Thursday, May 31, 2012

Funding

Hmm..indeed we can use school funding for purchasing experiment software...That's sounds cool and motivating :)

Wednesday, May 30, 2012

Some Relevant Research Area

A technique for detecting invariant
Loop Invariant Generation


Worst case execution analysis

Worst-Case Execution Time Analysis







Small test case

To test an idea, it is good to test the water with some small example, then to scale it up.
This not only apply in research, but in many aspect of life.

Introduction - with insightful view

When reading the introduction, i would be more happy to view from a view point where nobody has provided. And feel bored when read something repeated in many papers like safety property is "something bad never happen", I hope those could be introduced as little ad as concise as possible.


Saturday, May 26, 2012

Interleave and Skip

In the interleave expression, the Skip Process can't be run unless all processes in the interleave process becomes Skip.
Skip||a->Skip||b->Skip
For example, for above expression, it has two possible next steps, which is events {a,b},but the Skip cannot go.

The reason the operational semantics of sequential expression is defined as
P->(CompleteEvent)P'
-------------------------
P;Q->(tau)P';Q

and Skip process is defined as

--------------
Skip->(CompleteEvent)Stop

now let P=Skip||a->Skip||b->Skip
if P allows that Skip to go, all other expression will force to stop!


Thursday, May 24, 2012

Go for simplistic research prototype

Recently am building a research prototype, I have tried to add as many fancy non-technical features as it could be, it turns up that most are wasteful result. Now, I would try to go for a simplistic style. Logically, this is a better way to go. Let's try.

Soundness and Completeness


When decompose/abstract the system S to
S1||S2||S3

Soundness:
Whatever true in S1||S2||S3 true in S is sound
S1||S2||S3 is subset of S

Completeness:
Whatever true in S true in S1||S2||S3 is complete
S is subset of S1||S2||S3

When Soundness and Completeness hold
Checking S is equivalent of checking S1||S2||S3

Strategy Change

Hmm..I keep looking for the 1st rank idea, but fail.

Now, I would try the other way round, start from simple, essential, and at the same time thinking for hard.

Let's see how its work.

SJ 的理解模式 - 简易输入,积极思考,理解所解决的基础问题

我花了两天来理解那个paper,
接着呢, 我就讲给SJ 听。

简易输入
首先,我老想用paper里面的example, 和show 里面的proof.
但是都被他避免掉了。

他只要理解他给的simple example是怎么work的
然后呢了解了这一系列以后,就开始思考
它是怎么work的,它有什么用。


积极思考,理解paper所解决的基础问题
他思考的结果,就用了一个图形来表示。
behaviour范围增大,error范围增大。
基本就是,给予一个research result 的 input(without proof),
来思考这个result是怎么work的。
之前他看了这份paper,马上就了解了,它是一个synthesis的问题。
感觉就是他有一个核心的思考系统,也自己的思考,来判断这些paper的功用。
把paper和所知道的,做一个贯穿。

Active judging
就是所谓的active judging吧。依靠paper来理解,不自己思考,就是passive judging吧。
总结来说,基于简单的research input(就是我所说的), 来进行思考。

怪不得有一次有人说要给看thesis, 他说看paper就够了,thesis太多了。


Wednesday, May 23, 2012

Paper Writing..

I try to understand the structure of
Interface Decomposition of Service Compositions


In the introduction..it divides into
(1)Intro to two concepts
(2)Current Status
(3)Problem
(4)Solution
(5)Contribution
(6)Implication/benefits of the methods

Each section(a paragraph) about 10-12 lines, 3-4 sentences.

Phases in understand a paper

I have been reading the paer
Interface Decomposition for Service Compositions

I cannot understand certain approach in the paper, I keep reading it, but I still not able to understand it.

The main question in mind
 How it overcome the implied scenario?

(1)Email the author
The author refer me to a proof in the paper, this is the first breakthrough. But I can't understand the proof through.

(2)Use simple way(instead of the hard way) to understand it
I download the FSP, and find a implied scenario, and encode it in FSP, I gradually understand the approach in the paper.
The simple way here is: Hands on, and use the method on simple example. This almost apply in all scenarios.
The hard way here is understand it from the proof entirely.

(3)Defends it, know what exactly(a concise idea/argument) make it works.
But when discuss with a senior, who question about it's validity, I can't make an argument to defend it instantly. I keep handing on, and I more and more understand it, and make a very consice argument that why it works.





Saturday, May 12, 2012

Brain game

想看看有没有神马游戏可以帮忙脑子运动运动一下。。
发现了http://www.lumosity.com

Debugging

昨天debugging了一个software一整天,终于发现了问题。
这个bug的原因是assumption错误。


下面是给自己Note:
False Assumption: definitionRef会直接利用里面Definition的Process
Correct: denitionRef 会叫GetProcess把里面的Definition的Process重新包装再返回。
如何发现:是在VS studio对于definitionRef 的MoveOneStep 的debugging下发现的。

Thursday, May 10, 2012

Productivity with State Traversal


1. State Identifying stage:
Identify the total amount job you have to do;
2. State Reducing stage:
A. Quality
Ask yourself, can you don’t do this, and achieve the same results or similar results, if yes, then DISCARD it.
B. Environmental
Ask yourself, can you don’t do this, and people will think it is same results or similar results, if yes, then DISCARD it. (Don’t push too much effort in the area where people won’t feel it)
4. Method choosing stage: 
Identify the method for doing the job.
(Use as much resource that can be reused)
(The shortest amount of time for the similar real and environmental quality).
(And as you don’t like to open the same thing by traversing again and again, make them a shortcut, thus not only to save your time, and let you more happy when doing work)
5. Time labelling stage:
Label the deadline for each stage of the works, by the assumption of uninterrupted continuous effort puttting in. The output of the effort should be the normal output, which means that it just need your continuous concentration without needing you to push your mental muscle or body muscle.
6. State traversing stage:
Travel the state according 4 and 5. Record down the time usage in order to better estimate the time in the future.

Wednesday, May 9, 2012

解决问题


常常很喜欢为了搞懂一门知识,不停的看书;
但是对于书后面的问题,却懒于解决。

但是能够利用书上面的知识,去解决问题,是彻底搞懂一门知识的捷径。

面对这些问题,我又会想,和增进自己的研究能力又神马帮助。

在课业上:
问题:书本给的
解决方向:用书本教的
答案:书本给予,一般都有standard的answer

在研究上,
问题:自己提出的,
解决方向:先用直觉,尝试,然后就这样trial and error,
但是很大部分,可以用像 induction, proof by contradiction, maximal argument 等等来解决。
答案:自己去验证

对比一下,所以看这些书后面的问题的时候,我们应该想想,
问题:解决这样的问题有什么好处,为什么会想到这样的问题
解决方向:如果给我这个问题,没有课业内容,我会如何解决;
还有除了课本教的,还有没有别的解决方案
答案:先自己判断了正确后,再和书本对比,看看能不能学习到什么新的东西。