Table 1. Performance of Fit∂a∂i on simulated data sets.

Demography	nchr	α (shape)	β (scale)	0 ≤ |s| < 10−5	10−5 ≤ |s| < 10−4	10−4 ≤ |s| < 10−3	10−3 ≤ |s| < 10−2	10−2 ≤ |s|
Truea	—	0.184	3266b	0.182	0.096	0.146	0.219	0.357
Constant size	2596	0.180 ± 0.010	3712.2 ± 980.2	0.186 ± 0.009	0.095 ± 0.002	0.144 ± 0.006	0.213 ± 0.013	0.363 ± 0.016
	24	0.185 ± 0.028	3613.1 ± 4196.7	0.182 ± 0.017	0.097 ± 0.009	0.148 ± 0.023	0.221 ± 0.043	0.353 ± 0.060
Two-fold expansion	2596	0.191 ± 0.007	2606.0 ± 410.7	0.178 ± 0.008	0.098 ± 0.002	0.152 ± 0.004	0.230 ± 0.009	0.341 ± 0.010
	24	0.187 ± 0.023	3259.8 ± 2612.6	0.181 ± 0.016	0.097 ± 0.008	0.149 ± 0.019	0.223 ± 0.036	0.350 ± 0.050
LuCamp	2596	0.182 ± 0.008	3411.9 ± 558.5	0.184 ± 0.008	0.096 ± 0.001	0.145 ± 0.004	0.216 ± 0.008	0.358 ± 0.009
	24	0.186 ± 0.027	3435.4 ± 3249.9	0.182 ± 0.017	0.097 ± 0.009	0.148 ± 0.022	0.222 ± 0.042	0.351 ± 0.060

95% intervals were estimated as ± 1.96 SD of 100 replicates in each simulation set. chr, chromosome.

^aValues show the true DFE used to simulate the data.

^bHere the simulation was scaled to an ancestral population size of N = 10,085 diploids.