標籤:

Hormander分析不等式

Omega subset C^n為擬凸域

Omega=cup_{j=1}^{infty}K_j,hat K_j=K_j subset K_{j+1}.

eta_j in C_0^{infty}(Omega),eta_j |_{K_{j-1}}=1,supp eta_j subset K_j.

psi in C_0^{infty}(Omega)

|e^{psi}|^2=1+sum_{j=1}^{infty}sum_{k=1}^n |frac{partial eta_j}{partial ar z_k}|^2. |partial eta_j|^2 leq |e^{psi}|^2.

phiin C_0^{infty}(Omega) cap PSH(Omega).

phi_1=psi-2phi,phi_2=psi-phi,phi_3=psi.

sum_{j=1}^n frac{partial^2 psi}{partial z_j partial ar z_k} xi_jar xi_k geq 2(|partial psi|^2+e^{psi})|xi|^2.forall zin Omega,forall xi in C^n.

L^2_{(p,q)}(Omega,phi_1) longrightarrow L^2_{(p,q+1}(Omega,phi_2) longrightarrow L_{(p,q+2)}^2(Omega,phi_3).

H_1 longrightarrow H_2 longrightarrow H_3

H_i=L^2_{(p,q+1-i)}(Omega,phi_i).

||cdot ||_{H_j}=||cdot||_{phi_j},phi_j longrightarrow e^{-phi_j}d lambda. dlambda lesbegue測度.

forall fin D_{(p,q+1)}(Omega),D=C_0^{infty}.

T^*: H_2 longrightarrow H_1.

<f,T^*g>_{phi_1}=<Tf,g>_{phi_2},f in H_1,gin H_2.

f(z)=sum_{|I|=p,|J|=q}f_{I,J} dz_{I}wedge dz_J.

g(z)=sum_{I,J}sum_{j=1}^n g_{I,j,J}dz_I wedge dar z_j wedge d ar z_J.

Tf(z)=arpartial f(z)=sum_{|I|=p,|J|=q}sum_{j=1}^n(-1)^pfrac{partial f_{I,J}}{partial ar z_j}dz_I wedge d ar z_j wedge ar z_J.

<Tf,g>_{phi_2}=int_{Omega}sum_{j=1}^n(-1)^p frac{f_{I,J}}{partial ar z_j}ar g_{I,j,J}e^{-phi_2}dlambda

sum_{j=1}^n(-1)^pint_{Omega}frac{f_{I,J}}{partial ar z_j}(ar g_{I,j,J}e^{-phi_2})dlambda

(integral by part)=sum_{j=1}^n(-1)^{p-1}int_{Omega}f_{I,J}frac{partial}{partial ar z_j}(ar g_{I,j,J}e^{-phi_2})dlambda.

另一方面

<f,T^*g>_{phi_1}=sum_{|I|=p,|J|=q}int_{Omega}f_{I,J}overline{(T^*g)_{I,J}}e^{-phi_1}dlambda

f(z)=sum_{|I|=p,|J|=q}f_{I,J}dz_Iwedge dar z_J.

T^*g(z)=sum_{|I|=p,|J|=q}(T^*g)_{I,J}dz_I wedge dar z_J.

又由於<f,T^*g>_{phi_1}=<Tf,g>_{phi_2},從此式比較係數得:

T^*g(z)=(-1)^{p-1}sum_{|I|=p,|J|=q}sum_{j=1}^ne^{phi} frac{partial}{partial z_j}(e^{-phi_2} g_{I,j,J})dz_I wedge d ar z_J.

進一步化簡:

T^*g(z)=(-1)^{p-1}sum_{|I|=p,|J|=q}sum_{j=1}^ne^{phi} frac{partial}{partial z_j}(e^{-phi_2} g_{I,j,J})dz_I wedge d ar z_J=(-1)^{p-1}e^{-phi}sum_{|I|=p,|J|=q}sum_{j=1}^n(frac{partial h_{I,j,J}}{partial z_j}-frac{partial phi}{partial z_j}g_{I,j,J})dz_Iwedge d ar z_J+(-1)^{p-1}e^{-phi}sum_{|I|=p,|J|=q}sum_{j=1}^nfrac{partial phi}{partial z_j}g_{I,j,J}dz_Iwedge d ar z_J

=X+Y.

因此:T^*f=X+Y,X=T^*f-Y

||X||_{phi_1}leq ||T^*f||_{phi_1}+||Y||_{phi_1}.

||X||_{phi_1}^2 leq 2(||T^*f||_{phi_1}^2+||Y||_{phi_1}^2),f=g.

