﻿ Hsu Pei, 1980-2004
 UFDC Home  |  Search all Groups  |  General Manuscripts & Archives  |  Kai Lai Chung Papers |   Help

Hsu Pei, 1980-2004

Material Information

Title:
Hsu Pei, 1980-2004
Physical Description:
Unknown
Language:
English
Creator:
Hsu Pei
Chung, Kai Lai
Physical Location:
 Box: 1 Folder: Hsu Pei, 1980-2004

Subjects

Subjects / Keywords:
Mathematics -- History -- 20th century

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
System ID:
AA00007242:00001

Full Text

April 2-, 1980
Hsu Pei,

The summer quarter begins in the last part of June, but there
may be no advanced courses given which will suit you. After you
come you can find out what to do. If you want to live in a dormitory
(this is the best way) you may have to register in the summer. In
any case you will have ample time to prepare for your research. I do
not know whether you have had courses in iarkov processes. You can
start with my paper with Rao, A new setting for potential theory, which
I lectured on (only about one third), and fill in the background from
other books. Blumenthal and Getoor's book is more than sufficient.
Meyer's T'rocessus de Markov is also very good. I understand that Della-
cherTe's book was studied in the seminars. If so here is a good exer-
cise for you to try. The result is known but maybe you can find a bet-
ter solution. Define the hitting time TA as usual, A a Borel set,
and also
TA = inf (t>O: X(t-) E A).
It is easy to show that for a Hunt process, the second is always greater
than or equal to the first. Under Hypothesis (B), they are equal a. s.
For a discussion of IHyp. (B) see Meyer's Frontiere de Martin (or ask
Yen). In our paper we tried to prove (B) under our conditions, but
we had to add a little more. Ahyway the exercise above is very interest
ing because it is intuitively meaningful. It is surprising that the
proof of equality under (B) is not so easy. One can learn a lot of
things by trying to prove such a result.
Here are some open problems in our paper which would be interesting.
(1) Find a good condition under which the set Z is empty. One can
even begin by assuming all paths continuous, but maybe the con-
ditions &n u should be strengthened also.
(?) We proved that (B) is equivalent to: for each compact K:
PK1 = UA where (AC K
But maybe there is a simpler proof of this without first prov-
ing (B). Then the latter will follow.

Someone 6ent me some stencilled notes on problems concerning our
paper referred above. Unfortunately there were several errors. This
kind of work is not easy and needs time for a deep understanding and
to learn the techniques. It is a good way to think of such problems
while studying the books. I hope there will be no further felay for
your trip. Wh 1 ling waiting you may as well get started.

Sincerely*

7?f- 7/5/ 1
What made you want to edit a book of my work? Springer asked me some
years ago and WSPC more recently. Of course I can do it myself.
It sounds more "fun" (vulgar americanism. ) to have a former pupil as
editor and commenfries by others. I may well do it but first a few
A
questions ese You can reply by mail or fax (650)857-9532. No email.
(1) Prepare a list of seleommwher due and by hom send
it to me. I will subtract andladd. The final decision is mine.
(2) Have you published any book yet? If not you will be shortchanged by
national"'I suppose between Chinamen you can deal with him.
(3) Other business B can be discussed later. If the published
articles are to be just photo reproduced there will be "problems". WSPC
did a good job in printing my book. I have not seen any book by Yau's
Co. Tell them to send one for my inspection.
Good Luck (you need it to deal with those sharpies).

P. S. / call your interest in foreign language (I approve): can you -re-
my recent article) in Italian? That must be in.- I SELECT. read
t Phone only when abs. nec. Yau's cohorts have been te@ering me
with repeated unwelcome phone calls from Pekin etc.! Those Chinamen
never learn. Q l ea

FAX 6307895908

Hsu Pei: V 7/2/A
I would like to receive your list of selection before August 20, because
we shall be away soon after till September x. It need not be complete.
Your suggestion of a Chinese article is good: my very first math fubl.
1936/7 is available in copy and it would please me if you can translate
it with my support. I will add a sad note myself, as my then good friend
co-author (son of famous Chinesefwhom you probably did not know) com-
mitted suicide sMKKxax before the article was written. I will also
give you a reprint of my (so far unique) Italian article which will be
included (without translation). There may be one or two short C. R.
notes in French to be reproduced. This historical, reminiscent aspect
of the collection is what interests me most. Beethoven said that he
loved his 8th Symphony better than the publicly acclaimed 7th (or ?th).
% Many ot r authors preferred their lesser works to popular ones. As Chin
Sese scholar (not modern stupid Americani) you will understand this.
Regarding commentaries, of course you must do one for my Excursion paper.
I may do a couple for my dead co-authors --- but I have not really thou-
ght about these possibilities yet --- there is Tim However, can you
set some "deadline" for the completion of your work, and I insist the
publisher vl9 sets a deadline for the publication with heavy penalty
fmr nonfullfilment (as in my recent contract with a big firm). These
are the problems a novice author tends to overlooks but I will alert
you when the time comes. Yau's firm would be OK if it (he) pays good
royalties (which he did not to some authors, I heard).
Your invitation is appreciated but let us simplify the matter by
meeting for lunch in a nearby pace, Awerican st p. To-day is Sat.
next Thur. or Fri. is OK, just phone me at (650)857- M as early
as you can and I will tell you when and where. See you t en.

