I'm sorry about the confusion about the "factor of 1,000". The fitness formula produces values less than or equal to 1.0. For publication figures, I have multiplied by 1000 (I don't multiply any other parameters by 1000). In Figure 3, the value of p_I is correctly given as 10, but the initial F_I used in the program is actually 10/1000. I haven't used a non-null initial population in a long time, so I forget a lot of that.
In Figure 3, the initial values of prediction, error, and fitness are shown as 10.0, 0.0, and 10.0. Have these been multiplied by 1000 as the fitness was in Figures 2 and 8?