summaryrefslogtreecommitdiff
path: root/macros/latex209/contrib/lms/lmssampl.tex
blob: f946b5d26ffd3f384d0da4d3ffde5de0e550f7ab (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
% lmssampl.tex
% Copyright (C) 1993,1994,1995 Cambridge University Press
% v0.3, 4th January 1995 

\documentstyle{lms}

%%% Uncomment ONE of the combinations \onetrue, \twotrue... \seventrue,
%%% depending on the LaTeX configuration on your system. The files in 
%%% parentheses are required to tex the sample pages and guide.

%%% You will need to have a local copy of the files shown in parentheses
%%% for those documentstyles. If you have any problems at all, remember
%%% that your best option is to choose the version which has `noams' as
%%% part of its title, and if you wish the typesetter to substitute ams
%%% fonts for your fonts, please indicate this in the text.

%%% oldfss  = LaTeX without NFSS
%%% nfssone = LaTeX with New Font Selection Scheme v1
%%% nfsstwo = LaTeX with New Font Selection Scheme v2

%%% noams   = No AMS fonts, sample pages will substitute some characters
%%% amsone  = AMS v1 fonts
%%% amstwo  = AMS v2 fonts (the recommended option)

%%% Uncomment ONE of the following combinations:

 \onetrue     % oldfss/noams
% \twotrue     % oldfss/amsone (mssymb.tex)
% \threetrue   % oldfss/amstwo (amssym.def,amssym.tex)
% \fourtrue    % nfssone/noams
% \fivetrue    % nfssone/amstwo (amsfonts.sty,amssymb.sty)
% \sixtrue     % nfsstwo/noams
% \seventrue   % nfsstwo/amstwo (amsfonts.sty,amssymb.sty)

%%%% DON'T MODIFY THE FOLLOWING CODE; SKIP TO `START HERE' %%%%

\ifCUPmtlplainloaded
  \ifoldfss
%   \usefraktur
    \usescript
    %
    % set up 9pt Bbb for bibliography
    \font\ninemsy=mtym10 at 9pt
    \def\bibliobbb{\textfont\msyfam=\ninemsy}
  \fi
%
\else
%
%%% one (oldfss/noams)
  \ifone
    \def\frak{\protect\cal}
    \let\goth=\frak
    \def\Bbb{\protect\bf}
    \def\scr{\protect\cal}
  \fi

%%% two (oldfss/amsone)
  \iftwo 
    \input mssymb\relax % %%%MR own catcodes
    \makeatletter
    %
    % if the following is already uncommented in your version of
    % mssymb.tex, you may comment this out to % FINISH HERE to save
    % a font family
    %
    % set up Euler Fraktur/Gothic font (\frak, \goth)
    \font\teneuf=eufm10
    \font\seveneuf=eufm7
    \font\fiveeuf=eufm5
    \newfam\euffam
    \textfont\euffam=\teneuf
    \scriptfont\euffam=\seveneuf
    \scriptscriptfont\euffam=\fiveeuf
    \def\frak{\ifmmode\let\next\frak@\else
      \def\next{\errmessage{Use \string\frak\space only in math mode}}\fi
        \next}
    \def\goth{\ifmmode\let\next\frak@\else
      \def\next{\errmessage{Use \string\goth\space only in math mode}}\fi
        \next}
    \def\frak@#1{{\frak@@{#1}}}
    \def\frak@@#1{\fam\euffam#1}
    % FINISH HERE
    %
    % set up 9pt Bbb for bibliography
    \font\ninemsb=msym9
    \def\bibliobbb{\textfont\msyfam=\ninemsb}
    %
    \def\scr{\protect\cal}
    \let\le\leqslant \let\ge\geqslant
    \let\leq\leqslant \let\geq\geqslant
    \makeatother
  \fi

%%% three (oldfss/amstwo)
  \ifthree
%   \input test/oldfss/amssym.def\relax
%   \input test/oldfss/amssym.tex\relax %%%MR own catcode
    \input amssym.def\relax
    \input amssym.tex\relax %%%MR own catcode
    %
    % set up Euler Script font (\scr)
    \font\scrten=eusm10   \skewchar\scrten='60
    \font\scrseven=eusm7  \skewchar\scrseven='60
    \font\scrfive=eusm5   \skewchar\scrfive='60
    \newfam\scrfam
    \textfont\scrfam=\scrten
    \scriptfont\scrfam=\scrseven
    \scriptscriptfont\scrfam=\scrfive
    \def\scr{\fam\scrfam\scrten}
    %
    % set up 9pt Bbb for bibliography
    \font\ninemsb=msbm9
    \def\bibliobbb{\textfont\msbfam=\ninemsb}
    %
    \let\le\leqslant \let\ge\geqslant
    \let\leq\leqslant \let\geq\geqslant
  \fi