Hu Sen also wants the book, he says he knows you Pwell".

k ,

847-491-8906 Prof. Hsu, P. Math Northwestern

8/20/2001
Hsu Pei,
I have looked over your list and checked with a few reprints in my
file. Your selection is good especially in the later stage. In the earlier
stage, long before your time, there are a few "popular" articles
I shall add, and also a couple of my personal favorites9 Among the for-
mer category, have you ever heard of the Robbins-Monro method? Jack Wolfo-
-itz was going around complaining about the unsolved problem of normal limit
as hard as Riemann hypothesis. After I did it the statisticians immediate-
ly changed their tone saying that normality is not as important as their
own (trivial) estimate .... That is the way of "statistiCs". There is a
whole book of Cambridge Tract'(by an Indian) on it.
Before I send you my revision, please let me know whether the articles are
to be re-typed on Latex or just photographed; take a look at my edition of
P. L. Hsu's works. I was told by Springer Verlag that there are new compute
er technology .... which I do not understand but maybe you do.
If you prefer AMS/IP/ please tell me what books 44 -(they) have produced and
send me a good one now for inspection. If 'fc L er is the same as Yau's
Hongkong business (which tried to get my book with Zhao), you better make
sure that they pay good royalties. Of course we are not doing it for dirty
money but there is no ,eason to let them cheat us out of their profit. X7 4
I am getting a reasonable one from a new edition of my Elementary Probabi-
lity. If need be, but after you have dealt with them, I can talk with them.
A
Regarding the total number of pages, the final decision is ours, not theirs.
It 1/ooks there will be additional pages after my addition and subtraction.
For instance, have you seen my "Mathematics and Applications"? I want it
there as well as two Reminiscences of Levy and Do(e)blin. Many readers will
read these rather than the other articles. My review of Doob's book may
also be considered. We are going to N. Y. 8/23, and will be back 9/8. Mail
will reach me at ( ) 30 White Oak Lane, Southampton, N. Y. 11968
/ Tel. (631)283-5911 [before noon N. Y. time]
If possible send me a fax before or on 8/22. Regards,

P. S. Do you want to include our unique joint artile (which you may com-
ment on!)

i-I v
^ w^^

October 1, 2001
Hsu Pei: The law of diminishing returns is operating so I stop. Just
two more inclusions) one of which is up to you: our article on Neumann
"C b 26 63 c/9Fzc]
problem; the other "Crudely stationary counting processes" which is
a A
"refreshing". You are so efficient so maybe you can compile the revised
list, with further suggestions + or as you see fit, send it to me before
soMe So
I wa=u? suggest commentators --- please do also from your side. I will
A A
make copies the minor corrections oft the list and mail it to you in due
time. Nothing important so please do not wait.
As we have already discussed, let the so-called editor respond before
we make any revision of our list. By the way, please feel free to make
other suggestions at any time.
I have made a copy of my Chiinese debut and found it rather difficult to
read. If you agree it can simply be reproduced as a sort of museum piece
perhaps with an English abstract? Of course if you wish to translate it
that will be fine and I can help you with the obscurities there. Re
the Italian article which you will receive soon, why not just print it
as is --- about time that 4 americans learn some foreign language!
May I ask you to telephone Jing Qing at your area code (630)960-9095?
Ask her the status of the new end enlargedd edition of Green, Brown, and
Probability that she is editing. If you don't feel like it just have
one of your (Chinese) students do it ON MY BEHALF. OK? Thanks. IAll
give you a copy when it appears. 6 u'w

S= "Have a Good Day'." q I

d p yiip

10/2/2001
Hsu Pei: Ig 4
I have no objection to includA that Lip paper. The Chinese title
is / If you want a copy send
14 dollars and I'll get you one. I th ught Wu Rong gave you a free one.
Continuous Parameter Markov Chains, Proc. ICM 14-21 August 1958
My file of reprints is a mess. If you really want any of them, I
can try to find it and send it to you. I will send a copy of my Chinese
paper when I get to the office (rarely).
I (we) have been to Marseilles many times, last time c. 1990 I
gave a talk when there were at least three Chinese students. Do not let
the take you to their "cafe" as the food (is inedible --- my daughter
said the same. What is the name of the director of institute there: Pr..
who dOes stochastic PDE --- say hello to him. Bon voyage.
If you really like history I can tell you many good stories,
e. g. about Hadamard at the Boston ICM, 1950, He was in Pekin a year be-
fore I went to Tsinghua.

Rick &.CA

P t n et s r % = z .e Z -

63 1 3.

