Gerhard Fuernkranz wrote:
After studying the stoc1.c code briefly and reading the Purgathofer paper, I'm actually wondering, where the values in rprop are coming from? Did you determine them empirically?
Yes. In earlier work I was using a higher level of randomness, and the rprop array allowed tuning this. In the current code, the randomness is turned down a fair way, so the influence of rprop is very much smaller. Having JOINT on increases the randomness.
One of the issues is that I am currently judging the look of the screen by it's appearance on a display. Often in practice the look is different when printed out on a particular device, and some more tuning may be desirable at that point.