%%% four (nfssone/noams)
  \iffour
    \def\frak{\protect\cal}
    \let\goth=\frak
    \def\Bbb{\protect\mathbf}
    \def\scr{\protect\cal}
  \fi

%%% five (nfssone/amstwo)
  \iffive
    \makeatletter
%   \input test/nfss1/amsfonts.sty\relax %%%MR assumes \makeatletter
%   \input test/nfss1/amssymb.sty\relax
    \input amsfonts.sty\relax %%%MR assumes \makeatletter
    \input amssymb.sty\relax
    %
    \new@fontshape{eus}{m}{n}{%
       <5>eusm5<6>eusm6<7>eusm7<8>eusm8<9>eusm9<10>eusm10%
       <11>eusm10 at 10.95pt<12>eusm10 at 12pt<14>eusm10 at 14.4pt%
       <17>eusm10 at 17.28pt<20>eusm10 at 20.736pt%
       <25>eusm10 at 24.8832pt}{}%
    %
    \new@fontshape{eus}{b}{n}{%
       <5>eusb5<6>eusb6<7>eusb7<8>eusb8<9>eusb9<10>eusb10%
       <11>eusb10 at 10.95pt<12>eusb10 at 12pt<14>eusb10 at 14.4pt%
       <17>eusb10 at 17.28pt<20>eusb10 at 20.736pt%
       <25>eusb10 at 24.8832pt}{}%
    %
    \extra@def{eus}{\skewchar#1='60}{}
    %
    \newmathalphabet{\scr}
    \addtoversion{normal}{\scr}{eus}{m}{n}
    \addtoversion{bold}{\scr}{eus}{b}{n}
    %
    \let\le\leqslant \let\ge\geqslant
    \let\leq\leqslant \let\geq\geqslant
    \makeatother
 \fi

%%% six (nfsstwo/noams)
  \ifsix
    \let\cal\mathcal
    \def\frak{\protect\mathcal}
    \let\goth=\frak
    \def\Bbb{\protect\mathbf}
    \def\scr{\protect\mathcal}   
  \fi 

%%% seven (nfsstwo/amstwo)
  \ifseven
    \makeatletter
%   \input test/nfss2/amsfonts.sty\relax %%%MR asuumes \makeatletter
%   \input test/nfss2/amssymb.sty\relax
    \input amsfonts.sty\relax %%%MR asuumes \makeatletter
    \input amssymb.sty\relax
    %
    \DeclareMathAlphabet{\scr}{U}{eus}{m}{n}
    \SetMathAlphabet{\scr}{bold}{U}{eus}{b}{n}
    %
    \let\cal\mathcal
    \let\le\leqslant \let\ge\geqslant
    \let\leq\leqslant \let\geq\geqslant
    \makeatother
  \fi
%
\fi % end of CUPmtlplainloaded


%%% for guide only

\ifoldfss    
  \newcommand{\mitbf}[1] {\mbox{\boldmath ${#1}$}}
  \newcommand{\rmn}[1] {{\rm {#1}}}
  \newcommand{\itl}[1] {{\it {#1}}}
  \newcommand{\bld}[1] {{\bf {#1}}}
\fi

\ifnfssone
  \newmathalphabet{\mathit}
  \addtoversion{normal}{\mathit}{cmr}{m}{it}
  \addtoversion{bold}{\mathit}{cmr}{bx}{it}
  \newmathalphabet{\mathcal}
  \addtoversion{normal}{\mathcal}{cmsy}{m}{n}
  \newcommand{\mitbf}[1] {\hbox{\mathversion{bold}${#1}$}}
  \newcommand{\rmn}[1] {{\mathrm {#1}}}
  \newcommand{\itl}[1] {{\mathit {#1}}}
  \newcommand{\bld}[1] {{\mathbf {#1}}}
\fi

