I suggest you rewrite the equation as
c * sum_i( p(i)^a ) - sum_i( ( p(i)^a ) * b(i) = 0
and use Newton's method or other such to solve it.
For there to be a solution, c has to be in the interval
spanned by the b(i).
--
This address is for information only. I do not claim that these views
are those of the Statistics Department or of Purdue University.
Herman Rubin, Department of Statistics, Purdue University
hru...@stat.purdue.edu Phone:
(765)494-6054 FAX:
(765)494-0558