[Limdep Nlogit List] errors about mixed logit

[Jiang, W.]

Hi everyone, i have a question that when i am doing mixed legit model, i cannot get the result from coding:

RPLOGIT
;Lhs= Mobility
;Choices= move
;Rhs= a,b,c
;Pts =100
;Halton
;Fcn=a(n,0.5)
$

The results show that: Error    803: Hessian is not positive definite at start values.
Error   1027: Models - estimated variance matrix of estimates is singular

Because i used revealed preferences data rather than having experimental design, so i don't have utility for both choices, only for one. But i do have a binary choice in the end. is there any suggestions about the error? thank you!

Best
Wen