\ifnfsstwo
  \newcommand{\mitbf}[1] {\hbox{\mathversion{bold}${#1}$}}
  \newcommand{\rmn}[1] {{\mathrm {#1}}}
  \newcommand{\itl}[1] {{\mathit {#1}}}
  \newcommand{\bld}[1] {{\mathbf {#1}}}
\fi

%%% START HERE %%%

\extraline{First author supported in part by a grant 
           from the National Science Foundation}

\newtheorem{theorem}{Theorem}
\newtheorem{lemma}{Lemma}
\newtheorem{corollary}{Corollary}
\newunnumbered{remark}{Remark}
\newunnumbered{remarks}{Remarks}
\newunnumbered{notation}{Notation}
\newunnumbered{definition}{Definition}

\classno{35B60}
\begin{document}

\title[Dirichlet's Problem When The Data Is An Entire Function]
      {Dirichlet's Problem When The Data\\ Is An Entire Function}

\author{Dmitry Khavinson \and\ Harold S. Shapiro}

\maketitle

\section{Introduction}
\label{sec-Introduction}

This paper may be regarded as a sequel (and correction) to \cite{Incorrect
version}, and we use similar notations. Thus $x=(x_1,\ldots,x_n)$ and
$y=(y_1,\dots,y_n)$ denote points of~${\Bbb R}^n$ and $z=x+iy$ a point
of~${\Bbb C}^n$. We use standard multi-index notations; thus for
$\alpha=(\alpha_1,\dots,\alpha_n)$ with $\alpha_j$ non-negative integers,
$z^\alpha=z_1^{\alpha_1}\!\dots z_n^{\alpha_n}$,
$|\alpha|=\alpha_1+\dots+\alpha_n$, $\alpha!=\alpha_1!\dots\alpha_n!$ and
$|z|=(|z_1|^2+\dots|z_n|^2)^{1/2}$.

$D$ denotes $(D_1,\dots,D_n)$ with $D_j=\partial/\partial z_j$, and
$\partial$~denotes $(\partial_1,\ldots,\partial_n)$ with
$\partial_j=\partial/\partial x_j$ (or $\partial/\partial y_j$,
etc., as the case may be).

${\scr P}_{m,n}$ denotes the set of polynomials on $n$~letters with complex
coefficients, of degree at most~$m$, and ${\scr H}_{m,n}$ the set of
homogeneous polynomials of degree~$m$ in ${\scr P}_{m,n}$ augmented by~$0$
(so that ${\scr H}_{m,n}$ is a vector space over $\Bbb C$). The number of
variables~($n$) will usually be suppressed in the notation, and we shall then
write ${\scr P}_m{\scr H}_m$.

For $f\in{\scr P}_m$, $f^*$~is the polynomial obtained from~$f$ by
conjugating its coefficients.

$E_n$ denotes the set of entire functions on~${\Bbb C}^n$, and $X_n$ the entire
functions of exponential type. $E_n$~and~$X_n$ may be considered as topological
vector spaces, so as to be duals of one another, in a standard way (compare
\cite{Incorrect version}). $F_n$~is the Hilbert space of entire functions~$f$
on~${\Bbb C}^n$, $f=\sum c_\alpha z^\alpha$ normed by
\begin{equation}
\|f\|^2 = \sum \alpha! \, |c_\alpha|^2.
\label{eq:Fischer norm}
\end{equation}

{\em Whenever\/ $\|\,{\cdot}\,\|$ appears in this paper, it designates this
norm.} We denote by~$\langle\,\, , \,\rangle$ the corresponding inner product
in~$F_n$. Finally, for $z$~and~$w$ in~${\Bbb C}^n$, $z\cdot w$~denotes
$\sum_1^n z_j w_j$.

The main objective of this paper is to prove Theorem~\ref{theorem-Ellipsoid}
below. The special case $a_1=\dots=a_n=1$ (or, rather, a formulation equivalent
to this) is stated as Theorem~2 in \cite{Incorrect version}. Unfortunately, the
proof offered there is incorrect (the error, on p.~522, lies in applying
Lemma~1 to the series of polynomials $\sum h_m$: here $h_m$~is in~${\scr
P}_m$, but not homogeneous, so Lemma~1 is not applicable). Even more
unfortunately, the generalization of Theorem~2 of \cite{Incorrect version}
presented as the Corollary on p.~525 is also based on an invalid deduction. The
error here is the assertion that the analog of Theorem~3 for the space~$X_n$
rather than $E_n$ can be proved by a similar argument---it cannot. Thus far we
do not know whether this Corollary is true as stated, or not.

