## Consider these values for EX-EY.
for (d in c(0.5,1,1.5))
{
  ## Consider these sample sizes (for both the X and Y sample).
  for (n in c(10,20,30))
  {
    ## Consider these values for var(X) and var(Y) (we are assuming
    ## that the variances are equal here).
    for (v in c(1,2,3))
    {
      ## Generate 1000 data sets from the appropriate alternative
      ## distribution.
      X = array(d + sqrt(v)*rnorm(n*1000), c(1000,n))
      Y = array(sqrt(v)*rnorm(n*1000), c(1000,n))

      ## Calculate T-statistics for the simulated data sets.
      MX = apply(X, 1, mean)
      MY = apply(Y, 1, mean)
      VX = apply(X, 1, var)
      VY = apply(Y, 1, var)
      Sp2 = ((n-1)*VX + (n-1)*VY) / (2*n-2)
      TS = (MX-MY) / sqrt(2*Sp2/n)

      ## This is an estimate of the power.
      pw = mean(TS > qt(0.95, 2*n-2))

      ## Print the result.
      print(c(d, n, v, pw))
    }
  }
}