630 789 39o8

Hsu Pei: Dec. 9, 2001
Your solution of the problem is mxx brilliant but you made a big misprint:
"inequality")' The day before I got it Ruth found the EQUALITY in Riordan's
book where messier proofs are given. I can now do it by a cute induction
but yours is of course better. Is it new? This EQUATION was first given
in 1795 [do you know the history of this year in Paris] at the Academie
(with date when it was "lu") by two famous mathematicians whose names you
will see in my article, cited in Todhunter's History where we probably saw
it in China around 1935 (the year before the goddam japs invaded They
still have not apologized when Kiang leming went there --- he was too weak
to insist on it but maybe he wanted to get jap yens ...?] A few years ago
Bru (remember? O44btAa copy of the article but it was written in such a way
(1795) that I could not decipher it. Now I read it and found that they stat-
ed their first (obviously basic) Lemma that is that EQUATION. They gave 5
special cases but not a complete proof, saying things like "there is no dif-
ficulty (to extend)... and so ,they Lwij ho(i the dettails). Now with
your proof it is in fact "easy to see"J if fe formulateJ # as you did. But
(Lale ttc arfe j ^rs)
now I want to see how these two guys used the equation to prove the general
proposition in our paper, namely: knowing that the Sun has risen n times
in the first n+m aas, what is the probability that it will rise in r days
among the next r+s days. We did that and the result is given --- I have
not yet checked to see if Mey gave the explicit result (as I said, the ar-
ticle is hard to decipher due to its style and "notation"). However maybe
you can see how their Lemma would yield such a solution. 1955 was 66 years
ago but I clearly remember that it took us a long time to do the calcula-
tion an iinally it was done by a kind of induction. I must now read that
part (which I did not) to see ...
So I must delay for a few days returning your list with suggestions. Ee
my paper with Joe, of course I started it but he did something more: any
Hunt process reversed becomes a left-continuous moderate Markov process. By
the way "moderate' larkov" was first defined in my (cta .iath.) paper. So let
us keep this and maybe ask Paul Andre Meyer to comment on it [I think he
was the referee]. If necessary we shall sacrifice some other article.

A\ /4P ?

Dec. 18Q 2001
Hsu Pei: it g
I will send you the two other pages of your list with suggestions. Your
procedure of asking them-guys is excellent but let me alert you to a cou-
le of things. I have suggested a few co-authors who may be too old to re-
spond, let them be. French guys are notoriously bad correspondents, even
Neveu (my former "assistant" on a project) though he is very hospitable
in Paris. Another person whose name I will withhold (guess) may respond
in a self-serving semi-lying way i- he sees that he has been shown up
A nup
(subtly) in my paper which WE ask him to comment. Some are so sloppy
A
,e. g. Snell if you know him, very nice guy but sloppyYbut I am asking
him to comment on my "Prob. and Doob" --- he interviewed D. before I gave
that talk in Urbana. Etc. etc. Let's see what, transpires. Paul'Andre
is also a tricky type though not mechant. [Walsh used to complain that
he favored French over American --- John has a point].
OK my chores are done. I may later change my mind ...
You did not mention in your latest fax whether you hak received the
173S 9en er 1,
Chinese reprints. I worked hard on our proof of the Spnrise theorem, '1Ze?
and am asw most pleased to discover that we had in fact proved that IDEN-
TITY in a different way than all the other proofs including the quickest
one you found. I shall be glad to comment on it! r
If you have any questions please let me know. 'A

INPORTANT QUESTION: We have exchanged numerous fax(es?). P ease tell
me if you have ever experienced any difficulty in sending yours to us?
I have no problem at all sending mine to you (although some jerks have
told me that they did not receive some of my fax --- they might be lying.]
So I "need to know" your experience/
(US government officialese)

io4 \ 40O 7? 30.-.5

Jan. 16, 2004
Hsu Pei: .

Re the stable tail, it should be proved without using ch. f. or

Laplace or Suiordinator all that jazz. Feller tried hard (see Vol. 2) with-
out success. I proposed the problem to the rogs: Yor and Biane. Zero.

Read Levy's original proof with those Poisson jumps la er the frogs made

a big deal about. The result is stated in Kol-Gne without proof. Of course

if one uses the full force of the analysis of domain of attraction etc.

maybe it will come out but nobody has ever seen it anywhere. FPtzsim-

mons and an Italian followed Levy's WAY (TAO) and maybe ... but that's

not kosher as I said. Now if you want to test your own "force", try first
the eaty symmetric case exp(-Itla) and even the inversion formula (all in

my Course) heuristically (1) to see what comes out. See Exercise 7 in Sec.
6.4 (?) where part of it is done: of course from the ch. f. near 0 (as you
thought). Now get the rest of the tail if you can. I will tell Hou to

double your "compensation" and shame those weak-sisters frogs.
Now a few (last) words about making copies. Who told you the big LIE
that it can't be done WELL?? I made hundreds of pages from bound OLD vol-

unes years ago when you were a student. Everybody has done it. Have YOU
tried yourself? As I said if "too busy" thepay a (Chinese) student to

do it. Anyway, that is YOUR problem, eccoci.

'AtV V

% \y/