The (we hope) correct proof of Theorem~\ref{theorem-Ellipsoid} below (Theorem~2
of \cite{Incorrect version}) is based on elementary potential theory, not using
techniques of \cite{Incorrect version} based on the Fischer
norm~(\ref{eq:Fischer norm}). Using the latter technique, we have not succeeded
in proving this theorem in full generality, but only within classes of entire
functions of limited growth (Theorem~\ref{theorem-Ellipsoid} below). But, in
return, we obtain an analogous result not only for the Laplace operator, but
for a fairly large class of differential operators. Thus, the Corollary on
p.~525 of \cite{Incorrect version} is shown to be true for each homogeneous
polynomial~$P$ of a certain `amenable' class (see Section~\ref{sec-Ellipsoid}
below), provided the given~$f$ is restricted to an appropriate subclass of the
entire functions. This is done in Theorem~\ref{theorem-Ellipsoid} of the
present paper.

\section{Dirichlet's problem for the ellipsoid}
\label{sec-Ellipsoid}

\begin{theorem}
\label{theorem-Ellipsoid}
Let\/ $\Omega=\{x\in{\Bbb R}^n : \sum_{j=1}^n a_j^{-1}x_j^2 < 1\}$, where
$a_j>0$. If $f$~is entire on\/ ${\Bbb C}^n$, the solution of the Dirichlet
problem
\begin{equation} 
\vcenter{\openup\jot\ialign
         {\strut\hfil$\displaystyle#$&$\displaystyle{}#$\hfil&\quad#\hfil\cr
\Delta u &= 0 &in\/ $\Omega$,\cr
       u &= f &on $\partial\Omega$,\cr}}
\label{eq:Dirichlet}
\end{equation}
extends to a harmonic function on\/~${\Bbb R}^n$. {\rm (}Hence it extends to
an entire function on\/~${\Bbb C}^n$ satisfying\/ $\sum_1^n D_j^2 u = 0$, and
equal to~$f$ on the variety\/ $\{z\in{\Bbb C}^n : \sum_1^n a_j^{-2}z_j^2 =
1\}$.{\rm $\,$)}
\end{theorem}

\begin{proof}
We can write the Taylor expansion of~$f$ as $f=\sum_0^\infty f_m$, where
$f_m\in{\scr H}_m$. The Dirichlet problem analogous to~(\ref{eq:Dirichlet})
with $f_m$ in place of~$f$ has a unique solution $u_m\in{\scr P}_m$. (This
is well known, and is recalled for the reader's convenience in
Section~\ref{sec-Conclusion} below.) To complete the proof, we shall show that
$\sum_0^\infty u_m$ converges uniformly on compact subsets of ${\Bbb R}^n$.

Let $\Gamma$ denote $\partial\Omega$. Let
\begin{equation}
u_m = u_{m,0}+u_{m,1}+\dots+u_{m,m}
\label{eq:u definition}
\end{equation}
denote the decomposition of $u_m$ into homogeneous polynomials; thus $u_{m,j}$
is in~${\scr H}_j$ and harmonic.

We shall now prove that, for every $R>0$, there is a constant $A(R)$ such that
\begin{equation}
\sum_{m=0}^\infty \sum_{k=0}^m |u_{m,k}(x)| \le A(R),
\quad \mbox{for $|x| \le R$},
\label{eq:A(R) definition}
\end{equation}
which implies the desired convergence of~$\sum_0^\infty u_m$.
\end{proof}

\begin{lemma}
\label{lemma-max limit}
Let $F_m=\max\{|f_m(x)| : x\in\Gamma\}$. Then $F_m^{1/m}\to 0$.
\end{lemma}

\begin{proof}
The proof uses only that $\Gamma$~is a compact subset of~${\Bbb C}^n$,
contained in, say, the ball $B$:~$\{|z|\le\rho\}$.

