0|, |M|+j1, jj, and

(A.28)|M|+j<k+j,>0,0(|M|+j-1)k+j-1.

Expressing Kλjgλ via linear combinations in (A.25) satisfying (A.26) and applying them to Mλjgλ in (A.27), we conclude that v~λ and hence Kλjgλ is a linear combination of

h~Kjλ:={A(f)M0λj0uλ1|M0|in0Mλjfni}Gλ,

where M,j still satisfy (A.28). When |M0|=0, we actually have h~Kjλ=(λjuλ)Gλ. It is straightforward that |h~Kj|α,β|u|r,s. When |M0|>0, the estimate for |h~Kj|α,β is obtained analogous to that of hKj. ∎

We will also need to deal with sequences of compositions.

Proposition A.8.

Let Dm be a sequence of domains in Rd satisfying the cone property of which the constants C*(Dm)>1 are bounded. Assume that DmD and D also has the cone property. Let Fmλ=I+fmλ:DmDm+1 for m=1,2,. Suppose that

fiλ(0)=0,|fi|Di;1,01.

Assume that rs1 and {r}{s}. Then

{uλFλF1λ}D1;r,sCr{uD;r,s+Q~r,s(u,~fm)
+|u|D;1,0(fmDm;r,s+ijQ~r,s(~fi,~fj))}.

Proof.

We first derive the factor Cr in the inequality. Let Giλ=FiλF1λ. Let h,kλ be a k-th order derivative of uλGλ. We express h,kλ as a sum of terms of the form

(A.29)h2λ:=h~2λ{(λj0k0uλ)Gλ}1iTλjikiλFniλT<iTkiFniλ.

Here ki+jik-1, T+jij, and h~2λ is 1 or a product of first-order derivatives of Fiλ in x, which may be repeated. Let T,k be the maximum number of derivative functions of uλ,Fλ,,F1λ that appear in (A.29). We have T,0=0 and T,1=+1. By the chain rule, T,kT,k-1+k+1. Let N,k be the maximum number of terms (A.29) that are needed to express h,kλ as a sum of the monomials (A.29). Note that uλ(x),Fiλ(x) are functions in d+1 variables. By the chain rule, N,1(d+1). By the product and chain rules,

N,kN,k-1T,k-1(d+1)(k+1)(d+1)N,k-1<Ck.

For a fixed i, the first-order derivatives of Fiλ in x cannot repeated more than k times in h2λ. Thus, we have

|h~2λ(x)|(1+|f1|1,0)k(1+|f|1,0)k<2k.

Also the T in (A.29) is less than k. By Lemma A.2, we have

|h2|α,βNk+12k{ur,s+Qr,s(u,fi)+|u|1,0(|fi|r,s+Qr,s(fi,fj))}.

Here Nk+1 arises from the number of terms when computing the Hölder norms after taking k derivatives. ∎

Next, we consolidate the expressions such as Q~r,s(f,~g) in Lemma A.7 and Proposition A.8 by a simpler expression Qr,s(f,g), defined by (A.3), and the new expression

Qr,s(f,g):=Qr,s(f,g)+j=1[s]Qr-j+1,j+{s}*(f,g),s1.

Lemma A.9.

Let fCr,s(D),gCr,s(D) with D,D having the cone property. Then

(A.30)Qr-1,s(f,g)CrQ^r,s(f,g),Qr-1,s-1(λf,g)CrQ^r,s(f,g),
(A.31)Qr-1,s(f,g)Cr{|f|1,{s}gr,0+|f|1,0gr,s
+|g|1,{s}|f|r,0+|g|1,0|f|r,s+Qr,s(f,g)},
(A.32)Qr-1,s-1(f,λg)Cr{|f|1,{s}|g|r-1,1+Qr,s(f,g)}, 1s<2r,
(A.33)Qr-1,s-1(f,λg)Cr{|f|1,{s}|g|r-1,1+|f|0,1gr,s-1
+|g|0,1fr,s-1+Qr,s(f,g)},s1,
(A.34)Qr-1,s-2(λf,λg)Cr{|f|0,1gr,s-1+|g|0,1fr,s-1
+|f|{r},s|g|0,0+|g|{r},s|f|0,0+Qr,s(f,g)}.

Here the norms of f (resp. g) and its derivatives are in D (resp. D).

Proof.

We verify the first inequality in (A.30) by

|f|r-1-j,0|g|0,j+{s}Qr,s(f,g)

and

|f|0,j+{s}|g|r-1-j,0|f|0,j+{s}|g|r-j,0+|f|r-j,j+{s}|g|0,0
Q^r,s(f,g).

Here the first inequality follows from (A.5). The second inequality in (A.30) follows from

|g|r-1-j,0|λf|0,j+{s}Qr,s(f,g)

and

|g|0,j+{s}|λf|r-1-j,0|g|0,j+1+{s}|f|r-1-j,0+|g|0,0|f|r-1-j,j+1+{s}Q^r,s(f,g).

