用LIMDEP软件中 极大似然估计法估计参数，当分布函数书对数线性形式时是顺利的，没有error，如下：
Maximize ; Labels = CONS,C_BID
        ; Start = B        ; Maxit = 100
        ; Fcn =
        BVH=LGP(-DOT[Z]-C_BID*LOG(BID2U))|
        BVL=LGP(-DOT[Z]-C_BID*LOG(BID2D))|
        BVM=LGP(-DOT[Z]-C_BID*LOG(BID1))|
        R1*R2*Log(1-BVH)
        +R1*(1-R2)*Log(BVH-BVM)
        +(1-R1)*R2*Log(BVM-BVL)
        +(1-R1)*(1-R2)*Log(BVL)
        $
Entering iterative search for function optimizers.
        Initial iterations to improve starting values
        NOTE: Convergence in initial iterations is rarely
        at a true function optimum. This may not be a
        solution (especially if initial iterations stopped).
        Exit from iterative procedure.    1 iterations completed.
        Check convergence values shown below.
        Gradient value: Tolerance= .1000D-05, current value= .3009D-06
        Function chg. : Tolerance= .0000D+00, current value= .1419D+03
        Parameters chg: Tolerance= .0000D+00, current value= .1905D-06
        Smallest abs. param. change from start value = .
        At least one parameter did not leave start value.
        Normal exit from iterations. Exit status=0.
        Exit from iterative procedure.    1 iterations completed.
        Exit status appears above.
        Exit status for this model command is   .0

但是当分布函数是线性模型F(B;θ)=Λ（α+βi *xi-β_bid*B）=1⁄(1+exp(-(α+βi* xi-β_bid*B)))时，
Maximize ; Labels = CONS,C_BID
        ; Start = B
        ; Maxit = 100
        ; Fcn =
        BVH=LGP(-DOT[Z]-C_BID*BID2U)|
        BVL=LGP(-DOT[Z]-C_BID*BID2D)|
        BVM=LGP(-DOT[Z]-C_BID*BID1)|
        R1*R2*Log(1-BVH)
        +R1*(1-R2)*Log(BVH-BVM)
        +(1-R1)*R2*Log(BVM-BVL)
        +(1-R1)*(1-R2)*Log(BVL)
        $
跑出来一篇error：
Error:   590: Obs.=     1 Cannot compute function: Logminus
Error:   137: Iterations: function not computable at crnt. trial estimate
Error:   802: Cannot compute function at start values. Exit status=4.
Function=  .70187257136D+02, at entry,  .00000000000D+00 at exit
Error:   590: Obs.=     1 Cannot compute function: Logminus
Error:   137: Iterations: function not computable at crnt. trial estimate
Error:   143: Models - estimated variance matrix of estimates is singular
Error:   447: Current estimated covariance matrix for slopes is singular.

求问这些error是为什么……新手第一次发帖，有什么不对的地方请指出来……求大牛解答T。T毕业论文很焦虑啊