Why does my custom tobit not converge?

[Henry Strawforrd]

I am learning ml and wrote a simple tobit following the likelihood formula from here: https://jdemeritt.weebly.com/uploads...764/tobit1.pdf

Sorry, wrong title! It does converge but give different parameter estimates

Here is an MWE

// Make data

set seed 12553
clear
set obs 10000
gen y = runiformint(0,10)
gen x = runiformint(0,1)
gen z = abs(rnormal() * 50)


// Lower censoring

gen Di = (y!=0)

cap program drop mytobit_gf0
program mytobit_gf0
    args todo b lnfj
    tempvar xb sigma non_censored censored
    
    qui gen double `xb' = (`b'[1,1]*x + `b'[1,2]*z + `b'[1,3])
    qui gen double `sigma' = `b'[1,4]
    
    qui gen double `non_censored' = Di * (-ln(`sigma') + ln(normalden((y-`xb')/`sigma')))
    qui gen double `censored' = (1-Di) * (1 - normal(`xb'/`sigma'))
    
    qui replace `lnfj' = `non_censored' + `censored'                    
end

ml model gf0 mytobit_gf0 (y = x z) /sigma        
ml check    
ml search                    
ml maximize, difficult

tobit y x z, ll(0)

[Henry Strawforrd]

It does actually give the same results after fixing a typo!