We have for $t\in{\Bbb C}$,
\[ 
f(tz) = \sum_0^\infty t^m f_m(z). 
\]
Fixing $z\in B$, $f_m(z)$~are the Taylor coefficients of the entire function
$t\mapsto f(tz)$ on~$\Bbb C$. By the Cauchy--Hadamard estimate,
\[ 
|f_m(z)| \le \frac{\max \{|f(tz)| : t \le T\}}{T^m} 
\]
holds for all $T>0$. Hence,
\[ 
\max_{x \in B} |f_m(z)|\le\frac{\max\{|f(\zeta)|:|\zeta|\le\rho T\}}{T^m} 
\]
Taking $m$th~roots and letting $m\to\infty$ gives
\[ 
\limsup_{m\to\infty}\bigl(\max_{x\in B} |f_m(z)|\bigr)^{1/m} \le T^{-1} 
\]
for arbitrary~$T$, implying the assertion.
\end{proof}

\begin{remark}
The referee has remarked that it would be of interest to obtain a sharp form of
Lemma~\ref{lemma-max limit}, and has kindly supplied a proof that the
exponent~$n/2$ in~(\ref{eq:A(R) definition}) can be improved to $(n-2)/2$.
\end{remark}

\begin{corollary*}
Let $v$, $v_k$ and $\Sigma$ be as in Lemma\/~{\rm \ref{lemma-max limit}}, and let
$D$~be a bounded open set in\/~${\Bbb R}^n$ containing the ball\/
$\{|x|\le\rho\}$. Then, for $x\in\Sigma$,
\begin{equation}
|v_k(x)| \le {\scr C}_n k^{n/2} \rho^{-k} \cdot
\max_{x\in\partial D} |v(x)|, \quad k \ge 1.
\label{eq:bound on v}
\end{equation}
Also, $|v_0(x)| = |v_0(0)| \le \max\{|v(x)| : x\in\partial D\}$.
\end{corollary*}

\begin{proof}
The statement concerning~$v_0$ is obvious, so suppose $k\ge1$, and without
loss of generality, assume $\max\{|v(x)| : x\in\partial D\}$ is~$1$.

Then $|v(x)| \le 1$ for $|x|=\rho$, by the maximum principle, so $|v(\rho x)|
\le 1$ for $x\in\Sigma$. By the lemma, we have for $x\in\Sigma$,
\[ 
|v_k(\rho x)| \le {\scr C}_n k^{n/2}, 
\]
which gives (\ref{eq:bound on v}), since $v_k \in {\scr H}_k$.
\end{proof}

\begin{proof}[of Theorem, completed] 
We have, for $x\in\Gamma$,
\[ 
|f_m(x)| \le \varepsilon_m^m, 
\]
where $\varepsilon_m$ is a sequence which tends to~$0$. Hence, for
$x\in\Gamma$, $|u_m(x)|\le\varepsilon_m^m$. By the Corollary, the $u_{m,k}$
in~(\ref{eq:u definition}) satisfy
\[ 
|u_{m,k}(x)| \le {\scr C}_n k^{n/2} \rho^{-k} \varepsilon_m^m |x|^k,
   \quad k \ge 1, 
\]
and $|u_{m,0}| \le {\scr C}_n \varepsilon_m^m$, for all $x \in {\Bbb R}^n$,
where $\rho=\min_i a_i$. In particular, for $|x| \le R$, 
\[ 
|u_{m,k}(x)| \le {\scr C}'_n \cdot A^k\varepsilon_m^m R^k 
\] 
holds for every choice of
$A>\rho^{-1}$. Thus (\ref{eq:A(R) definition})~follows since $\sum_{m=0}^\infty
\varepsilon_m^m \sum_{k=0}^m (AR)^k$ is clearly convergent (for assuming, as we
may, $AR>1$, the inner sum is $\le C(AR)^{m+1}$, etc.). This completes the
proof of Theorem~\ref{theorem-Ellipsoid}.
\end{proof}

\begin{remarks}
Variants of the theorem can easily be obtained from the above estimates, for
example, {\em if $f$~is of exponential type, so is~$u$}; indeed, $f$~is of
exponential type if and only~if $\max\{|f_m(z)| : z\in K\}$ does not exceed
$(A/m)^m$ (where $A=A(K)$~is some constant), holds for some (hence every)
compact~$K$ having $0$~as an interior point. The proof now follows in the
same way as before.
\end{remarks}

