search for: steptol

...: I have data on an absorption measurement which I want to fit with an voigt profile: fn.1 <- function(p){ for (i1 in ilong){ ff <- f[i1] ex[i1] <- exp(S*n*L*voigt(u,v,ff,p[1],p[2],p[3])[[1]]) } sum((t-ex)^2) } out <- nlm(fn.1, p = c(fo, GG, GL), hessian = TRUE, steptol = 1e-5, iterlim = 1000) foN <- out$estimate[1] GGN <- out$estimate[2] GLN <- out$estimate[3] This works fine but the my start value of S is to poor, so I like to fit S in a second run, with the initial values from the first run (two runs because I know that S as an parameter is an pr...

...ut I presume this would be cleaner and faster if implemented at a lower level ... nlm2 <- function(f, p, hessian=FALSE, typsize=rep(1,length(p)), fscale=1, print.level=0, ndigit=12, gradtol=1e-6, stepmax=max(1000 * sqrt(sum((p/typsize)^2)), 1000), steptol=1e-6, iterlim=100, check.analyticals=TRUE, ...) { tmpf <- function(x) { f(x,...) } nlm(tmpf,p,hessian=hessian,typsize=typsize,fscale=fscale,print.level=print.level, ndigit=ndigit,gradtol=gradtol,stepmax=stepmax,steptol=steptol,iterlim=iterlim, check.analyticals=check.analyt...

R version: RW0632 1. gnlm package, nlr() error message: "Last global step failed to locate a point lower than x. Either x is an approximate local minimum of the function, the function is too non-linear for this algorithm, or steptl is too large." last line: steptol (also, should it read "steptol is too small?") 2. help(glm) " Fitting Generalized Linear Models glm(formula, family=gaussian, data, weightst=NULL, subsett=NULL," typos in "weights" and "subset" Thank you! Rashid Nassar -.-.-.-.-.-.-.-....

...onverged after 100 iterations. What might be done to improve the performance of something like "nlm" without substantially increasing the overhead for other problems? Thanks. spencer graves ############################# nlm. <- function(f=fgh, p=c(-1.2, 1), gradtol=1e-6, steptol=1e-6, iterlim=100){ # R code version of "nlm" # requiring analytic gradient and hessian # # Initial evaluation f.i <- f(p) f0 <- f.i+1 # Iterate for(i in 1:iterlim){ df <- attr(f.i, "gradient") # Gradient sufficiently small? if(sum(df^2)<(gradtol...

I am trying to view the source code of the function nlm in the stats package of R 2.4.1. I downloaded the source from CRAN and opened nlm.R, and it calls a .Internal function: .Internal(nlm(function(x) f(x, ...), p, hessian, typsize, fscale, msg, ndigit, gradtol, stepmax, steptol, iterlim)) This is the same thing I saw when entering the function name at the R command prompt. Where will I find the actual code? Thanks.

...;- function(A) { A1=A[1] A2=A[2] A3=A[3] regm1 <- nnet(Xcc,Ycr,entropy=T,size=A1,decay=A2,maxit=2000,trace=F,Hess=T,rang=A3,skip=T) dif=sum((predict(regm1,Xcc)-Ycr)^2) return(dif) } somar=DEoptim(pred_regm1,c(1,0.00001,0.01), c(25,0.999,0.95), control = DEoptim.control(steptol=25,trace = FALSE,itermax=500,parallelType=1)) but implemented parallel process, i have next error: Error en checkForRemoteErrors(val) : 4 nodes produced errors; first error: could not find function "nnet" how I can solve my mistake? Thanks and kind regards -- David Zamora Ávila...

...0, upper=50, maximum=FALSE, tol=tol, alpha.lower=alpha.lower, F.value=F.value, df.1=df.1, df.2=df.2) # Try to accomplish the same task with a different R function. nlm(f=Low.Lim.NC.F, p=LL.0, fscale=1, print.level = 0, ndigit=12, gradtol = 1e-6, stepmax = max(1000 * sqrt(sum((LL.0/10)^2)), 1000), steptol = 1e-6, iterlim = 1000, check.analyticals = TRUE, alpha.lower=alpha.lower, F.value=F.value, df.1=df.1, df.2=df.2) # The answer in each case is 3.0725. Thus, a noncentral F with # 5 and 200 df with a noncentrality parameter 3.0725 has at its .975 # quantile a value of 4 (this has been verified in...

Dear R list members, Since today seems to be the day for optimization questions, I have one that has been puzzling me: I've been doing some work on sem, my structural-equation modelling package. The models that the sem function in this package fits are essentially parametrizations of the multinormal distribution. The function uses optim and nlm sequentially to maximize a multinormal

Hi Everyone, Is there a simple way to force nlm to take larger initial steps? Setting print.level = 2 allows me to inspect the step size at each iteration, but I appear not to have made any appreciable impact on it by changing values of typsize, fscale, steptol or stepmax. The steps repeatedly come out tiny, 1e-9 typically, and the algorithm is terminating not because the gradient is zero (it is not, according to the numerical values) but because the function value is not changing between iterations. Am I missing something obvious? Many thanks, Jonatha...

...tried to change the optional parameters in the nlm call but this does not help to achieve code=1. What may be a reason for such situation? Could I think that the obtained estimates are what I am looking for? I am using R 1.9.1: nlm(N.Loglik,initial.values,fscale=1,print.level = 1, gradtol = 1e-6,steptol = 1e-6,iterlim = 500,y=y) Much thanks, Victoria. [[alternative HTML version deleted]]