系统科学基础课程的教材草稿

最近正在写系统科学基础的教材,欢迎大家提意见。我会一直更新这个工作版本

目前,第二部分,系统科学的数学物理基础已经基本完成,除了随机过程。

第一部分,系统科学导论,还需要很大的功夫,找例子,分类,复述,写评论。
第三部分,系统科学的基础理论,非线性动力学部分也基本完成。复杂网络没有动笔,但也不是难事。其他的部分还需要一些时间。
第四部分,计算附录以及概念地图学习方法,应该也花不了太多时间。

精简教育:Teach Less, Learn More

教学设计上,课程内容与方向的设计,从小学甚至更早开始,就需要一个一般原则,而不是注重各种各样的奇技淫巧,歪门邪道。从知识的角度需要学习的东西很多,学不过来,但是,从对学科的一般认识以及进一步自己学习的角度,需要学习的东西很少很少。

From the beginning of a person’s education, say elementary school or even earlier, there should be a principle, or a general guide line, based on which the whole set of courses curriculum, teaching methods and styles, learning methods and techniques, should be designed. Without such a principle, all those fancy or not-so-fancy techniques and all those creative or not-so-creative teaching methods, are just clever but useless/dishonest tricks. From the perspective of learning as to acquire knowledge, there are too much to learn in every field; from the perspective of learning what is the field and building up a basis from which one can start to learn things in that field by ones’ own, however, there is really not much to learn.

这个一般的原则,也就是正道,我认为是:精简教育,学习最核心的东西。那么什么是最核心的东西呢?学习对一个学科的一般认识(也就是通常所说的:学习某专业就要像一个某某家一样思考)还有进一步自己学习的基础。

This principle, I believe, is: Teach less, learn more. Learn only the core part of a field. So what is the core part? I believe it is to form a general scope of the field, or what commonly expressed as “when you learn a field, you should think like the best scholars in this field”, and learn sufficient basic concept and skills so that one can learn more about this field on ones’ own.

进一步自己学习的基础还有待整理。这里,我写下来我对学科的一般认识。

While waiting for answers for the question of what should be the basic but sufficient concepts and techniques of each individual fields, here I want to talk a little bit on what I think a top expert in the field of the following several fields look at those fields.

数学家:把现实世界抽象成数学问题,把抽象出来的数学问题解决;
物理学家:把现实世界抽象成物理问题,把物理问题转化成数学问题,解决抽象出来的数学问题;
计算机科学家:把现实世界抽象成数学问题,用算法的方式和角度解决这个数学问题;

Mathematicians: Convert real-world phenomena into well-defined mathematical problems, if not well-defined then develop mathematics first to make the problem well-defined, and then solve the abstracted math problems.
Physicists: Convert real-world phenomena into physical problems and then express the physical problems as math problems, and then solve the math problems.
Computer scientists: Convert the real-world phenomena into math problems, and then solve the math problems in an algorithmic fashion.

通常的语言使用者,母语:读和写,其中最主要的是把自己的想法和说的话,转化成文字;
专业的文字使用者:有想法和想说的话,可以用来表达,加上把这些想法和想说的话转化成文字的能力;
通常的语言使用者,非母语:听说读写,利用或者不利用母语,把自己的想法和想说的话,转化成文字和声音。

Language users, native: Reading and writing, convert their ideas and what they want to express into written words.
Profession language users: the previous, plus find something about which they have some ideas or have the desire/inspiration to say something.
Language users, non-native: Listening, speaking, reading and writing, using their native language or not, to convert their ideas and what they want to express into spoken and written words of the target language.

其他科学家,以后再添上。

When I, myself, have the inspiration and also indeed have something non-trivial to say, I will add my understandings of other fields.

所有的不是以这个核心能力为目标的教育和教学都是有害的。学习者可能变的越来越有学识,但是同时变得越来越古板,没有创造性,越来越像一个google服务器——一个供检索的知识容器。

All teaching, or more generally education, if is not based on this general principle, do more harm than good to the learners. People might become more and more educated, but meanwhile become less and less creative, more and more like a machine, or I would like to call it, a google server.

我希望有一天,有人能够把这些真正有必要学习的,对学科的一般认识,进一步自己学习的基础,都整理出来。

I wish one day, there will be others who also respect this picture of education and go through all fields to comb out those core parts, including the general scope of the fields and also the minimum set of basic but sufficient concepts and techniques.

真正的减负在于用更少的时间学会更好的更核心的东西,并且学会高效的学习方法。

If we want our next generation to learn happier, it can not be achieved by simply reducing the work load such as school hours and size of homework, but can only be reached by teaching only the core parts, and teaching it better.

Prof. Alberto Canas added that once the principles/big ideas have been identified, problem-based learning and project-based learning should be used to bring those big ideas down to the earth.

So I suggested the following formula that: Teaching = guidelines/principles/big pictures/big ideas + examples + logic structure. Concept mapping and concept maps can be helpful in constructing/presenting the first and the last, problem-based learning can be helpful in implementing the second.

三亚路线图会议回来收获小结

陈天平老师关于Blind Signal Separation的工作非常有意思。有三个原始信号,我们只知道通过线性混合之后的三个信号,问有什么方式能够恢复原始的信号。神经网络算法能够完成这件事情。其中用来做混合的矩阵的大部分信息也能够得到。

听到企业、商业界人士的一些很不错的想法,很大胆的想法,很有意思。将来理解的更透彻的时候再写感想。

得到几位前辈的的教诲,很有收获,在这里默默表示感谢。

Fokker-Planck方程的Lie代数解法和Lattice FPE解法

我现在再利用Fokker-Planck方程描述量子输运现象,正在寻找FPE的解析和数值求解的方法。

在黄祖洽和丁鄂江的《输运理论》上的5.12节读到FPE的Lie代数解法,非常有意思。对于不含时的线性的FPE,
\begin{equation}
\frac{d}{dt}P = L P
\end{equation}
求结果过程就是写出算符
\begin{align}
e^{Lt}
\end{align}
的具体显示形式。由于这是一个一般的线性算符,写出具体形式是不容易的。注意到这个就是一个指数映射,所以,如果我们把$$L$$看成一个李代数,并且找到合适的表示,那么这个演化算符就是这个李代数对应着的李群,其表示可以通过李代数的表示得到。

巨神奇的是,这个方法,是方福康老师发展起来的。

最近,通过许爱国师兄发现,格气和元胞自动机的方法,可以用来求解FPE,叫做Lattice Fokker-Planck Equation。很有意思。

在我的工作中,这两个方法,都值得尝试一下。

吴金闪这个人的小结

核心能力:认识问题深刻;思考广泛没有约束;看文献整理文献提炼问题的能力超常;技术上对于矩阵和矢量,还有Monte Carlo模拟,Green函数,Fokker-Planck方程已经内生化,基本上不需要思考就能拿起来用;思想上对于真正的学术研究有追求。

主要问题:泛学术化——做事情的方式和研究工作一样直接、一针见血或者企图一针见血,甚至日常交往也用Socrates(苏格拉底)诘问法;对人心的理解比较幼稚——认为是好的东西就能够得到推广,尽力帮助别人别人就会感激从而也获得别人的帮助,乃至更多的人获得帮助。不过最近有提高。

其实,对于这些主要问题也不是不知道,以前总认为,做学问的人,就算有这些问题,又怎样呢,反正很多其他的事情可以不去处理的,或者别人会帮这个做学问的人处理好的。现在渐渐地发现,其实,人和江湖是分不开的。为自己的成长感到伤心。不过,也许,这种伤心就是成长的一部分。