swarch模型hanmilton1994论文的程序

hanmilton1994论文的程序，改变swarch模型的状态个个数，如将3状态改变为2状态，ns=3变为ns=2，为什么程序的运行出现问题？将3状态改变为2状态，ns=3变为ns=2时候已经将初始值改为 let th[11,1] =
           0.34775  0.25589  0.55992  0.06036  0.16047  0.051751
           0.99195  0.99907  5.7857  0.44348  3.7547 ;
出错信息显示为：Stack trace:
   matpm called from C:\gauss\ARCH\Maxseek, line 190
   echoo called from C:gauss\ARCH\Maxseek, line 227
请问如何解决，QQ396569605可详细告知。

还需要改个地方，你把ipm=3，改成ipm=1。
ipm是估计概率转移矩阵的参数，=3是用作者自己编写的那一套代码去估计，后面程序里有说明；
=1是直接估计2*2矩阵中的p11和p22
另外，我用的是gauss7.0，改了之后运行无误
         FINAL ESTIMATES

Value of log likelihood:      -2845.4269

Coefficients:
      0.33000284       0.26802494        23.460591       0.11939691       0.25590515    0.00015927405        10.503320       0.10594459       0.44324573        3.3957401        3.7547000

Constant term in regression      0.33000284
Autoregressive coefficients in regression      0.26802494
Initial variance not neeeded
Constant term in ARCH process       23.460591
Coefficients on lagged epsilon squared in ARCH process      0.11939691       0.25590515
(Transposed) matrix of transition probabilities
  2.5368221e-008     0.0089831327
      0.99999997       0.99101687

The state with no adjustment to ARCH process is state 1, with transition
probability   2.5368221e-008
Vector of variance factors for states 2 through       2.0000000       0.10594459
Coefficient on negative lagged change for asymmetric effect      0.44324573
degree of freedom for t distribution is       5.3957401

Gradient vector:
  6.9179806e-005     0.0011302750  -7.8088182e-005    -0.0020425821    -0.0018203135    0.00015770401  -7.8054892e-005     -0.014099421   -0.00073166419   -0.00026004046       0.00000000
Hessian not positive definite; eigenvalues are
      0.00000000    0.00017917889      0.024908049        1.1920800        70.196701        256.68116        352.48718        424.74669        1439.5269        1983.4422        17026.812
eigenvector associated with smallest eigenvalue is
      0.00000000       0.00000000       0.00000000       0.00000000       0.00000000       0.00000000       0.00000000       0.00000000       0.00000000       0.00000000       0.00000000        1.0000000

Matrix of Markov transition probabilities:

  2.5368221e-008     0.0089831327
      0.99999997       0.99101687

Ergodic probs for full state vector:
      0.00000000   2.2585755e-010   7.9978217e-005     0.0088231762   2.2585744e-010     0.0089031542     0.0088231762       0.97337051

Ergodic probs for primitive states:
    0.0089031547       0.99109685

Log likelihood:
      -2845.4269

等待高手！

增加样本容量怎么更改代码呢