Closed jeremy-pereira closed 2 years ago
That looks like a typo or a lingering remnant of some previous design change. Eventually, mutations should overwhelm the effects of this starting bias, but you're right, this does look like biased starting conditions. I'll fix that. Thanks for finding this.
Edit: fixed in main branch.
I'm looking to create my own evolution simulation and this code base seemed like an excellent place to start, so I was perusing the code and I came across the gene declaration and the
makeRandomWeight
code.Why
0xefff
? Are you deliberately trying to create weights that are biased towards the negative? Or is it a typo for0xffff
?