We verify (A.31) by

|f|r-1,0|g|0,{s}|f|r,0|g|1,{s}

and for j>0,

|f|r-1-j,0|g|0,j+{s}|f|r-j,0|g|0,j+{s}+|f|0,0|g|r-j,j+{s}.

Inequality (A.32) follows from

|f|r-j-1,0|tg|0,j+{s}Qr,s(f,g)

and

|f|0,{s}|tg|r-1,0|f|1,{s}|g|r-1,1.

For (A.33), we need to verify extra terms. Suppose 0<j<[s]. Then

|f|0,j+{s}|tg|r-j-1,0|f|0,j+{s}|tg|r-j,0+|f|r-j,j+{s}|g|0,1

and

|f|0,j+{s}|tg|r-j,0|f|0,j+1+{s}|g|r-j,0+|f|0,1|g|r-j,j+{s}.

We verify (A.34) by

|tf|{r},0|tg|0,j+{s}|f|{r},0|g|0,s+|f|{r},s|g|0,0

and

|tf|r-j-2,0|tg|0,j+{s}|f|r-j-2,0|g|0,j+2+{s}+fr,s-1|g|0,1.

Thus, the proof is complete. ∎

We now state a simple version of the above inequalities, which suffices applications in this paper. The next two propositions are immediate consequences of Lemma A.7, Proposition A.8, Lemma A.9, and the following crude estimate:

Q~r,s(~f,~g)C(|f|1,s*gr,s+|g|1,s*fr,s
+fs+1,sgr,s*+gs+1,sfr,s*).

Proposition A.10.

Let s*=0 for s=0 and s*=1 for s1. Let Fλ=I+fλ and (Fλ)-1=I+gλ be as in Lemma A.7. Assume that

fλ(0)=0,|f|1,0θCN,|f|1,s*<1.

Let ρ1=(1-θ)ρ with 14<ρ<2 and 0<θ<12. Then

gρ1;r,sCr{fρ;r,s+fs+1,sfr,s*}

and

{uλ(Fλ)-1}ρ1;r,sCr{uρ;r,s+|u|ρ;1,s*fρ;r,s+us+1,sfr,s*
+ur,s*fs+1,s+|u|ρ;1,0fs+1,sfr,s*}.

Proposition A.11.

Let r,s,s* be as in Proposition A.10. Let Fiλ=I+fiλ map Di into Di+1, where D are as in Proposition A.8. Assume that fiλ(0)=0 and |fi|1,s*1. Then

{uλFmλF1λ}D1;r,sCrm{ur,s+u1,0ijfis+1,sfjr,s*
+i|u|1,s*fir,s+us+1,sfir,s*
+ur,s*fis+1,s},

where fia,b=fiDi;a,b and ua,b=uDm+1;a,b.

References

[1] L. Bers, Riemann surfaces (mimeographed lecture notes), New York University, New York 1957–1958. Search in Google Scholar

[2] F. Bertrand and X. Gong, Dirichlet and Neumann problems for planar domains with parameter, Trans. Amer. Math. Soc. 366 (2014), no. 1, 159–217. 10.1090/S0002-9947-2013-05951-XSearch in Google Scholar

[3] F. Bertrand, X. Gong and J.-P. Rosay, Common boundary values of holomorphic functions for two-sided complex structures, Michigan Math. J. 63 (2014), no. 2, 293–332. 10.1307/mmj/1401973052Search in Google Scholar

[4] S.-C. Chen and M.-C. Shaw, Partial differential equations in several complex variables, AMS/IP Stud. Adv. Math. 19, American Mathematical Society, Providence 2001. 10.1090/amsip/019Search in Google Scholar

[5] S.-S. Chern, An elementary proof of the existence of isothermal parameters on a surface, Proc. Amer. Math. Soc. 6 (1955), 771–782. 10.1090/S0002-9939-1955-0074856-1Search in Google Scholar

[6] G. Frobenius, Ueber das Pfaffsche Problem, J. reine angew. Math. 82 (1877), 230–315. 10.1515/crll.1877.82.230Search in Google Scholar

[7] D. Gilbarg and N. S. Trudinger, Elliptic partial differential equations of second order, Classics Math., Springer, Berlin 2001. 10.1007/978-3-642-61798-0Search in Google Scholar

[8] X. Gong and K. Kim, The ¯-equation on variable strictly pseudoconvex domains, Math. Z., to appear. 10.1007/s00209-017-2011-zSearch in Google Scholar

[9] X. Gong and S. M. Webster, Regularity for the CR vector bundle problem I, Pure Appl. Math. Q. 6 (2010), no. 4, 983–998. 10.4310/PAMQ.2010.v6.n4.a1Search in Google Scholar

[10] X. Gong and S. M. Webster, Regularity for the CR vector bundle problem II, Ann. Sc. Norm. Super. Pisa Cl. Sci. (5) 10 (2011), no. 1, 129–191. 10.2422/2036-2145.2011.1.05Search in Google Scholar

