I could solve this problem by setting the boundary options as follows. par1Bnd<-c(0.001,0.999) par2Bnd<-c(-10,10) DomainMat<-matrix(c(par1Bnd,par2Bnd), nrow = 2, ncol=2, byrow=TRUE) result<-genoud(Qmin, nvars=2, starting.values=c(0.5,0), Domains=DomainMat, control=list(ndeps=c(1e-7,1e-2)), boundary.enforcement=2)
To understand how to set up each option, you can refer to the links: http://sekhon.berkeley.edu/rgenoud/genoud.html http://sekhon.berkeley.edu/rgenoud/ In the above setting, "Domains" decides the boundary, and "starting.values" indicates the initial values of parameters. "control" is the list of control parameters in "optim" (For details, refer to http://sekhon.berkeley.edu/stats/html/optim.html). "boundary.enforcement" prevents the algorithm from going away from the boundary. On 13 October 2010 15:01, Wonsang You [via R] < ml-node+2993619-1186574920-137...@n4.nabble.com<ml-node%2b2993619-1186574920-137...@n4.nabble.com> > wrote: > I have to make correction in my error message which I introduced in my > original message. Sorry for my mistake. > Finally, I had the following error message after running the function > 'genoud'. > > Error in optim(foo.vals, fn = fn1, gr = gr1, method = optim.method, control > = control) : > non-finite finite-difference value [1] > > When I execute 'traceback()' to trace where the error occured, I got the > following results. Unfortunately, I could not figure out what was the > problem from the above information. > > 6: optim(foo.vals, fn = fn1, gr = gr1, method = optim.method, control = > control) > 5: function (foo.vals) > { > ret <- optim(foo.vals, fn = fn1, gr = gr1, method = optim.method, > control = control) > return(c(ret$value, ret$par)) > }(c(0.220878697173384, -13.3643173824871)) > 4: .Call("rgenoud", as.function(fn1), new.env(), as.integer(nvars), > as.integer(pop.size), as.integer(max.generations), > as.integer(wait.generations), > as.integer(nStartingValues), as.real(starting.values), as.vector(P), > > as.matrix(Domains), as.integer(max), as.integer(gradient.check), > as.integer(boundary.enforcement), as.double(solution.tolerance), > as.integer(BFGS), as.integer(data.type.int), > as.integer(provide.seeds), > as.integer(unif.seed), as.integer(int.seed), > as.integer(print.level), > as.integer(share.type), as.integer(instance.number), > as.integer(MemoryMatrix), > as.integer(debug), as.character(output.path), > as.integer(output.type), > as.character(project.path), as.integer(hard.generation.limit), > as.function(genoud.optim.wrapper101), as.integer(lexical), > as.function(fnLexicalSort), as.function(fnMemoryMatrixEvaluate), > as.integer(UserGradient), as.function(gr1func), as.real(P9mix), > as.integer(BFGSburnin), as.integer(transform), PACKAGE = "rgenoud") > 3: genoud(Qmin, nvars = 2, starting.values = InitVal, max.generations = 10, > > wait.generations = 3, n = n, yper = yper, pertype = pertype) at > wFGN.R#75 > Wonsang You > Leibniz Institute for Neurobiology > > > ------------------------------ > View message @ > http://r.789695.n4.nabble.com/How-to-fix-error-in-the-package-rgenoud-tp2993489p2993619.html > To unsubscribe from How to fix error in the package 'rgenoud', click > here<http://r.789695.n4.nabble.com/template/TplServlet.jtp?tpl=unsubscribe_by_code&node=2993489&code=eW91QGlmbi1tYWdkZWJ1cmcuZGV8Mjk5MzQ4OXwxODQ5NDg1ODM2>. > > > [[alternative HTML version deleted]] ______________________________________________ R-help@r-project.org 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.