\begin{definition}
Let $\Lambda$~denote the class of positive sequences $\{\lambda_m\}_0^\infty$
with $\lambda_m \searrow 0$. To each sequence $\lambda=\{\lambda_m\}$
in~$\Lambda$ we define $B_\lambda$ to be the set of all entire functions
$f = \sum_0^\infty f_m$ on~${\Bbb C}^n$ (where, as usual, $f_m \in {\scr
H}_m$) such that
\begin{equation}
\|f_m\| = o(\lambda_m^m)m^{m/2}, \quad m\to\infty,
\end{equation}
and
\begin{equation}
\|f\|_\lambda = \sup \lambda_m^{-m}m^{-m/2}\|f_m\|.
\label{eq:lambda definition}
\end{equation}
It is easy to check that $B_\lambda$ is a Banach space with the norm
$\|f\|_\lambda$. Moreover, it is separable, indeed $\sum_{m=0}^k f_m$
converges to~$f$ as $k\to\infty$, for all $f \in B_\lambda$.
\end{definition}

\begin{lemma}
If $g \in {\scr H}_m$, then
\[ 
\sum_1^n \|D_j g\|^2 = m\|g\|^2. 
\]
\end{lemma}

\begin{proof}
$mg = \sum_1^n z_j D_j g$ by Euler's formula, so
\[ 
m\|g\|^2 = \sum_1^n \langle z_jD_jg,g \rangle = \sum_1^n \|D_jg\|^2. 
\]
\end{proof}

\begin{notation}
Throughout this section, $Q$ denotes $\sum_1^n z_j^2$.
\end{notation}

\begin{remark}
In terms of the Dirichlet problem, Theorem~\ref{theorem-Ellipsoid} says the
following. {\em For~$f\in B_\lambda$, where $\lambda$~satisfies (\ref{eq:lambda
definition})} (and hence, see the following remark, {\em for every entire~$f$
of order~$<4$}), {\em the problem~(\ref{eq:lambda definition}), where\/
$\Omega$~is the unit sphere, has a solution~$u$ that is (the restriction
to\/~$\Omega$ of) an entire function in~$B_\lambda$.} It would not be hard to
modify the proof to obtain an analogous result for ellipsoids rather than
spheres (and $B_\lambda$ replaced by some related class of entire functions),
by modifying Fischer's inner product so that $\sum_1^n D_j^2$ and
multiplication by $\sum_1^n a_j^{-2}z_j^2$ become adjoint operators. However,
we have been unable to obtain Theorem~\ref{theorem-Ellipsoid} (even for
spheres) by such methods. On the other hand (and this is the point of the
following section) these methods allow a generalization from $\sum_1^n D_j^2$
to a large class of differential operators~$P(D)$ ($P$~being a homogeneous
polynomial).
\end{remark}

\begin{remark}
To give some feeling for what (\ref{eq:lambda definition})~means, let us show:
{\em every entire function of order less than four is in~$B_\lambda$, for some
$\lambda\in\Lambda$ satisfying\/} (\ref{eq:lambda definition}). Indeed, suppose
$f$~is entire and
\begin{equation}
|f(z)| \le Ae^{|z|^\rho}, \quad z\in{\Bbb C}^n,
\end{equation}
where $A$ and $\rho$ are positive constants and $\rho<4$.
\end{remark}

\section{Concluding remarks}
\label{sec-Conclusion}

\subsection{}

The basic question underlying this paper is that of finding global continuation
of the solution to Dirichlet's problem when such continuation is known both for
the equation of~$\partial\Omega$ and for the `data function'~$f$. Even when
extreme regularity is assumed, for example, $\partial\Omega$~algebraic and
$f$~entire, few results are known (even in two dimensions) about the maximal
domain to which the solution extends harmonically, let alone the nature of the
singularities that may arise. This is in contrast to the situation for Cauchy's
problem, where, for example, complete results are known in two dimensions,
based on the Schwarz function (compare \cite{Ref9,Ref12}). Moreover, G.~Johnson
\cite{Ref8} has obtained complete results for the Cauchy problem when the
initial data is an entire function restricted to a quadric surface (this for a
class of differential operators including the Laplacian). So far, there are no
results of this precision available for the Dirichlet problem.

We have already spoken of the question of whether ellipsoids are characterized
by Theorem~\ref{theorem-Ellipsoid}. In this connection, recall that when
$\Omega$~is an ellipsoid, the solution of Dirichlet's problem with data
in~${\scr P}_m$ also lies in~${\scr P}_m$ (this is a classical result
from the study of ellipsoidal harmonics, and we used it in proving
Theorem~\ref{theorem-Ellipsoid}). This has a kind of converse, which one
readily sees as follows.

