Is 1/x the best method to maximize values? When I have a 0 value, it becomes infinite. To compensate, I am adding 0.01 to the values before 1/x, however, I know this is not ideal in every case (and I’ve added a small conditional to only add 0.01 when the value is 0). Is there another workaround? One other thing I noticed is if I increase the delta to 0.0001 for example, the Wallacei charts become illegible as the graph axis reflects a large value:
Any advice would be greatly appreciated!
Yes, based on our experience 1/x is more effective than inputting a negative value. Adding a small constant value to x before 1/x is a common approach to prevent 1/0. If you do so, I recommend doing it for all X values and do not use conditional statements as in that case, that small value won’t be considered as a true constant variable in the simulation. Another way of dealing with these scenarios is to take advantage of the Null indicator. check out this video here.. So basically you can use the null indicator to teach Wallacei that 1/0 is a bad solution and not to consider it in the evolution.
And regarding the graphs you posted, I am not quite sure if I understood your question clearly but the bottom graph refers to a scenario that from the early generations to the last generations, variation decreased significantly, (aka convergence). That is why you see a line like this, this is actually due to the fat and flat early SD graphs and thin late SD graphs. And since they are all plotted on the same chart (to understand the trend) this will happen.