The Monte Carlo Test
In computing circles, two different exploratory methods are in use, and a sharp distinction must be made between them. One is the "iterative approximation procedure." You can forget the horrible name at once: this process is the systematic searching by application of which we have just tried to find out the square root of 100.
The other method has the far prettier name of the "Monte Carlo Test." The system it uses is a complete absence of system. As in roulette at Monte Carlo, the computer tries its luck with "random numbers," selected without reason and thrown to it just to see what it will do with them.
The Monte Carlo System or, as it is sometimes called, the Random Walk method, is used to imitate real situations which are just as accidental. Electric circuits which have to work under constantly changing conditions can be checked in a Monte Carlo test, the testing values being chosen just as arbitrarily as they are in daily life. For example, the working load that a telesubscriber may ring up another) can be tested. The network is programmed into a computer and then telephone connections are made by numbers selected at random until it can be confirmed that the network is adequate in some sections, while in others it needs extensions.
It was the destiny of the electronic computer, with its unconventional methods, to secure a place in the history of science for this kind of numerical experimenting. You will see what astonishing things can be done by such methods. First of all, however, we would like to tell you in passing where the computer gets the "random numbers" it needs for its experiments. For this purpose, it has a "random numbers generator," a kind of automatic dice-throwing apparatus - a program issuing an arbitrary sequence of numbers. We shall not explain to you here how this program works, for a description would be quite involved.
©
by PhiloPhax & Lauftext
& Redaktion Lohberg
Kybernetik
- Was ist das?
First printed in Germany: 1963