摘要
在用Taylor方法研究种群空间格局时,拟合Taylor幂法则V=am^b通常采用将原函数线性化为lnV=Ina+blnm,然后用线性最小二乘法拟合,求得参数a,b。这种方法不是最优的,本文提出用麦夸方法(Marquardt's methos)最优拟合Taylor幂法则。实例验证也表明麦夸方法优于通常采用的“对数线性化——最小二乘法”,采用前者的拟合残差远小于采用后者时的拟合残差。
When assessing and interpreting population spatial patterns with Taylor's Power Law model,transforming the power function to provide a mathematically equivalent linear equation is a usual method for fitting Taylor's Power Law curve.However,it is not optimum fitting to the untransformed function,i.e.,Taylor's Power Law model itself.In this paper,Marquardt's optimum algorithm is introduced for determing the values of the parameters (a & b)of Taylor's Power Law model,which will give the minimum least-square fit to a set of data.It is proved to be superior to the usual method with examples.
出处
《北京林业大学学报》
CAS
CSCD
北大核心
1990年第1期49-54,共6页
Journal of Beijing Forestry University
关键词
麦夸法
最优拟合
Taylor幂法
害虫
Marquardt's algorithm,optimum fitting,Taylor's Power Law model
