Today I decided to do something with konquest, which is a free Linux game. It isn't very complex but some people find it very addicting. My personal interest is to see if I can make it 3D and use blender to implement the game. I personally would like to be able to see ships that are in route to attack and their headings.
I will make a version that is a blend file and stay true to the history of the game by adding the ability to see ships moving as an option with perhaps some degree of fog of war effect that only allows you to see within a certain radius of a planet that you control. Since blender is available for Windows, it would be possible to run the game on a Windows machine, and I did test. Blender seems to run fine on Windows.
I start with the command line action:
apt-get source konquest
This gives me all the code for the kde games in a packed format. I unpacked it to a directory and took a look. It is pretty much as I suspected and seems to be well structured. The basic logic of the game seems to be that you select ships to attack a planet and if you dominate then you expand. It isn't likely to be more than that, I would assume, with some random number inclusion to make it interesting. The part where I would expect some complexity is in the target selection by the "bot" players and how they select which planet to attack and with what force and how they determine what level of defense must be kept to protect from the other "bots" and live players. I suspect it will take me several hours to create it, and another day to let it sink in so I can make some changes and polish it. The point is to have something that is reasonably interesting to do as I consider a mathematical technique for the TSP which I suspect might work, but I am not certain. In the process of creating the simulation, there are aspects of TSP ( Traveling Salesman Problem ) when plotting a strategy that would determine the path of most effective attack in a sequence and not just the potential of win on a single planet which is closest.
I intend to use a type of genetic algorithm to approach the TSP in a new way and also consider aspects of an optically based DNA computer that functions as content addressable memory and how that would be applied in the solution of a TSP. I may make a USB connection to a CAM to determine if the entire system functions as I suspect, but that aspect would be Linux only and though I intend the game to be completely compatible with blender, the extension is only designed for USB on Linux. The cost and time of doing a device driver dll for Windows is prohibitive and when I have contracted them in the past they were quoted to me a 50K$ up front and this was 1993, so I would guess that a Windows device implementation could run as high as 250k$. So, that is not even reasonable to consider. If somebody wants to know how effective CAM is in solving problems they can set up a Linux system.
I have an inkling of a new mathematical technique that has been escaping me and perhaps if I play at this for a while it will come to me. There seems to be a better way to solve multiple simultaneous equations that are interactive and recursive and many of these concepts are inter twined. It might seem completely unrelated to what I am developing, however it is actually precisely on target. In the genome there are many genes active at the same time and they act upon other genes and moderate systems and pathways. It is a similar situation. The action of many "bots" interact with specific targets and in the complexity of it become emergent or complex or merge to a stable attractor.
The complexity of a "simple" biological system can be 10,000 factorial complex and since 70 factorial is so complex that it exceeds the number of atoms in the known universe, it is vastly more complex that can typically be expressed in any common relationship.
0 comments:
Post a Comment