\subsection{}

Concerning the material in Sections
\ref{sec-Introduction}~and~\ref{sec-Ellipsoid}, some questions remain.
Especially, it seems of interest to know when the set of solutions
of~$P^*(D)(P-1)f=0$, $f\in B_\lambda$, not merely is finite-dimensional (for
which we gave sufficient conditions, in terms of $P$~and~$\lambda$) but
consists of $0$~alone. Perhaps the uniqueness assumption in
Theorem~\ref{theorem-Ellipsoid} could be omitted---we know of no
counterexample.

\begin{acknowledgements}
This work was done while the first author was visiting Stockholm in the spring
of~1991. The first author is indebted to the Royal Institute of~Technology for
support and for providing a congenial research environment.
\end{acknowledgements}

\begin{thebibliography}{99}

\bibitem{Ref1}
{\bibname V. Bargmann}, `On a Hilbert space of analytic functions and an
associated integral transform', {\it Comm.\ Pure Appl.\ Math.}\ (1961) 
187--214.
%
\bibitem{Ref2}
{\bibname F. A. Berezin}, `Covariant and contravariant symbols of operators', 
{\it Math.\ USSR-Izv.}\ 6 (1972) 1117--1151.
%
\bibitem{Ref3}
{\bibname C. A. Berger \and L. A. Coburn}, `Toeplitz operators and quantum
mechanics', {\it J. Funct.\ Anal.}\ 68 (1986) 273--299.
%
\bibitem{Ref4}
{\bibname C. A. Berger \and L. A. Coburn}, `Toeplitz operators on the
Segal--Bargmann space', {\it Trans.\ Amer.\ Math.\ Soc.}\ 301 (1987)
813--829.
%
\bibitem{Ref5}
{\bibname V. Guillemin}, `Toeplitz operators in $n$~dimensions', {\it
Integral Equations Operator Theory\/} 7 (1984) 154--205.
%
\bibitem{Ref6}
{\bibname J. Janas}, `Toeplitz and Hankel operators on Bargmann spaces',
{\it Glasgow Math.\ J.} 30 (1988) 315--323.
%
\bibitem{Ref7}
{\bibname J. Janas}, `Unbounded Toeplitz operators in the Bargmann--Segal
space', {\it Studia Math.}, to appear.
%
\bibitem{Ref8}
{\bibname G. Johnson}, `The Cauchy problem in ${\Bbb C}^n$ for 
second-order PDE with data on a quadric surface', in preparation.
%
\bibitem{Ref9}
{\bibname D. Khavinson \and H. S. Shapiro}, {\it The Schwarz potential in\/
${\Bbb R}^n$ and Cauchy's problem for the Laplace equation}, 
Research Report TRITA-MAT-1989-36 (Royal Institute of Technology, 1989).
%
\bibitem{Ref10}
{\bibname D. J. Newman \and H. S. Shapiro}, `A Hilbert space of entire
functions related to the operational calculus', mimeographed, Ann Arbor, 1964.
%
\bibitem{Ref11}
{\bibname D. J. Newman \and H. S. Shapiro}, {\it Fischer spaces of entire
functions}, Proc.\ Sympos.\ Pure Math II (Amer.\ Math.\ Soc., Providence, RI, 1968) 360--369.
%
\bibitem{Ref12}
{\bibname H. S. Shapiro}, {\it The Schwarz function and its generalization to 
higher dimensions\/} (Wiley, 1991).
%
\bibitem{Incorrect version}
{\bibname H. S. Shapiro}, `An algebraic theorem of E.~Fischer, and the
holomorphic Goursat problem', {\it Bull.\ London Math.\ Soc.}\ 21
(1989) 513--535.
%
\bibitem{Ref14}
{\bibname H. S. Shapiro}, `Analytic continuation of the solution to 
Dirichlet's problem', in preparation.
%
\bibitem{Ref15}
{\bibname B. Yu.\ Sternin \and V. E. Shatalov}, `Continuation of solutions 
of elliptic equations and localization of singularities', preprint, 1991.
\end{thebibliography}

%% Note that there is no paragraph spacing between two affiliations
%% on one line

\affiliationone{Department of Mathematics\\ 
University of Arkansas\\ Fayetteville, AR 72701\\ USA}                                             
%
\affiliationtwo{Mathematiska Institutionen\\ 
Kungl.\ Tekniska H\"ogskolan\\ S-100 44 Stockholm\\ Sweden} 

\end{document}