何毓琦的个人博客分享 http://blog.sciencenet.cn/u/何毓琦 哈佛(1961-2001) 清华(2001-date)

博文

On Research #10 - some specific advice to Ph.D students 精选

已有 20678 次阅读 2007-11-25 23:26 |系统分类:科研笔记

On Reseaqrch #10 - Specific Advice for Chinese Ph.d students looking for Guidance

My series of 9 articles on research and education

http://sciencenet.cn/blog/user_content.aspx?id=2224

http://sciencenet.cn/blog/user_content.aspx?id=2265

http://sciencenet.cn/blog/user_content.aspx?id=2501

http://sciencenet.cn/blog/user_content.aspx?id=2768

http://sciencenet.cn/blog/user_content.aspx?id=2950

http://sciencenet.cn/blog/user_content.aspx?id=3859

http://sciencenet.cn/blog/user_content.aspx?id=8186

http://sciencenet.cn/blog/user_content.aspx?id=8412

http://sciencenet.cn/blog/user_content.aspx?id=9328

as well as the two articles on US vs. Chinese phds and postdoc positions

http://sciencenet.cn/blog/user_content.aspx?id=11101

http://sciencenet.cn/blog/user_content.aspx?id=11258

http://sciencenet.cn/blog/user_content.aspx?id=11434 (Chinese translation of 11258)

has prompted a number of direct inquiries to me typified by:

“ I am a phd student in ABC university in China. My advisor has over fifty

ph.d students and can give me little advice or guidance.

1.Can you help me?

2.My field of interest is in XYZ topic of which you are an expert.

Please tell me if my selected tropic is proper and correct? Can

you give me some pointers?   “  

The problem with this kind of open-ended request is that if I want to

answer the inquiry faithfully and not mislead anyone then I literally have

to take on the person as my own ph.d student and begin a 3-4 year intense

relationship as described in my articles above (in particular article #

2950). This of course is not possible. In fact, I have even stopped taking

on Tsinghua students because I cannot assume the responsibility of an

advisor in the way I wanted at my advanced age (我已经收了关门徒弟).

However, I do sympathize with the plight of these Chinese phd students

who needed help.

Thus, let me give a specific set (yet generally applicable) of advices

below to help answer these inquires at least partially.

First I’ll start with my own experience. (Note: what I say below is not

because I want to glorify my own achievement some 46 years ago. But I

relate the experience to illustrate what specific advices to follow. Read

should ponder my own experience and this paragraph  again after you

have read my specific advices below)  During 1959-1960 when I started

at Harvard intending to do a thesis on control and system theory, Harvard

had essentially no one in the subject area. I was lured to graduate study

there by a pamphlet entitled Computer-Control Engineering issued by the

Harvard computation laboratory under Howard Aiken (note: Howard

Aiken was generally considered to be the Father of digital computers and

built the Mark series of computers during 1944-1955. However, his view

on computer control was largely misdirected. And by the late fifties he

was nearing retirement no longer active in research even though he still

had a great reputation.) I quickly found out that there was really no

competence on the subject of control at Harvard. One junior faculty in the

subject was leaving since he was denied tenure. Only one other lecturer, a

recent graduate, was on a year-to year appointment to teach a course on

feedback control. Thus, there was essentially no one to guide me. In

desperation, I begin to read any papers I could find on control being

published at the time. I came across a paper by R.E. Kalman and J.

Bertram on dead beat control – the problem of controlling a general nth

order linear sampled data system from any initial condition to zero in n step

Associated with the solution of the problem was a condition then known

as the Kalman-Betram condition. After studying the problem on my own,

I came to realize that the K-B condition really is a condition on linear

independence in linear algebra. So, I wrote up the idea together with a

application and submitted it as

an extension to the Kalman paper. Simultaneously, I wrote to Kalman

asking for more of his work. At the time, Kalman was not yet famous (his

famous Kalman filtering paper was still one year away from publication).

He was appreciative that someone, a graduate student, had studied and

was interested in his work. He not only sent me several of his preprints,

but also recommended my extension of his work for presentation and

publication, at the first American Automatic Control Conference in 1960,

It is there I first met Kalman. During a conversation, he realized that I had

thoroughly studied his work and was in fact one of the few people at that

time appreciated his contribution (Note: At that time almost all the well

known work on control were done in the so-called frequency domain

utilizing Fourier and Laplace transform methods. Kalman’s approach

using time domain and differential equation model for dynamic system

was very unorthodox and viewed with suspicion by the establishment. But

for a graduate student who has not been brain-washed by orthodoxy, new

ideas were easier to absorb.) Kalman further invited me to co-author a

paper with him considerably generalized the idea of linear independence

in dynamic systems to the concept of “controllability” – a fundamental


idea in control. The paper quickly became a classic of the field. I also


benefited from getting an advanced prepublication copy of the book by R.


Bellman – Adaptive control, a guided tour - from my classmate , Stuart


Dreyfus who was working as a programmer for Bellman at that time. I


learned a great deal before the general public from the book. These two


things more than any other enabled me to finish my thesis, and in addition


you can say my career in the field was launched.


Looking at the above experience, let me recommend the following

approach to every Chinese ph.d student who asked me or wanted to asked

me the question at the beginning of this blog:

1.Select a paper or papers of real interest to you with the subject

matter that you hope to make your thesis topic. It will be useful if

the paper is authored by young and beginning scholars (say

assistant or associate professors who are in the process of

establishing their reputations. They have more time and more

interested in publicizing their work with others).

2.Work hard and study the paper by yourself  thoroughly to

understand it to the point you can say something intelligent and

new about it. THEN AND ONLY THEN write to the author of the

paper. This way the author will know you are someone who

appreciates him/her(a 知音). He will be much more motivated to

correspond with you knowing you are serious and not looking for an easy way

out.


3.Ask for his preprints (papers not yet published) and discuss with

him his work in progress in an intelloigent way.

4.Once you established rapport with one or two scholars this way,

you are well on your way to be guided by experts on your thesis

research.

5.The principle to observe here is that your relationship with your

advisor is a form of give-and-take. S/he should get something from

you (the minimum being your deep appreciation of his work) in

return for his attention and guidance. For example, I have said this

before in my first article mentioned above (#2224) that the second

of my three conditions for accepting a thesis is “that I should learn

something new from the thesis”). Scientific correspondence  and

discussion is a two way street.





Lastly you should always remember that THERE IS NO MAGIC

FORMULA OR ROYAL ROAD TO LEARNING BUT HARD FOCUS

WORK.  There is no free lunch in success. Maps, directions, and travel

book will help you to avoid blind alleys and wasted effort, but there is no substitute for being there yourself.  Good luck!




http://blog.sciencenet.cn/blog-1565-11577.html

上一篇:怎样才能到国外做博士后 - translation of my postdoc article
下一篇:关于SCI 及榮誉会員的選举方法
收藏 分享 举报

1 李天成

该博文允许注册用户评论 请点击登录 评论 (11 个评论)

数据加载中...

Archiver|手机版|小黑屋|科学网 ( 京ICP备14006957 )

GMT+8, 2017-9-19 23:26

Powered by ScienceNet.cn

Copyright © 2007-2017 中国科学报社

返回顶部