On Sun, Aug 22, 2010 at 7:30 AM, Gabor Grothendieck <ggrothendi...@gmail.com> wrote: > In ?nls.control it says: > > warnOnly a logical specifying whether nls() should return instead of > signalling an error in the case of termination before convergence. > Termination before convergence happens upon completion of maxiter > iterations, in the case of a singular gradient, and in the case that > the step-size factor is reduced below minFactor. > > however, > > 1. here we have a singular gradient but it gave an error rather than > the expected behavior of returning: > >> set.seed(1) >> y <- exp(x) + rnorm(10) >> >> nls(y ~ a * exp(b * x), start = c(a = 0, b = 0), > + control = nls.control(warnOnly = TRUE)) > Error in nlsModel(formula, mf, start, wts) : > singular gradient matrix at initial parameter estimates > > 2. and here we have a failure in numericDeriv and although that is not > documented as one of the conditions I think it would be desirable that > it be included as something that does not result in an error when > warnOnly = TRUE: > >> nls(y ~ a * exp(b * x), start = c(a = 10, b = 0), > + control = nls.control(warnOnly = TRUE)) > Error in numericDeriv(form[[3L]], names(ind), env) : > Missing value or an infinity produced when evaluating the model >
In the above: x <- 1:10 ______________________________________________ R-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-devel
