我有一个比较复杂的对数似然函数，希望能够通过Matlab做MLE，Matlab的工具包里有一个maxlik.m的函数，函数说明是这个样子的：

% PURPOSE: minimize a log likelihood function

% ------------------------------------------------------------------

% USAGE: result = maxlike(func,b,info,varargin)

% or: result = maxlike(func,b,[],varargin) for default options

% Where: func = function to be minimized

% b = parameter vector fed to func

% info structure containing optimization options

% .delta = Increment in numerical derivs [.000001]

% .hess = Hessian method: ['dfp'], 'bfgs', 'gn', 'marq', 'sd'

% .maxit = Maximium iterations [100]

% .lambda = Minimum eigenvalue of Hessian for Marquardt [.01]

% .cond = Tolerance level for condition of Hessian [1000]

% .btol = Tolerance for convergence of parm vector [1e-4]

% .ftol = Tolerance for convergence of objective function [sqrt(eps)]

% .gtol = Tolerance for convergence of gradient [sqrt(eps)]

% .prt = Printing: 0 = None, 1 = Most, 2 = All [0]

% varargin = arguments list passed to func

% ------------------------------------------------------------------

% RETURNS: results = a structure variable with fields:

% .b = parameter value at the optimum

% .hess = numerical hessian at the optimum

% .bhist = history of b at each iteration

% .f = objective function value at the optimum

% .g = gradient at the optimum

% .dg = change in gradient

% .db = change in b parameters

% .df = change in objective function

% .iter = # of iterations taken

% .meth = 'dfp', 'bfgs', 'gn', 'marq', 'sd' (from input)

% .time = time (in seconds) needed to find solution

我想问一些其中的func是不是要传递一个函数句柄？比方说我的似然函数是L，是不是这里应该用@L?

b应该是待估参数向量吧，比方说我的待估参数有两个sigma和mu,是不是应该用[sigma,miu]?

最后那里的varargin位置怎么填呢？是不是把似然函数里边除了待估参数外的其他变量的符号用[]括起来然后填进去？

这个maxlik函数我没有用过，求各位高手一定帮帮忙啊！！！

[此贴子已经被作者于2005-12-7 17:01:41编辑过]