From: yan zhong on 6 Apr 2010 15:47 Hi, I try to estimate the parameters of a short-term interest rate model (with garch & jump in the process) by using fmincon to optimize my log-likelihood. The total parameters need to estimate is quite large (9 in total). There is also one linear inequality constraint for garch parameters. I supplied gradient and use BFGS updating scheme for Hessian, the used options as follows: options = optimset('LargeScale' , 'on'); options = optimset(options,'Algorithm','interior-point'); options = optimset(options,'AlwaysHonorConstraints','none'); options = optimset(options,'DerivativeCheck','on'); options = optimset(options,'Display','iter'); options = optimset(options,'FunValCheck','on'); options = optimset(options,'GradObj','on'); options = optimset(options,'Hessian','bfgs'); options = optimset(options,'MaxFunEvals' , 1e+7); options = optimset(options,'MaxIter' , 1e+7); options = optimset(options,'TolFun',1e-6); However, the error msg came back as: ??? Error using ==> linear_drift_level_garch_JumpObj Too many input arguments. Error in ==> optimfcnchk>checkfun at 355 [f,g] = userfcn(x,varargin{:}); Error in ==> fmincon at 526 [initVals.f,initVals.g(:)] = funfcn{3}(X,varargin{:}); Caused by: Failure in initial user-supplied objective function evaluation. FMINCON cannot continue. I check my objective function and gradient, they look ok. I do not know why cause such problem. Any help will be truly approciated! Yan
From: Steven Lord on 6 Apr 2010 17:48 "yan zhong" <yanp.zhong(a)gmail.com> wrote in message news:hpg347$807$1(a)fred.mathworks.com... > Hi, > I try to estimate the parameters of a short-term interest rate model (with > garch & jump in the process) by using fmincon to optimize my > log-likelihood. The total parameters need to estimate is quite large (9 in > total). There is also one linear inequality constraint for garch > parameters. I supplied gradient and use BFGS updating scheme for Hessian, > the used options as follows: *snip* > However, the error msg came back as: > ??? Error using ==> linear_drift_level_garch_JumpObj > Too many input arguments. You showed the group a bunch of code -- except for the objective function signature and how you call FMINCON using your objective function, which are the two most important pieces of code! -- Steve Lord slord(a)mathworks.com comp.soft-sys.matlab (CSSM) FAQ: http://matlabwiki.mathworks.com/MATLAB_FAQ
|
Pages: 1 Prev: fopen Next: How do i save rgb2ind image so that ican use it in otherfunction? |