Re: [Gretl-users] Missing values in MLE

 The idea is to force certain quantities to be positive so that lambda   cannot be negative, eg:    <hansl>  series y = xDepVars[yIdx]  series x = xDepVars[xIdx]  scalar m = mean(y) # assume this is positive   matrix b = zeros(3,1)    mle ll = ?-ln(lambda) - xDepVars[yIdx]/lambda  series lambda = m  scalar c_  = exp(b[1])  scalar rng = exp(b[2])  scalar err = exp(b[3])  series lambda = c_ + rng * x + err * lambda(-1)  params b  end mle --robust    matrix c = exp(b)  matrix V = $vcv .* (c*c') # delta method  matrix s = sqrt(diag(V))    matrix cs = c ~ s  string parnames = "C_,rng,err"    modelprint cs parnames  </hansl>?