返回信息流The study of search spread has revealed how parameter
tuning can adjust the relative scale of the effective particle
distribution. The particle distribution in CK is not known,
but the free parameters w, φ are empirically set so that the
sampling standard deviation is very close to the informer
separation. In contrast the standard deviation of trial positions
in BB and DR3 is 60–65% of the informer separation. These
choices of scale factors might in fact be necessary to temper
possible undesirable characteristics of the dynamics.CK is
tuned close to instability and has a large scale factor—is this
because its convergence properties are too strong? DR3 is
tuned close to stagnation;12 does this indicate that it is too
divergent? BB is tuned close to collapse; does this indicate
that search is generally too diverse?
黑体字部分感觉自相矛盾啊 第一句就是CK不稳定是否因为收敛性太强了。。求高人指点[ema1]
这是一条镜像帖。来源：北邮人论坛 / ml-dm / #10209同步于 2013/3/11
ML_DM机器人发帖
谁能帮我解释一下这段话什么意思 关于PSO算法的 前后矛盾 凌乱
gin0726
2013/3/11镜像同步0 回复
订阅后，新回复会通过你的通知中心匿名送达。
0 条回复
暂无回复 · 你可以订阅本帖等待新回复。