Y=(-1)^{p-1}e^{-phi}sum_{|I|=p,|J|=q}sum_{j=1}^nfrac{partial phi}{partial z_j}g_{I,j,J}dz_Iwedge d ar z_J.

||Y||_{phi_1}^2=int_{Omega_1}|(-1)^{p-1}e^{-phi}sum_{j}frac{partial psi}{partial z_j} g_{I,j,J}|^2e^{-phi_1}dlambda

=int_{Omega}|sum_j frac{partial psi}{partial z_j }g_{I,jJ}|^2e^{-phi}dlambdaleq int_{Omega} |f|^2|arpartial phi|^2e^{-phi}dlambda=||f||_{phi_1}^2

X =(-1)^{p-1}e^{-phi}sum_{|I|=p,|J|=q}sum_{j=1}^n(frac{partial h_{I,j,J}}{partial z_j}-frac{partial phi}{partial z_j}g_{I,j,J})dz_Iwedge d ar z_J.

delta_j g=e^{phi}frac{partial}{partial z_j}e^{-phi}g=frac{partial g}{partial z_j}-frac{partial phi}{partial z_j}g.

g,h in C_0^{infty}(Omega).

<g,frac{partial}{partial ar z_j}>_{phi}=-<delta_j g,h>_{phi}.

int_{Omega}g overline{(frac{partial h}{partial ar z_j})}e^{-phi}dlambda=int_{Omega}overline{(frac{partial h}{partial ar z_j})}(ge^{-phi})dlambda

=int_{Omega}ar h e^{phi}frac{partial}{partial z_j}(g e^{-phi})dlambda

-<delta_j g,h>_{phi}.

[delta_j,ar partial_k]=frac{partial^2 phi}{partial z_jpartial ar z_k}, [A,B]=AB-BA.

[delta_j,delta_k]g=[frac{partial}{partial z_j}-frac{partialphi}{partial z_j},frac{partial}{partial ar z_k}]g

=[-frac{partial phi}{partial z_j},frac{partial}{partial ar z_k}]g

=frac{partial^2 phi}{partial z_jpartial ar z_k}g.

<delta_j g,delta_k h>_{phi}=<frac{partial g}{partial ar z_k},frac{partial h}{partial ar z_j}>_{phi}+<g,frac{partial^2 phi}{partial z_jpartial ar z_k}h>_{phi}

實際上

<g,ardelta_j h>_{phi}

=<g,ar delta_jdelta_k h>_{phi}=<g,delta_k(ardelta_j h)+frac{partial^2 phi}{partial z_jpartial ar z_k}h>_{phi}=<arpartial_k g,ar partial_j h>_{phi}+<g,frac{partial^2 phi}{partial z_jpartial ar z_k}h>_{phi}.

下面計算||X||^2_{phi}

X =(-1)^{p-1}e^{-phi}sum_{|I|=p,|J|=q}sum_{j=1}^n(frac{partial h_{I,j,J}}{partial z_j}-frac{partial phi}{partial z_j}g_{I,j,J})dz_Iwedge d ar z_J.

phi_1=phi-2psi

||X||_{phi_1}^2=int_{Omega}sum_{I,J}|sum_j delta_j f_{I,j,J}|^2e^{-2psi} e^{-phi_1}dlambda

sum_{I,J}sum_{j,k}int_{Omega}<delta_jf_{I,j,J},delta_kf_{I,k,J}>e^{-psi}dlambda

結合前面的計算告訴我們:

||X||_{phi_1}^2leq ||T^*f||^2+int_{Omega}|f|^2|arpartial phi|^2e^{-psi}dlambda.

||X||^2_{phi_1}=sum_{I,J}sum_{j,k}<delta_if_{I,i,J},delta_kf_{I,k,J}>_{phi}

=sum_{I,J}sum_{j,k}<frac{partial}{partial ar z_k}f_{I,j,J},frac{partial}{partial ar z_j}f_{I,k,J}>_{phi}+sum_{I,J}sum_{j,k}<f_{I,j,J},frac{partial^2 phi}{partial z_j partial ar z_k}f_{I,k,J}>_{phi}

涉及到之後的計算,進一步,我們將會得到:

||X||^2_{phi_1}=sum_{I,J}sum_{j,k}<delta_if_{I,i,J},delta_kf_{I,k,J}>_{phi}

=sum_{I,J}sum_{j,k}<frac{partial}{partial ar z_k}f_{I,j,J},frac{partial}{partial ar z_j}f_{I,k,J}>_{phi}+sum_{I,J}sum_{j,k}<f_{I,j,J},frac{partial^2 phi}{partial z_j partial ar z_k}f_{I,k,J}>_{phi}=sum_{I,J}sum_jint_{Omega}|frac{partial f_{I,J}}{partial ar z_j}|^2e^{-phi}dlambda-||Sf||_{phi}^2+sum_{I,J}sum_{j,k}<f_{I,j,J},frac{partial^2 phi}{partial ar z_j partialar z_k f_{I,k,J}}>_{phi}

關鍵是計算Sf

我們有線性映射:

L_{(p,q+1)}^2(Omega,phi_2) longrightarrow L^2_{(p,q+2)}(Omega,phi_3)

Sf=delta f=(-1)^{p}sum_{|I|=p,|J|=q+1}sum_{j=1}^n frac{partial f_{I,J}(z)}{partial ar z_j} dz_Iwedge    dar z_j wedge dar z_J

