Maximum likelihood estimator bug

[Alex Stead]

Hi,

I'm trying to make a simple maximum likelihood estimator. This is something I've done several times before, but this time I'm getting an error and I can't figure out why. I'm using Stata/MP 16.0.

In this case I'm trying to estimate a multivariate normal model. Here's the code:

gen c1= invnorm(uniform())
gen c2= invnorm(uniform())
gen c3= invnorm(uniform())

capture program drop mvn_ll
program mvn_ll
        version 16.0
        args todo b lnf
        tempvar b1x b2x b3x lnsigma1 lnsigma2 lnsigma3
        mleval `b1x'        =     `b',        eq(1)
        mleval `b2x'         =     `b',        eq(2)
        mleval `b3x'         =     `b',        eq(3)
        mleval `lnsigma1'    =    `b',        eq(4)        scalar
        mleval `lnsigma2'    =    `b',        eq(5)        scalar
        mleval `lnsigma3'    =    `b',        eq(6)        scalar
        tempname Y MU SIGMA
        matrix `Y' = ($ML_y1 \ $ML_y2 \ $ML_y3)
        matrix `MU' = (`b1x'\ `b2x' \ `b3x')
        matrix `SIGMA' = (exp(`lnsigma1')^2, 0, 0 \ 0,exp(`lnsigma2')^2, 0 \ 0, 0, exp(`lnsigma3')^2)
        mlsum `lnf' = lnmvnormalden(`MU',`SIGMA',`Y')
end

ml model lf mvn_ll (c1=) (c2=) (c3=) (lnsigma1:) (lnsigma2:) (lnsigma3:)

ml max

When I run ml check, I'm getting the following message

Test 1:  Calling mvn_ll to check if it computes log likelihood and
         does not alter coefficient vector...
         FAILED; mvn_ll returned error 111.

Here is a trace of its execution:
------------------------------------------------------------------------------
-> mvn_ll __00000I __00000J __00000K __00000L __00000M __00000N __00000O
        - `begin'
        = capture noisily version 16: mvn_ll __00000I __00000J __00000K __00000L __00000M __00000N __00
> 000O
          ---------------------------------------------------------------------------- begin mvn_ll ---
          - version 16.0
          - args todo b lnf
          - tempvar b1x b2x b3x lnsigma1 lnsigma2 lnsigma3
          - mleval `b1x' = `b', eq(1)
          = mleval __00000P = __00000J, eq(1)
vector __00000J not found
          ------------------------------------------------------------------------------ end mvn_ll ---
        - `end'
        = set trace off
------------------------------------------------------------------------------
Fix mvn_ll.
r(111);

And I can't figure out why that is. It's been a while since I've done this kind of thing, so I may be forgetting something basic. Can anyone shed some light on this?

[Jeff Pitblado (StataCorp)]

You specified ml model lf mvn_ll but the program definition for mvn_ll is for a d0 evaluator instead of an lf evaluator.