similar to: General-purpose GPU computing in statistics (using R)

Hi, Given a positive integer N, and a real number \lambda such that 0 < \lambda < 1, I would like to generate an N by N stochastic matrix (a matrix with all the rows summing to 1), such that it has the second largest eigenvalue equal to \lambda (Note: the dominant eigenvalue of a stochastic matrix is 1). I don't care what the other eigenvalues are. The second eigenvalue is

i want to estimate parameters with maximum likelihood method with contraints (contant numbers). for example sum(Ai)=0 and sum(Bi)=0 i have done it without the constraints but i realised that i have to use the contraints. Without constraints(just a part-not complete): skellamreg_LL=function(parameters,z,design) { n=length(z); mu=parameters[1]; H=parameters[2]; Apar=parameters[3:10];

Dear Martin, I have been playing a lot with the glkerns() function in the "lokern" package for "automatic" smoothing of time-series data. This kernel smoothing approach of Gasser and Mueller seems to perform quite well for estimating the function and its derivatives (first and second derivatives). In fact, this is one of the best methods based on my simulation studies for

Hi, I am currently trying to execute the following command: f<-factanal(factors=k$Components$nparallel,covmat=m,n.obs=2287,rotation="varimax",control=list(opt=list(method=c("BFGS")))) but keep getting the error: L-BFGS-B needs finite values of 'fn' I can't figure out what I am doing wrong here, why isn't optim being told to use BFGS instead of L-BFGS-B...

I've seen with other software the capability for the optimizer to switch algorithms if it is not making progress between iterations. Is this capability available in optim()? Thanks, Stephen Collins, MPP | Analyst Health & Benefits | Aon Consulting [[alternative HTML version deleted]]

Hi, I am trying to solve this equation prob = exp(-3.33 + 0.0102*x)/(1+exp(-3.33 + 0.0102*x)). I want to write a function where I call the function and enter the 'prob' value and the output should be the 'x'. Im not sure how to write this. I have a basic structure but im not sure if its correct. calc <- function(prob){ prob <- exp(-3.33+0.0102*x)/(1+exp(-3.33 + 0.0102*x))

Hi, Suppose x and y are vectors whose elements are known. I know that there is a sinusoidal relation between them. In other words, y=a*sin(bx) where b is probably a function of pi. How do I find a and b in R? Hakan [[alternative HTML version deleted]]

Hi guys Is there a way in R to simulate/generate random numbers from a negative skewed and fat tailed distribution ? I would like to simulate a set of (discrete) data. Regards, Carlos Carlos http://www.nabble.com/file/p25783889/graph.png graph.png -- View this message in context: http://www.nabble.com/Simulate-negative-skewed%2C-fat-tailed-distribution-tp25783889p25783889.html Sent from the

I am currently using the package 'adapt' for multivariate integration. However this package seems to be removed from CRAN (It is still referred to in the help file for integrate(stats) though). I assume it has been deprecated for a reason? Is there an alternative for multivariate numerical integration? ----- Jeroen Ooms * Dept. of Methodology and Statistics * Utrecht University Visit

Dear R users, I have two nonlinear equations, f1(x1,x2)=0 and f2(x1,x2)=0. I try to use optim command by minimize f1^2+f2^2 to find x1 and x2. I found the optimal solution changes when I change initial values. How to solve this? BTW, I also try to use grid searching. But I have no information on ranges of x1 and x2, respectively. Any suggestion to solve this question? Thanks, Kate

I am fittting a linner regression with constrained parameters, saying, all parameters are non-negative and sum up to 1. I have searched historical R-help and found that this can be done by solve.QP from the quadprog package. I need to assess the significance of the coefficient estimates, but there is no standard error of the coefficient estimates in the output. So I can not compute the p-value.

Dear all, I want to compute the eigenvalues of a complex matrix for some statistics. Comparing it to its matlab/octave sibling, I don't get the same eigenvalues in R computing it from the exact same matrix. In R, I used eigen() and arpack() that give different eigenvalues. In matlab/octave I used eig() and eigs() that give out the same eigenvalues but different to the R ones. For real

Dear R users, i try to use function dmvnorm(x, mean, sigma, log=FALSE) from R package mvtnorm to calculate the probability of x under the multivariate normal distribution with mean equal to mean and covariance matrix sigma. I become the following Error in solve.default(cov, ...) : system is computationally singular: reciprocal condition number = 1.81093e-19 What could be the reason of it?

Hello. I am trying to invert a matrix, and I am finding that I can get different answers depending on whether I set LAPACK true or false using "qr". I had understood that LAPACK is, in general more robust and faster than LINPACK, so I am confused as to why I am getting what seems to be invalid answers. The matrix is ostensibly the Hessian for a function I am optimizing. I want to get

Hi all, I have a system of 3 equations with many defined parameters and 3 variables I need to find solutions to. I actually know the solutions I'm aiming for (0.07,0.287,0.0061) but R tends to give me (0,0,0). I tried the "BB" package but don't really follow how to refine my solutions from that; these are wrong so far. Here's my code from trying that: > f<-function(x){

Hi, I have a complex-valued vector X in C^n. Given another complex-valued vector Y in C^n, I want to find a permutation of Y, say, Y*, that minimizes ||X - Y*||, the distance between X and Y*. Note that this problem can be trivially solved for "Real" vectors, since real numbers possess the ordering property. Complex numbers, however, do not possess this property. Hence the

Hi, I apologize for the long message but the problem I encountered can't be stated in a few lines. I am having some problems with the function constrOptim. My goal is to maximize the likelihood of product of K multinomials, each with four catagories under linear constraints on the parameter values. I have found that the function does not work for many data configurations. #The likelihood

Dear R wizards: I am playing (and struggling) with the example in the constrOptim function. simple example. let's say I want to constrain my variables to be within -1 and 1. I believe I want a whole lot of constraints where ci is -1 and ui is either -1 or 1. That is, I have 2*N constraints. Should the following work? N=10 x= rep(1:N) ci= rep(-1, 2*N) ui= c(rep(1, N), rep(-1, N))

Does anybody know of a function that implements the derivative (gradient) of the multivariate normal density with respect to the *parameters*? It?s easy enough to implement myself, but I?d like to avoid reinventing the wheel (with some bugs) if possible. Here?s a simple example of the result I?d like, using numerical differentiation: library(mvtnorm) library(numDeriv) f=function(pars, xx, yy)

Hi all, could someone send me examples of the EM Clustering algorithm and/or some pdf describing it? Thanks, Douglas Sousa.