=sum_{I,K}sum_{j}(-1)^pfrac{partial f_{I,J}(z)}{partial ar z_j}sigma_k^{j,J}dz_I wedge d ar z_k.

||Sf||_{phi}^2=sum_{I,K}int_{Omega}|sum_{J}sum_j(-1)^pfrac{partial f_{I,J(z)}}{partial ar z_j}sigma_{k}^{I,J}|^2e^{-phi}dlambda

=sum_{I,K}int_{Omega}sum_{J,L}sum_{j,l}(frac{partial f_{I,J}(z)}{partial ar z_j}sigma_k^{i,J})(frac{partial f_{I,L}}{partial ar z_l}sigma_{l,L}^k)e^{-phi}dlambda

=sum_{I,J,L}sum_{j,l}frac{partial f_{I,J}}{partial ar z_j}frac{partial f_{I,L}}{partial ar z_l}sum_k sigma_k^{j,J}sigma_{lL}^k e^{-phi}dlambda

(我們知道sum_k sigma_k^{j,J}sigma_{l,L}^k=sigma_{l,L}^{k,J})

=sum_{I,J,L}sum_{j,l}frac{partial f_{I,J}}{partial ar z_j}frac{partial f_{I,L}}{partial ar z_l}sum_k sigma_{l,L}^{k,J} e^{-phi}dlambda.

(sigma^{j,J}_{l,L}=sigma^J_Lj 
eq lJ=j cup K

sigma_L^J=1J=l

0 
eq sigma_{lL}^{jJ} = sigma_{jlk}^{jJ}sigma_{ljk}^{jlk}sigma_{lL}^{lek}

=sigma_{lk}^Jsigma_L^{jk} )

=sum_{I,J}sum_{j 
otin J}int_{Omega}|frac{partial f_{I,J}}{partial ar z_j}|^2e^{-phi}dlambda-sum_{I,K}sum_{j
eq L}int_{Omega}frac{partial f_{I,J,k}}{partial z_j}frac{partial f_{I,l,K}}{partial z_l}e^{phi}dlambda.

sum_{I,J}sum_{jin J}int_{Omega}|frac{partial f_{I,J}}{partial ar z_j}|^2e^{-phi}dlambda

(j=lk)=sum_{I,K}sum_{j=l}int_{Omega}frac{partial f_{I,j,k}}{partial ar z_j}overline{frac{partial f_{I,j,k}}{partial ar z_l}}e^{-phi}dlambda.

||delta f||^2_{phi}=sum_{I,J}sum_jint_{Omega}|frac{partial f_{I,J}}{partial z_j}|^2e^{-phi}dlambda-sum_{I,K}sum_{j,l}int_{Omega}frac{partial f_{I,j,k}}{partial z_j}overline{frac{partial f_{J,j,k}}{partial z_l}}e^{-phi}dlambda.

結合這些計算我們得到了:

||X||^2_{phi_1}=sum_{I,J}sum_{j,k}<delta_if_{I,i,J},delta_kf_{I,k,J}>_{phi}

=sum_{I,J}sum_{j,k}<frac{partial}{partial ar z_k}f_{I,j,J},frac{partial}{partial ar z_j}f_{I,k,J}>_{phi}+sum_{I,J}sum_{j,k}<f_{I,j,J},frac{partial^2 phi}{partial z_j partial ar z_k}f_{I,k,J}>_{phi}

=sum_{I,J}sum_jint_{Omega}|frac{partial f_{I,J}}{partial ar z_j}|^2e^{-phi}dlambda-||Sf||_{phi}^2+sum_{I,J}sum_{j,k}<f_{I,j,J},frac{partial^2 phi}{partial ar z_j partialar z_k f_{I,k,J}}>_{phi}

這樣我們證明了:

Hormander L^2 inequality:

sum_{|I|=p,|J|=q}sum_{j,k=1}^nint_{Omega}frac{partial^2 phi}{partial z_j partial ar z_k}f_{I,j,J}overline{f_{I,k,J}}e^{-phi}dlambda+sum_{|I|=p,|J|=q}sum_{j,k=1}^nint_{Omega}|frac{partial f_{I,J}}{partial ar z_j}|^2e^{-phi}dlambda

leq 2||T^*f||_{phi_1}^2+||delta f||_{phi_3}^2+2int_{Omega}|f|^2|delta phi|^2e^{-phi}dlambda.

此不等式加上levy-form的標準的不等式:

L(phi,X)geq 2(|arpartial phi|^2+e^{-phi}2||X||^2) 其中phi多重次調和函數

可得:

forall f in D_{p,q+1}(Omega,e^{phi})

||f||_{phi_2}^2leq ||T^*f||_{phi_1}^2+||delta||_{phi_3}^2.


推薦閱讀:

【解析幾何】雙聯立(齊次化處理)解決定點問題
Gauss與AGM(V-3)
二重積分
【不等式】9-8不等式及其應用

TAG:數學 |