Othello study is coming to an end

The deadline for the IBM program challenge is approaching. Today, I have spent some time modifying it and handed it over. Alas, I have spent too much time on it and many lessons have been skipped.

After all, this is very difficult. In the past, many Othello papers on 8x8 could not come in handy, because IBM's questions may not be 8x8, but those theories, for example, a lot of starting points, pattern, and parity won't work. On the contrary, the simplest theories are: stability, mobility, and potential mobility. I found some programs and changed them. The results were not satisfactory. After all, the time is not too long, and it cannot be used for self-learning like others, because it will take a long time. Besides, the fixed 8*8 self-learning is also acceptable, once the rules of the competition change, the previous training is in vain, because we can only adopt some simple strategies, but we do not know if this is the case. I really hope I can squeeze into the top 20. However, in any case, I think this is very interesting and I will continue to study it later.

That's it. It's just a bit over. To sum it up, the program uses the most basic alpha-beta pruning and uses a valuation based on industry and potential industry dynamics, finally, the number of sub-numbers is searched. In addition, in terms of creativity, the Valuation Functions of two different strong Othello programs are combined to achieve good results. The proportional parameters are manually tried (very inconvenient, but there is no way); it is worth mentioning that a special process is made for a place that cannot be downloaded. I call it stone, and no one can get down at Stone, but the child around stone also seems to have a certain degree of "stability", because this stone reduces the possibility of turning it over, and we hope that our final pieces will all gather around stone. Let's think about it, if our child fills up spaces around stone, the number is very impressive. If there are 10 stone children, the number may reach 40 (probably ), therefore, this consideration is also added to the evaluation function; however, it may not be overlooked, that is, although the stone child reduces the probability of subpopulation turning around, however, it also reduces the probability that the peripheral sub-nodes may connect to other sub-nodes. Therefore, this should be taken into consideration when you set the score on the checkerboard. Also, to speed up the program running, the program uses a one-dimensional array to represent the checkerboard and only converts it at the beginning, and change the corresponding place each time when the next sub-function (changes a position), so as to avoid repeating every time the sub-function enters the next sub-function, however, this operation may not use 8x8 (8x8 can be represented by an integer, and the single and ten digits can represent the X and Y coordinates respectively, therefore, a lot of % and/operations are used, which increases the time overhead, because the time saving is not very obvious, but because of the use of a one-dimensional array, therefore, when writing some functions, it is relatively simple (such as seeking the dynamic power and determining whether it can be used ).

The above are some of my own attempts. Many of them are from the perspective of the website, and I have also referred to some very good foreign papers.

I have to thank those who are very good in my eyes and I admire them because they are making bold attempts in their own practical spirit, I am deeply inspired by the following ideas:

Buro, an American scientist, by logistello;

Nowcan should be able to search for black and white games on the Internet to link to his blog. His ideas are unique, his research spirit is admirable, and his program level is among the best in China;

Blacwet, also known as Li Shun, has English names: devil_li, dedevil-li, blacwet, livededevil, etc. After reading his self-introduction, he turned out to be my hometown and was very envious because he had been very familiar with computers for a long time, have their own achievements, and this is what I yearn. In China, people who study black and white games will mention him and his website. Indeed, he brings everyone into the Hall of Othello programming with his profound analysis and programming skills, to enlighten the elders. You can learn a lot here !!!

The author of the sad black and white games can hardly find the introduction of this person on the Internet. However, no one in China can surpass his black and white game program "sad black and white Games", and the source code is nowhere to download, it is really sad ".

In addition to the above, there are many people I don't remember, but some of them should be my masters, some are the "advanced" who have already explored this aspect, and some are the "fellow travelers" who are working on this and inspire me to keep moving forward. Thank you here. More importantly, it is hoped that the spirit of talented programmers will be nurtured to inspire their programming potential and catch up with the pace of genius.

Higer

In Zhongguancun, Beijing