Dear friends. I am using function constrOptim(c(0.5,0.3,0.5), fit.error, fit.error.grr, ui=-1*ui,ci=-1*ci)
and I am confronted with error message "initial value not feasible" I plug in the initial value of (0.5,0.3,0.5) to function fit.error and fit.error.grr and have pretty reasonable result. I inequality "ui %*% theta - ci >= 0" as suggested in the R manual and it is satisfied. In case that this is a typo of the manual, I let ui=-ui and ci=-ci and try constrOptim again but the same warning message pops up. Could you please point me a way out of this? I am actually trying to translate a fortran code to R and the function I want to replace is DBCPOL, which used 'the complex method" described in Nelder and Mead (1965) and Gill et al. (1981). I believe contrOptim is better than it because it is newer, is it? Best Wishes Yuchen Luo [[alternative HTML version deleted]] ______________________________________________ R-help@stat.math.ethz.ch mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.