和其他的变量回归时挺好的，到了lncba、lnmp两个变量就会出现could not calculate numerical derivatives -- flat or discontinuous region encountered。求懂的人解答

1----------------------------------------------------
mgarch dcc (lnnhci lncba),arch(1) garch(1)

Calculating starting values....

Optimizing log likelihood

(setting technique to bhhh)
Iteration 0:   log likelihood = -69.743524  
Iteration 1:   log likelihood = -65.659759  
Iteration 2:   log likelihood = -60.970076  
Iteration 3:   log likelihood = -56.083538  
Iteration 4:   log likelihood = -54.045829  
Iteration 5:   log likelihood = -53.915863  
numerical derivatives are approximate
flat or discontinuous region encountered
Iteration 6:   log likelihood =  -50.21448  (not concave)
could not calculate numerical derivatives -- flat or discontinuous region encountered
could not calculate numerical derivatives -- flat or discontinuous region encountered


2----------------------------------------------------
mgarch dcc (lnnhci lnmp),arch(1) garch(1)

Calculating starting values....

Optimizing log likelihood

(setting technique to bhhh)
Iteration 0:   log likelihood = -855.42985  
Iteration 1:   log likelihood = -843.16432  
Iteration 2:   log likelihood = -836.58961  
Iteration 3:   log likelihood = -826.86734  
Iteration 4:   log likelihood = -824.59888  
Iteration 5:   log likelihood = -824.59825  (backed up)
Iteration 6:   log likelihood = -824.58636  (backed up)
numerical derivatives are approximate
flat or discontinuous region encountered
Iteration 7:   log likelihood = -824.58631  (not concave)
could not calculate numerical derivatives -- flat or discontinuous region encountered
could not calculate numerical derivatives -- flat or discontinuous region encountered
r(430);