There are several analysis functions with Fritz 12 and the one that intrigued me was the “Monte Carlo” feature. It does not work with my setup and I get a message that my engines to not support this feature. Out of curiosity, I looked it up to see what I was missing.
In the financial world you often can’t make precise judgments of situations so the Monte Carlo method is used to get a ‘feel’ for different circumstances. In fact it can be used for any complex situation where precise analysis is impossible. Apparently this is what this feature attempts to do.
The Monte Carlo feature allows you to create position statistics. It doesn’t work for me because you need one of the Rybka engines. There are a couple on the old computer but before I locate them and transfer them to the laptop, I wanted to find out exactly what I’ve been missing, is it worth the trouble and do I need it.
As I understand it, with this function the engine plays a whole lot of blitz games against itself from a given position until you tell it to quit and then it comes up with a statistical evaluation of the position. Apparently what the results tell you are which moves are likely to give the best odds.
The Monte Carlo system does not work well in calculating exact variations (like tactics) and is most useful in examining positions where positional judgment in the key factor. One person gave as an example: Suppose you are trying to decide whether to trade or push a P. If there is no tactical result in the next several moves, the engine will have trouble deciding whether to trade or push. With the Monte Carlo system it may be that pushing leads to a position where little progress is possible. On the other hand, it may show trading will allow one side to gain an advantage. You can’t play an entire game using this system. Instead, you have to know when it’s a good position in which to activate the feature.
I can see where this feature would be valuable to a GM doing research or to a world class CC player who is spending days, or even weeks, on analyzing a move, but for me it’s more information than I need…or want. A Simple Explanation of the Monte Carlo Method in Physics
No comments:
Post a Comment