7th weeks notes

I really really do not want to write last weeks notes.
Actually, I almost do nothing about my project.

Prepared for the English exam, and the built the environment of matlab and lamp~~

6th notes

This week I rebuild the whole program. It is including: in the new project, I used the array to store the position(as map) and store the link and the cost in one arraylist(the cost at the first position)， and the most important, I make all the parameters can be easy to translate, that means: if i want to use some other functions, or some other operators or even use some other methods, it won't need to rebuild that much. For example, I haven't do that much yet, but my project already has some empty method for asymmetric TSP.

And this week, I do some testing. The result seems so strange, I testing with eil51, I get the result even better than the one on the website. And the result of att48, seem 200% over the one on the website. It seems so strange~~ I need more test with all this. NEXT WEEK.

And what's more, I got a really funny things.

Many other students have nearly one hour's face to face talk with the supervisor, a lot of knowledge introduction, a lot of experience introduction. But I do not have, before any discussion, I need send email first. And each talk won't be more than 15 minutes. Professor only told me, what's wrong with my project, and something like what may need to do to improve. I accept this. I think this is a "research". I should be like this.

But, when I ask him, could you recommend some paper I can read. He said: if you type the key word with "Evolution Algorithm with TSP", you can find a lot.
I feel errrrrrrrrrrrrrrr...... So funny~~~ I am trying to make myself trust that I am a phD student~~~ hah~~~~


Ps: I found a strange thing is that: all the best solution of mine is end by the first city. It is because of my coding problem?

5th Weeks notes

I feel really unhappy with this week's processing.

I spend the whole week on the proposal and the presentation.

But actually, I do not do a good job on this.

I do not pose on all of my proposal, actually, I totally do not agree with what I write.

And only one thing of my improvement IS "I think I indeed cannot do some about dynamic environment". It is too difficult and I do not have that much time to solve a problem like that. And, a tsp is enough.

But anyway, I think next week will be a great one !! I will do some strong things!!

4th weeks notes

In this week.
I finished a basic programing of GA to solve the problem of TSP, in two days. But unfortunately, this in not good enough.

I make a lot of change from the traditional TSP programming. Such as the choosing from fitness, and the group size of populations(normally is only one).

But actually, I got a really good result, only 2% over the optimazation result. (Make the group size as more than 1 is really useful to get a better result, to get away from the local optimal solution.)

The last week's problem has been solved, just because the population is too small and the wrong calculation about the fitness.

I need do some change about this program, and do some further work about it. And what's more, I need a proposal of my whole project.

Third week notes

In this week, I finished reading the book "How to solve it "
And had a talk with professor. I hope to do 3 aspect.


1. Try to solve problems by GA with some local optimal searching function together, to make the whole searching have a better efficiency.Maybe in the page 14,solving the function G2, GA combine with hill climbing is not a bad idea. I am not quite sure is there anything about this area I can do.

2. About using evolutionary algorithm to solve constrained problems, just as you said in the "9.3 summary"'s first paragraph "Nevertheless, we have to admit that the characteristics that make a constrained problem difficult for an evolutionary algorithm aren't clear." Do some research and study in this area, I think it will be quite challenge and difficult and valuable. I also a little bit wonder whether I am strong enough to do this. After all, I have to finish it in one year, not too much time left for me to get as deep as a PHD do.

3. Something about dynamic environment(and noise). In my opinion, if cannot deal with dynamic, a theory cannot be admitted as absolutely complete and prefect. After all, our environment is changing every second.

(Copy from the email.)

Professor hoped me to do something about discrete or continuous programming(on Friday).
And I begin try to implement TSP by java. It seems not that much easy.
Till Sunday, I only implement some basic function of it.


(I write this on Wed morning, and till Tuesday 4PM, I finished programing, and till Wed. 2AM, I still cannot find out the result is 400% over best)

Second week notes

In this week, on Tuesday, I have a talk wih professor Zbigniew Michalewicz. And finally, I could be supervised by him!

And at the rest of this week, I keep reading a book called "How to solve it - modern heuristics", trying to find out some more details about algorithm, and find out something I really interested.

After this week, I read more than half of it, but I think everything is interesting. I may need some time to make a choice. And I am not quite sure whether I can do something about GA combine with other local optimal algorithm.

And I am trying to implement TSP solution by GA. In this weekend, I write some about how to create a link list randomly which connecting each city and distance.

It may be a job for me to think about how to implement it by GA.

First week notes

From the second semester of 2009, I begin my Master Research Project.
In the first week, I send a lot emails and read a lot of information about some professor's. Finally, there are three supervisor I hope to choose.

1.Professor Zbigniew Michalewicz. Do something about GA and Evolutionary Computation. This is my first choice and most wish to get.
2.Professor Hong Shen. If I follow him, I can do a lot of tough work I want. I can publish a really really valuable final paper at last.
3.Dr. Anthony Dick. Do something about Computer Vision. it seems about SIRF. I don't know what it is at all. This is the last choice. But I do read it a lot.

Before this semester, I do a lot of homework about GA, although I did not do any implement. I extrodinarally hope to get something about GA, and then KEEP ON STUDY TO GET A PHD DEGREE.