I am trying to find the root of the following function. Basically, I am
trying to find n1, n2, prob1, and prob2 from a mixture of two negative
binomials.. I differentiated the log likelihood with respect to n and am
trying to solve this equation. zij and y are both a column of vectors. y is
the data and zij are the conditional probability of being in a particular
distribution. I tried uniroot but it failed ..is there anything in R that
can do this?? The problem is I generated the data I am using from a mixture
of two negative binomials.
f=function(n) { sum((1-zij)*(digamma(y+n)-digamma(n))) }
uniroot(f, c(1,100))$root
Error in uniroot(f, c(1, 100)) :
f() values at end points not of opposite sign
[[alternative HTML version deleted]]