Impose constraint on free parameter from ml estimation

[Henry Strawforrd]

I am using the constraint

constraint define 1 delta < 1

(or also "/delta" does not work)

which does not work because delta is free parameter in the ml estimation

ml model gf0 nl_tobit /gamma1 /lambda /gamma0 /delta /sigma, constraints(1)

How to reference such a parameter in the constraint?

[Maarten Buis]

https://www.stata.com/support/faqs/s...l-constraints/

[Henry Strawforrd]

Hm, I did not get it from this link.

[FernandoRios]

the command `constrain` does not handle nonlinear constrains. To impose that (if you really think they should be imposed) you need to use transformations

Example:
- Sigma needs to be positive. So instead of estimating sigma, you estimate logSigma. Exp(logsigma) will always be positive
- p needs to be between 0 and 1. Then you need to estimate xb, and use the transformation exp(xb)/(1+exp(xb)) . The transformation will be between 0 and 1.

For your case, perhaps:
you want delta to be less than 1
then estimate d_1 where delta = 1-exp(d_1)

F