[11] X. Gong and S. M. Webster, Regularity in the local CR embedding problem, J. Geom. Anal. 22 (2012), no. 1, 261–293. 10.1007/s12220-010-9192-6Search in Google Scholar

[12] H. Guggenheimer, A simple proof of Frobenius’s integration theorem, Proc. Amer. Math. Soc. 13 (1962), 24–28. 10.2307/2033765Search in Google Scholar

[13] N. Hanges and H. Jacobowitz, The Euclidean elliptic complex, Indiana Univ. Math. J. 46 (1997), no. 3, 753–770. 10.1512/iumj.1997.46.1397Search in Google Scholar

[14] T. Hawkins, The mathematics of Frobenius in context. A journey through 18th to 20th century mathematics, Sources Stud. Hist. Math. Phys. Sci., Springer, New York 2013. 10.1007/978-1-4614-6333-7_18Search in Google Scholar

[15] C. D. Hill and M. Taylor, Integrability of rough almost complex structures, J. Geom. Anal. 13 (2003), no. 1, 163–172. 10.1007/BF02931002Search in Google Scholar

[16] C. D. Hill and M. Taylor, The complex Frobenius theorem for rough involutive structures, Trans. Amer. Math. Soc. 359 (2007), no. 1, 293–322. 10.1090/S0002-9947-06-04067-0Search in Google Scholar

[17] L. Hörmander, The Frobenius–Nirenberg theorem, Ark. Mat. 5 (1965), 425–432. 10.1007/BF02591139Search in Google Scholar

[18] L. Hörmander, The boundary problems of physical geodesy, Arch. Ration. Mech. Anal. 62 (1976), no. 1, 1–52. 10.1007/BF00251855Search in Google Scholar

[19] K. Kodaira and D. C. Spencer, Multifoliate structures, Ann. of Math. (2) 74 (1961), 52–100. 10.2307/1970306Search in Google Scholar

[20] J. J. Kohn, Harmonic integrals on strongly pseudo-convex manifolds. I, Ann. of Math. (2) 78 (1963), 112–148. 10.2307/1970506Search in Google Scholar

[21] L. Ma and J. Michel, Regularity of local embeddings of strictly pseudoconvex CR structures, J. reine angew. Math. 447 (1994), 147–164. 10.1515/crll.1994.447.147Search in Google Scholar

[22] B. Malgrange, Sur l’intégrabilité des structures presque-complexes, Symposia mathematica. Vol. II (Rome 1968), Academic Press, London (1969), 289–296. Search in Google Scholar

[23] J. Moser, On invariant curves of area-preserving mappings of an annulus, Nachr. Akad. Wiss. Göttingen Math.-Phys. Kl. II 1962 (1962), 1–20. Search in Google Scholar

[24] R. Narasimhan, Analysis on real and complex manifolds, North-Holland Math. Libr. 35, North-Holland, Amsterdam 1985. Search in Google Scholar

[25] F. Nevanlinna and R. Nevanlinna, Absolute analysis, Springer, New York 1973. 10.1007/978-3-662-00249-0Search in Google Scholar

[26] A. Newlander and L. Nirenberg, Complex analytic coordinates in almost complex manifolds, Ann. of Math. (2) 65 (1957), 391–404. 10.2307/1970051Search in Google Scholar

[27] A. Nijenhuis and W. B. Woolf, Some integration problems in almost-complex and complex manifolds, Ann. of Math. (2) 77 (1963), 424–489. 10.2307/1970126Search in Google Scholar

[28] L. Nirenberg, A complex Frobenius theorem, Seminars on analytic functions I, Institute for Advanced Study, Princeton (1957), 172–189. Search in Google Scholar

[29] R. T. Seeley, Extension of C functions defined in a half space, Proc. Amer. Math. Soc. 15 (1964), 625–626. 10.1090/S0002-9939-1964-0165392-8Search in Google Scholar

[30] F. Treves, Hypo-analytic structures, Princeton Math. Ser. 40, Princeton University Press, Princeton 1992. Search in Google Scholar

[31] S. M. Webster, A new proof of the Newlander–Nirenberg theorem, Math. Z. 201 (1989), no. 3, 303–316. 10.1007/BF01214897Search in Google Scholar

[32] S. M. Webster, On the proof of Kuranishi’s embedding theorem, Ann. Inst. H. Poincaré Anal. Non Linéaire 6 (1989), no. 3, 183–207. 10.1016/s0294-1449(16)30322-5Search in Google Scholar

[33] S. M. Webster, The integrability problem for CR vector bundles, Several complex variables and complex geometry. Part 3 (Santa Cruz 1989), Proc. Sympos. Pure Math. 52, American Mathematical Society, Providence (1991), 355–368. 10.1090/pspum/052.3/1128608Search in Google Scholar

Received: 2016-11-20
Revised: 2017-11-23
Published Online: 2018-01-17
Published in Print: 2020-02-01

© 2022 Walter de Gruyter GmbH, Berlin/Boston

Downloaded on 12.3.2026 from https://www.degruyterbrill.com/document/doi/10.1515/crelle-2017-0051/html
Scroll to top button