Generate matrices meet conditions to obtain N matrices and rbind the N accepted matrices together-CodePudding

I want to generate 400 compositional data which meet conditions and then combine them together. The means and variance correlations between three variables used to generate the data are:

mu <- c(45.2, 15.4, 15.4) # means
names (mu) <- c("P", "S", "Fa") # names of the three variances
sted <- c(4.830935, 3.222764, 3.222764) # standard deviations
names (sted) <- c("P", "S", "Fa") # names of the three variances

# covariance matrix: 
corMat <- matrix (c(23.337931, -7.737931, -7.737931, 
                    -7.737931, 10.386207, -5.131034,
                    -7.737931, -5.131034, 10.386207),
                  ncol = 3, byrow = T)
dimnames (corMat) <- list (row_names = c("P", "S", "Fa"),
                           col_names = c("P", "S", "Fa"))

The procedure of stimulated data generating includes two steps, as follows:

Step 1: generate a datum with a normal distribution based mean and variance correlations between three variables:

set.seed(1)
library(MASS)
dat1 <- mvrnorm(n = 1, mu = mu, Sigma = corMat, empirical = F)

Step 2: check if dat1 has: P within the range of 45 <= P <= 55; S within 12 <= S <= 22; and Fa within 12 <= S <= 22. If dat1 meets the three conditions of P, S and Fa, accept dat1. Repeat steps 1 and 2 until achieving 400 accepted data and rbind them together.

I found some questions related to repeat break such as this, but I am still stuck in repeating steps 1 and 2 till obtaining 400 data and bind the data together. Please can anyone help? Thank you very much.

CodePudding user response：

The following solves your problem. However, the generated dat1 appears to never fulfill the conditions you want to impose. Should not dat1 be divided by 100?

nSets <- 0

data <- matrix(ncol=3)

while (nSets < 400)
{
  dat1 <- mvrnorm(n = 1, mu = mu, Sigma = corMat, empirical = F)

  # dat1 <- dat1/100 <-- maybe this is necessary?

  if (dat1["P"] >= 0.45 & dat1["P"] <= 0.55 &
      dat1["S"] >= 0.12 & dat1["S"] <= 0.22 &
      dat1["Fa"] >= 0.12 & dat1["Fa"] <= 0.22)
  {
    nSets <- nSets   1
    data <- rbind(data,dat1)
  }
}

data <-  data[-1,]