Fidelis Assis writes: > > talking of EDDC, I'm running into a bit of wierdness with the equation > > published in http://osbf-lua.luaforge.net/papers/osbf-eddc.pdf -- > > specifically the final CF(F) equation on page 5. Using the K1, K2 and K3 > > values suggested, I cannot reproduce the values in the graph below it. > > > > According to the graph, what I should see are values like this: > > > > Dfs Dfh CF > > 100 100 0 > > 0 100 1 > > 100 0 1 > > Yes, that's the "ideal" curve... good for explaining and visualizing the > effects of the CF. > > > what I actually get are: > > > > Dfs Dfh CF > > 100 100 1.19134681363401e-07 > > 0 100 0.124843632959153 > > 100 0 0.124843632959153 > > But when it comes to practice we need some adjustments... the second > term, ($WdotSumf / (1.0 + $K3 * $WdotSumf)), limits the max value to > approx. 1/$K3 for large $WdotSumf, that is 1/8 (0.125) in this case. > > You can get a good approximation of the ideal curve by setting $K3 to 1: > > s=100 h=0 w=3125 cf=0.999994300021052 > s=100 h=100 w=3125 cf=9.53671598439468e-07 > s=0 h=100 w=3125 cf=0.999994300021052 > > $K3 = 1 was my first attempt but experiments showed that higher values > produce better accuracy, with a maximum around 8.
aha, I get it. That makes sense. Thanks for the explanation. So using K3=8 limits the maximum CF(F) to ~0.125? Doesn't that restrict your probabilities to the [0.5-.125, 0.5+.125] range? Is that desirable? I would have assumed we'd still want to allow really "strong" tokens to reach nearly 0 or nearly 1.0. --j.
