Skip to main content

Table 8 Distance between inverse Fisher information matrix and sample covariance matrix for various sample sizes

From: Application of OU processes to modelling temporal dynamics of the human microbiome, and calculating optimal sampling schemes

\(\eta\) \(\sigma\) n
5000 1000 500 100 50 10
0.5 1 0.054 0.053 0.052 0.058 0.125 0.996
2 0.066 0.065 0.064 0.055 0.052 0.998
4 0.066 0.065 0.064 0.055 0.040 0.995
1 1 0.008 0.008 0.011 0.139 0.330 0.968
2 0.057 0.051 0.046 0.106 0.266 1.000
4 0.122 0.109 0.099 0.056 0.121 0.853
2 1 0.001 0.033 0.147 0.078 0.021 12.281
2 0.002 0.028 0.123 0.073 0.047 12.168
4 0.005 0.020 0.102 0.045 0.098 7.586
  1. For small sample sizes (mostly \(n\leqslant 100\)) there were a few simulations where \({\hat{\eta }}=\infty\), i.e. there is no estimated correlation between consecutive samples. These simulations were removed from the variance calculation.
\